The Cluster Active Archive: Studying the Earth's Space Plasma Environment (Astrophysics and Space Science Proceedings)

Astrophysics and Space Science Proceedings The Cluster Active Archive Studying the Earth’s Space Plasma Environment ...

Author: Harri Laakso | Matthew Taylor | C. Philippe Escoubet

24 downloads 679 Views 18MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Astrophysics and Space Science Proceedings

The Cluster Active Archive Studying the Earth’s Space Plasma Environment

Harri Laakso Editor ESA/ESTEC, Noordwijk, The Netherlands

Matthew G.T.T. Taylor Editor ESA/ESTEC, Noordwijk, The Netherlands

C. Philippe Escoubet Editor ESA/ESTEC, Noordwijk, The Netherlands

123

Editors Dr. Harri Laakso European Space Agency European Space Research & Technology Centre (ESTEC) Keplerlaan 1 2200 AG Noordwijk Netherlands [email protected]

Dr. C. Philippe Escoubet European Space Agency European Space Research & Technology Centre (ESTEC) Keplerlaan 1 2200 AG Noordwijk Netherlands [email protected]

Dr. Matthew Taylor European Space Agency European Space Research & Technology Centre (ESTEC) Keplerlaan 1 2200 AG Noordwijk Netherlands [email protected]

ISSN 1570-6591 e-ISSN 1570-6605 ISBN 978-90-481-3498-4 e-ISBN 978-90-481-3499-1 DOI 10.1007/978-90-481-3499-1 Springer Dordrecht Heidelberg London New York Library of Congress Control Number: 2009934513 c Springer Science+Business Media B.V. 2010 ° No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Cover design: eStudio Calamar S.L. Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

Preface

Magnetospheric physics started more than 50 years ago when many of the first satellites carried detectors to explore the space environment around the Earth and made plenty of new discoveries. The first findings inspired the development of a plethora of scientific satellites in the 1960s and 1970s exploring near-Earth space, which was found to have several distinct regions with their own structural and dynamical characteristics. Since then magnetospheric physics has remained a strong space science discipline, not only for its own sake but also because near-Earth space provides us with a unique space laboratory where we can investigate physical processes of tenuous collisionless plasmas with in-situ instruments and apply the resulting understanding to physical phenomena observed in more exotic environments of the Solar System and of the Universe. For long time magnetospheric research was carried out only by means of single spacecraft. In the 1970s the science community realized that one needs coordinated simultaneous observations from several satellites in order to make a major step forward in magnetospheric and space plasma physics. In particular the ISEE (three satellites), GEOS (two satellites) and DE (two satellites) programs provided many new discoveries that single spacecraft measurements could not achieve. In particular it was the dual spacecraft constellation of ISEE-1 and ISEE-2 that offered the first opportunity to improve the understanding of spatio-temporal features in physical processes of space environment while the other previous missions were separated by very large distances and were on different orbits, so one could not separate spatial and temporal features. In early 1980s the science community started dreaming about the next generation of multi-satellite constellation mission, the culmination of which came in 1982 when a group of scientists proposed a four-satellite mission, called Cluster, to ESA. After the assessment studies, the Cluster mission, together with the SOHO mission, were selected as the first cornerstone of the ESA’s Horizons 2000 Programme in 1986. Unfortunately, after 10 years of hard work the Cluster satellites were dramatically destroyed by the failure of the first flight of Ariane 5 on 4 June 1996. However, it was decided soon after that the mission was of such high importance to the science community that the satellites and payload must be rebuilt, and 4 years later the four Cluster spacecraft were successfully launched in pairs on two Soyuz rockets in July and August 2000. After 1 year of successful operations, the mission was extended

v

vi

Preface

for an additional 35 months, up to December 2005. Later the mission was further extended for 4 years until the end of 2009, and a new extension proposal has been submitted to request further operations beyond 2009. During the first extension, ESA decided to establish the Cluster Active Archive (CAA) to contain all processed and validated high-resolution scientific data, as well as raw data, calibration data and documentation from all the Cluster instruments. This is essential in order to preserve all Cluster observations in a stable, long-term archive for scientific analysis well beyond the end of the mission. The CAA activities started in 2004 and 2 years later in February 2006, the CAA opened services to the community. The Cluster mission has held science workshops twice a year since autumn 2001, which has helped bring together scientists to compare and exchange their observations and begin new collaborations. Before the CAA, scientists got access to the Cluster observations via their personal contacts with the instrument Principal Investigators and Co-Investigators. Now and in the future, the CAA provides the primary access route to the Cluster observations. Soon after the CAA opening it was realized that while some of the Cluster datasets are quite easy to use, there are plenty of complex datasets whose usage may not always be obvious. Therefore it was decided to organize a school for the users of the CAA datasets, which was organized together with the 15th Cluster science workshop in order to attract as many users as possible. The details of this and other workshops can be found at the CAA website http://caa.estec.esa.int/. The 15th Cluster workshop, attended by 120 people, was held in Hotel Playa la Arena (Puerto Santiago), Tenerife, Canary Islands, on 9–15 March 2008. The workshop provided the attending community with an in-depth overview of the CAA data products and tools. In addition, attendees were able to discuss with instrument team developers, the core CAA team and software developers. The packed schedule included presentations of instrument datasets, software tools and the CAA infrastructure and was interspersed with invited scientific talks and several dedicated science sessions on the bow shock, solar wind, magnetopause and cusp, the magnetotail and inner magnetosphere. The fact that the workshop/school was hosted in the same location as the accommodation was fundamental in the success of the meeting, allowing attendees to continue working with colleagues and discussing work in and around the workshop hours. This book contains presentations made at the workshop, including several articles about the CAA and its datasets as well as a few overview papers on the Cluster mission. In addition all participants were offered an opportunity to publish their scientific results on solar wind/magnetosheath, magnetopause, and magnetotail, presented during the workshop in this book. We would like to thank the following people for their contributions in reviewing the articles in this book: O. Alexandrova, A. Allen, M. Andr´e, T. Asikainen, ˚ A. Asnes, R. Benson, J. Blecki, J. Borovsky, D. Burgess, J. Cao, N. CornilleauWehrlin, P. Daly, I. Dandouras, J. Davies, M. Dunlop, J. Eastwood, S. Elkington, A. Eriksson, J. Faden, A. Fazakerley, J. Fennell, R. Friedel, T. Fritz, S. Fung, E. Gamby, V. G´enot, M. Goldstein, E. Grigorenko, B. Grison, M. Hapgood,

Preface

vii

C. Harvey, R. Horne, C. Jacquey, Y. Khotyaintsev, L. Kistler, H. Kucharek, G. Le, E. Lucek, J. McFadden, S. Milan, F. Mozer, J. Mukherjee, H. Opgenoorth, C. Owen, G. Parks, M. Parrot, A. Pedersen, C. Perry, J. Pickett, P. Puhl-Quinn, J. Rae, J.L. Rauch, A. Retino, A. Roux, V. Sergeev, A. Tjulin, K. Torkar, K. Trattner, A. Walsh, Y. Wang, J. Weygand, J. Wild, and Q. Zong. May 2009

Harri Laakso, Matthew G.T.T Taylor, and Philippe Escoubet

View on the Tenerife Island

Contents

Preface .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .

v

Part I Products and Services of the Cluster Active Archive 1

Cluster Active Archive: Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . H. Laakso, C. Perry, S. McCaffrey, D. Herment, A.J. Allen, C.C. Harvey, C.P. Escoubet, C. Gruenberger, M.G.G.T. Taylor, and R. Turner

3

2

ASPOC Data Products in the Cluster Active Archive . . . . . . . . .. . . . . . . . . . . 39 K. Torkar and H. Jeszenszky

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA) . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . 51 I. Dandouras, A. Barthe, E. Penou, S. Brunato, H. R`eme, L.M. Kistler, M.B. Bavassano-Cattaneo, A. Blagau, and the CIS Team

4

Digital Wave Processor Products in the Cluster Active Archive .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . 73 K.H. Yearby, H.St.C. Alleyne, S.N. Walker, I. Bates, M.P. Gough, A. Buckley, and T.D. Carozzi

5

EDI Data Products in the Cluster Active Archive . . . . . . . . . . . . .. . . . . . . . . . . 83 E. Georgescu, P. Puhl-Quinn, H. Vaith, M. Chutter, J. Quinn, G. Paschmann, and R. Torbert

6

The EFW Data in the CAA .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . 97 Y. Khotyaintsev, P.-A. Lindqvist, A. Eriksson, and M. Andr´e

7

FGM Data Products in the CAA .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .109 J.M. Gloag, E.A. Lucek, L.-N. Alconcel, A. Balogh, P. Brown, C.M. Carr, C.N. Dunford, T. Oddy, and J. Soucek

ix

x

Contents

8

PEACE Data in the Cluster Active Archive . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .129 A.N. Fazakerley, A.D. Lahiff, R.J. Wilson, I. Rozum, C. Anekallu, M. West, and H. Bacai

9

RAPID Products at the Cluster Active Archive .. . . . . . . . . . . . . . .. . . . . . . . . . .145 Patrick W. Daly and Elena A. Kronberg

10 STAFF Instrument Products Distributed Through the Cluster Active Archive .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .159 N. Cornilleau-Wehrlin, L. Mirioni, P. Robert, V. Bouzid, M. Maksimovic, Y. de Conchy, C.C. Harvey, and O. Santol´ık 11 Cluster Wideband Data Products in the Cluster Active Archive .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .169 J.S. Pickett, J.M. Seeberger, I.W. Christopher, O. Santol´ık, and K.M. Sigsbee 12 The WHISPER Relaxation Sounder and the CLUSTER Active Archive. . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .185 J.G. Trotignon, P.M.E. D´ecr´eau, J.L. Rauch, X. Valli`eres, A. Rochel, S. Kougbl´enou, G. Lointier, G. Facsk´o, P. Canu, F. Darrouzet, and A. Masson 13 ESOC Data Products in the CAA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .209 J. Volpp and D. Sieg Part II Tools for the CAA Data Analysis 14 QSAS, QM Science Analysis Software . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .225 A.J. Allen 15 Flexible Tools for Accessing the Cluster Archives . . . . . . . . . . . . .. . . . . . . . . . .233 E. Gamby, J. De Keyser, and M. Roth 16 AMDA, Automated Multi-Dataset Analysis: A Web-Based Service Provided by the CDPP .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .239 C. Jacquey, V. G´enot, E. Budnik, R. Hitier, M. Bouchemit, M. Gangloff, A. Fedorov, B. Cecconi, N. Andr´e, B. Lavraud, C. Harvey, F. D´eriot, D. Heulet, E. Pallier, E. Penou, and J.L. Pinc¸on 17 Cluster CAA Module for PaPCo .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .249 ˚ J. Faden, A. Asnes, R. Friedel, M. Taylor, S. McCaffrey, C. Perry, and M.L. Goldstein

Contents

Part III

xi

Measurement Techniques and Calibration Routines

18 Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .261 A. Masson, O. Santol´ık, M.G.G.T. Taylor, C.P. Escoubet, ˚ A.N. Fazakerley, J. Pickett, A. Asnes, X. Valli`eres, H. Laakso, and J.-G. Trotignon 19 Cluster-PEACE In-flight Calibration Status . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .281 A.N. Fazakerley, A.D. Lahiff, I. Rozum, D. Kataria, H. Bacai, ˚ C. Anekallu, M. West, and A. Asnes 20 Generation and Validation of Ion Energy Spectra Based on Cluster RAPID and CIS Measurements . . . . . . . . . . . . .. . . . . . . . . . .301 Elena A. Kronberg, Patrick W. Daly, Iannis Dandouras, Stein Haaland, and Edita Georgescu Part IV

Magnetospheric Missions

21 The Cluster Mission: Space Plasma in Three Dimensions . . . .. . . . . . . . . . .309 M.G.G.T. Taylor, C.P. Escoubet, H. Laakso, A. Masson, and M.L. Goldstein 22 Double Star: Mission, Instruments and Joint Observations ... . . . . . . . . . .331 M.W. Dunlop, C.P. Escoubet, Z.-X. Liu, C. Shen, H. Laakso, M.G.G.T. Taylor, A.N. Fazakerley, and the Double star PIs Part V

Observations of Solar Wind and Magnetosheath

23 Effect of Shock Normal Orientation Fluctuations on Field-Aligned Beam Distributions. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .349 K. Meziane, A.M. Hamza, M. Wilber, M.A. Lee, C. Mazelle, E.A. Lucek, T. Hada, and A. Markowitch 24 Wave Number Spectra in the Solar Wind, the Foreshock, and the Magnetosheath . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .363 Y. Narita, K.-H. Glassmeier, S.P. Gary, M.L. Goldstein, and R.A. Treumann 25 Cluster Hot Flow Anomaly Observations During Solar Cycle Minimum . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .369 G. Facsk´o, M. T´atrallyay, G. Erd˝os, and I. Dandouras

xii

Contents

26 On the Growth of Mirror Mode Waves in the Magnetosheath Based on Cluster Observations . . . . . . . .. . . . . . . . . . .377 M. T´atrallyay, G. Erd˝os, I. Dandouras, and E. Georgescu Part VI Observations of Magnetopause and Cusp 27 Mixed Azimuthal Scales of Flux Transfer Events . . . . . . . . . . . . . .. . . . . . . . . . .389 R.C. Fear, S.E. Milan, E.A. Lucek, S.W.H. Cowley, and A.N. Fazakerley 28 Perspectives Gained from a Combination of Polar, Cluster and ISEE Energetic Particle Measurements in the Dayside Cusp . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .399 T.A. Fritz 29 Energetic Particles in the Cusp: A Cluster/RAPID View . . . . .. . . . . . . . . . .415 T. Asikainen Part VII Observations of Magnetospheric Tail 30 Plasma Flow Reversals in the Magnetotail and Their Implications on Substorm Models. . . . . . . . . . . . . . . . . . .. . . . . . . . . . .429 A.T.Y. Lui 31 Spatial and Temporal Structures in the Vicinity of the Earth’s Tail Magnetic Separatrix Cluster Observations .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .435 E.E. Grigorenko, L.M. Zelenyi, M.S. Dolgonosov, and J.-A. Sauvaud 32 Cluster Observations of Energy Conversion Regions in the Plasma Sheet .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .453 O. Marghitu, M. Hamrin, B. Klecker, K. R¨onnmark, S. Buchert, L.M. Kistler, M. Andr´e, and H. R`eme 33 Acceleration of > 40 keV Electrons in Near-Earth Magnetotail Reconnection Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .461 ˚ A. Asnes, M.G.G.T. Taylor, and A.L. Borg 34 Evolution and Energization of Energetic Electrons in the Inner Magnetosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . .467 J.F. Fennell and J.L. Roeder

Contributors

L.-N. Alconcel Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, SW7 2BW, UK A.J. Allen Blacket Laboratory, Imperial College, London, SW7 2AZ, UK, [email protected] H.St.C. Alleyne Department of Automatic Control and Systems Engineering, University of Sheffield, Mappin Street, Sheffield, S1 3JD, UK, [email protected] M. Andr´e Swedish Institute of Space Physics, Uppsala, Sweden N. Andr´e CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France C. Anekallu Mullard Space Science Laboratory, University College London, UK T. Asikainen Department of Physics Centre of Excellence in Research, P.O. Box 3000, FIN-90014, University of Oulu, Finland, [email protected] ˚ A. Asnes ESA/ESTEC, Noordwijk, The Netherlands, [email protected] H. Bacai Mullard Space Science Laboratory, University College London, UK A. Balogh Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, SW7 2BW, UK and International Space Science Institute (ISSI), Hallerstrasse 6, 3012 Bern, Switzerland A. Barthe Universit´e de Toulouse, Centre d’Etude Spatiale des Rayonnements, B.P. 44346, 31028 Toulouse, France and AKKA Technologies, 6 Rue R. Camboulives, 31036 Toulouse, France, [email protected] I. Bates Department of Automatic Control and Systems Engineering, University of Sheffield, Mappin Street, Sheffield, S1 3JD, UK

xiii

xiv

Contributors

M.B. Bavassano-Cattaneo Istituto di Fisica dello Spazio Interplanetario, INAF, Via del Fosso del Cavaliere 100, 00133 Roma, Italy, [email protected] A. Blagau Institute of Space Sciences, P.O. Box MG-23, RO 77125 BucharestMagurele, Romania, [email protected] A.L. Borg Norwegian Defense Research Institute, Kjeller, Norway M. Bouchemit CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France V. Bouzid Laboratoire de Physique des Plasmas, Ecole Polytechnique, UPMC, CNRS, route de Saclay, 91128 Palaiseau, France P. Brown Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, SW7 2BW, UK S. Brunato Universit´e de Toulouse, Centre d’Etude Spatiale des Rayonnements, B.P. 44346, 31028 Toulouse, France and Noveltis, 2 Avenue de l’Europe, 31520 Ramonville-Saint-Agne, France, [email protected] S. Buchert Swedish Institute of Space Physics, Uppsala, Sweden A. Buckley Space Science Group, University of Sussex, Falmer, Brighton, BN1 9QH, UK, [email protected] E. Budnik Noveltis, 2, Avenue Europe, 31520 Ramonville Saint Agne, France P. Canu Laboratoire de Physique des Plasmas, V´elizy, France, [email protected] T.D. Carozzi Space Science Group, University of Sussex, Falmer, Brighton, BN1 9QH, UK, [email protected] C.M. Carr Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, SW7 2BW, UK B. Cecconi LESIA, Observatoire de Paris-Meudon, 5, place Janssen, 92195 Meudon I.W. Christopher Department of Physics and Astronomy, The University of Iowa, Iowa City, 52240, USA M. Chutter University of New Hampshire, Durham, USA, [email protected] N. Cornilleau-Wehrlin Laboratoire de Physique des Plasmas, Ecole Polytechnique, UPMC, CNRS, route de Saclay, 91128 Palaiseau, France and Station de Radioastronomie de Nanc¸ay, 18330 Nancay, France, [email protected]

Contributors

xv

S.W.H. Cowley University of Leicester, UK I. Dandouras Universit´e de Toulouse, Centre d’Etude Spatiale des Rayonnements, B.P. 44346, 31028 Toulouse, France, [email protected] F. Darrouzet Institut d’A´eronomie Spatiale de Belgique, Bruxelles, Belgique, [email protected] J. De Keyser Belgian Institute for Space Aeronomy, Brussels, Belgium P.M.E. D´ecr´eau Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France, [email protected] F. D´eriot CNES, Centre spatial de Toulouse, 18 avenue E. Belin, 31401 Toulouse, France M.S. Dolgonosov Space Research Institute of RAS, Moscow, Profsoyuznaya str, 84/32, Russia C.N. Dunford Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, SW7 2BW, UK M.W. Dunlop SSTD, Rutherford Appleton Laboratory, Chilton, Didcot, Oxfordshire, OX11 0QX, UK, [email protected] G. Erd˝os KFKI Research Institute for Particle and Nuclear Physics, H-1525 Budapest, P.O. Box 49, Hungary A. Eriksson Swedish Institute of Space Physics, Uppsala, Sweden C.P. Escoubet European Space Agency, ESTEC, D-SRE, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands G. Facsk´o Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France and KFKI Research Institute for Particle and Nuclear Physics, H-1525 Budapest, P.O. Box 49, Hungary, [email protected] J. Faden Cottage Systems, Iowa City, IA, USA, [email protected] A.N. Fazakerley Mullard Space Science Laboratory, University College London, Dorking, Surrey, RH5 6NT, UK, [email protected] R.C. Fear University of Leicester, UK, [email protected] A. Fedorov CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France J.F. Fennell The Aerospace Corporation, P.O. Box 92957, Los Angeles, CA 90009, USA, [email protected] R. Friedel Los Alamos National Labs, Los Alamos, NM, USA T.A. Fritz Center for Space Physics, Boston University, Boston, MA 02215 USA, [email protected]

xvi

Contributors

E. Gamby Belgian Institute for Space Aeronomy, Brussels, Belgium, [email protected] M. Gangloff CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France V. G´enot CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France E. Georgescu Max-Planck-Institute for Solar Research, Katlenburg-Lindau, Germany, [email protected] J.M. Gloag Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, SW7 2BW, UK M.L. Goldstein Geospace Physics Laboratory, NASA Goddard Space Flight Center, Code 673, Greenbelt, MD 20771, USA M.P. Gough Space Science Group, University of Sussex, Falmer, Brighton, BN1 9QH, UK, [email protected] E.E. Grigorenko Space Research Institute of RAS, Moscow, Profsoyuznaya str, 84/32, Russia, [email protected] C. Gruenberger European Space Agency, ESTEC, Noordwijk, The Netherlands T. Hada ESST, Kyushu University, Fukuora, Japan M. Hamrin Department of Physics, Ume˚a University, Ume˚a, Sweden and Swedish Institute of Space Physics, Ume˚a, Sweden A.M. Hamza Physics Department, University of New Brunswick, Canada C.C. Harvey Centre de Donn´ees de Physique des Plasmas, CNRS, Toulouse, France D. Herment European Space Agency, ESTEC, Noordwijk, The Netherlands D. Heulet CNES, Centre spatial de Toulouse, 18 avenue E. Belin, 31401 Toulouse, France R. Hitier Co-Libri, Cremefer 11290 Montr´eal, France C. Jacquey CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France, [email protected] H. Jeszenszky Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, 8042 Graz, Austria D. Kataria Mullard Space Science Laboratory, University College London, UK Y. Khotyaintsev Swedish Institute of Space Physics, Uppsala, Sweden, [email protected] L.M. Kistler Space Science Center, University of New Hampshire, Durham, NH 03824, USA, [email protected]

Contributors

xvii

B. Klecker Max-Planck-Institut f¨ur extraterrestrische Physik, Garching, Germany S. Kougbl´enou Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France, [email protected] H. Laakso European Space Agency, ESTEC, D-SRE, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands, [email protected] A.D. Lahiff Mullard Space Science Laboratory, University College London, UK B. Lavraud CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France M.A. Lee Space Science Center, University of New Hampshire, Durham, USA P.-A. Lindqvist Royal Institute of Technology, Stockholm, Sweden Z.-X. Liu Centre for Space Science and Applied Research, Chinese Academy of Sciences, Beijing 100080, China G. Lointier Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France, [email protected] E.A. Lucek Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, SW7 2BW, UK, [email protected] A.T.Y. Lui The Johns Hopkins University Applied Physics Laboratory, Laurel, MD 20723, USA, [email protected] M. Maksimovic LESIA, Observatoire de Paris-Meudon, 5 place Jules Jansen, 92195 Meudon Cedex, France O. Marghitu Institute for Space Sciences, Bucharest, Romania and Max-Planck-Institut f¨ur extraterrestrische Physik, Garching, Germany, [email protected] A. Markowitch Laboratoire de Physique et Astroparticules, Universit´e de Montpellier, France A. Masson RSSD, ESA, Noordwijk, The Netherlands and ESA/ESTEC, D-SRE, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands, [email protected] C. Mazelle Centre d’Etudes Spatiales des Rayonnements, Toulouse, France S. McCaffrey European Space Agency, ESTEC, Noordwijk, The Netherlands K. Meziane Physics Department, University of New Brunswick, Canada, [email protected] L. Mirioni Laboratoire de Physique des Plasmas, Ecole Polytechnique, UPMC, CNRS, route de Saclay, 91128 Palaiseau, France

xviii

Contributors

T. Oddy Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, SW7 2BW, UK E. Pallier CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France G. Paschmann Max-Planck-Institute for Extraterrestrial Physics, Garching, Germany, [email protected] E. Penou Universit´e de Toulouse, Centre d’Etude Spatiale des Rayonnements, B.P. 44346, 31028 Toulouse, France and CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France, [email protected] C. Perry Rutherford Appleton Laboratory, Chilton, Didcot, Oxon, UK J.S. Pickett Department of Physics and Astronomy, The University of Iowa, Iowa City, 52240, USA, [email protected] J.L. Pinc¸on LPCE, Laboratoire de Physique et Chimie de l’Environnement, 45071 Orl´eans, France P. Puhl-Quinn University of New Hampshire, Durham, USA, [email protected] J. Quinn Boston University, Boston, MA, USA, [email protected] J.L. Rauch Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France, [email protected] P. Robert Laboratoire de Physique des Plasmas, Ecole Polytechnique, UPMC, CNRS, route de Saclay, 91128 Palaiseau, France A. Rochel Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France, [email protected] J.L. Roeder The Aerospace Corporation, P.O. Box 92957, Los Angeles, CA 90009, USA K. R¨onnmark Department of Physics, Ume˚a University, Ume˚a, Sweden M. Roth Belgian Institute for Space Aeronomy, Brussels, Belgium H. R`eme Universit´e de Toulouse, Centre d’Etude Spatiale des Rayonnements, B.P. 44346, 31028 Toulouse, France and CNRS, UMR 5187, B.P. 44346, 31028 Toulouse, France, [email protected] I. Rozum Mullard Space Science Laboratory, University College London, UK

Contributors

xix

O. Santol´ık Department of Physics and Astronomy, The University of Iowa, Iowa City, Iowa, 52240, USA and Institute of Atmospheric Physics, ASCR, Bocni II/1401, CZ-14131 PRAHA 4, Czech Republic and Institute of Atmospheric Physics and Charles University, Faculty of Mathematics and Physics, Prague, 18000, Czech Republic J.-A. Sauvaud Centre d’Etudie Spatial des Rayonnements, Toulouse cedex4, France J.M. Seeberger Department of Physics and Astronomy, The University of Iowa, Iowa City, 52240, USA C. Shen Centre for Space Science and Applied Research, Chinese Academy of Sciences, Beijing 100080, China D. Sieg Hewlett-Packard GmbH at European Space Operations Centre, [email protected] K.M. Sigsbee Department of Physics and Astronomy, The University of Iowa, Iowa City, 52240, USA J. Soucek Department of Space Physics, Institute of Atmospheric Physics, Academy of Sciences of the Czech Republic, Prague, Czech Republic M. T´atrallyay KFKI Research Institute for Particle and Nuclear Physics, H-1525 Budapest, P.O. Box 49, Hungary, [email protected] M.G.G.T. Taylor European Space Agency, ESTEC, Noordwijk, The Netherlands, [email protected] R. Turner European Space Agency, ESTEC, Noordwijk, The Netherlands R. Torbert University of New Hampshire, Durham, USA, [email protected] K. Torkar Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, 8042 Graz, Austria, [email protected] J.G. Trotignon Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France, [email protected] H. Vaith University of New Hampshire, Durham, USA, [email protected] X. Valli`eres Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France, [email protected] J. Volpp European Space Operations Centre, Robert-Bosch-Strasse 5, D-64625 Darmstadt, Germany, [email protected]

xx

Contributors

S.N. Walker Department of Automatic Control and Systems Engineering, University of Sheffield, Mappin Street, Sheffield, S1 3JD, UK, [email protected] M. West Mullard Space Science Laboratory, University College London, UK and Royal Observatory, Belgium M. Wilber Space Sciences Laboratory, University of California, Berkeley, USA R.J. Wilson Mullard Space Science Laboratory, University College London, UK and Los Alamos National Laboratory, Los Alamos, USA K.H. Yearby Department of Automatic Control and Systems Engineering, University of Sheffield, Mappin Street, Sheffield, S1 3JD, UK, [email protected] L.M. Zelenyi Space Research Institute of RAS, Moscow, Profsoyuznaya str, 84/32, Russia

Part I

Products and Services of the Cluster Active Archive

Chapter 1

Cluster Active Archive: Overview H. Laakso, C. Perry, S. McCaffrey, D. Herment, A.J. Allen, C.C. Harvey, C.P. Escoubet, C. Gruenberger, M.G.G.T. Taylor, and R. Turner

Abstract The four-satellite Cluster mission investigates the small-scale structures and physical processes related to interaction between the solar wind and the magnetospheric plasma. The Cluster Active Archive (CAA) (URL: http://caa.estec.esa.int) will contain the entire set of Cluster high-resolution data and other allied products in a standard format and with a complete set of metadata in machine readable format. The total amount of the data files in compressed format is expected to exceed 50 TB. The data archive is publicly accessible and suitable for science use and publication by the world-wide scientific community. The CAA aims to provide user-friendly services for searching and accessing these data and ancillary products. The CAA became operational in February 2006 and as of Summer 2008 has data from most of the Cluster instruments for at least the first 5 years of operations (2001–2005). The coverage and range of products are being continually improved with more than 200 datasets available from each spacecraft, including high-resolution magnetic and electric DC fields and wave spectra; full three-dimensional electron and ion distribution functions from a few eV to hundreds of keV; and various ancillary and browse products to help with spacecraft and event location. The CAA is continuing to extend and improve the online capabilities of the system and the quality of the existing data. It will add new data files for years 2006–2009 and is preparing for the long-term archive with complete coverage after the completion of the Cluster mission.

H. Laakso (), S. McCaffrey, D. Herment, C.P. Escoubet, C. Gruenberger, M.G.G.T. Taylor, and R. Turner European Space Agency, ESTEC, Noordwijk, The Netherlands e-mail: [email protected] C. Perry Rutherford Appleton Laboratory, Chilton, Didcot, Oxon, UK A.J. Allen Imperial College, London, UK C.C. Harvey Centre de Donn´ees de Physique des Plasmas, CNRS, Toulouse, France

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 1, c Springer Science+Business Media B.V. 2010

3

4

H. Laakso et al.

1.1 Introduction The Cluster mission consists of four identical spacecraft, each carrying 11 instruments [2]. The satellites were launched in two pairs in July/August 2000, and after a 6-months commissioning phase, the mission moved into an operational phase in February 2001. The mission has been approved to operate until summer 2009, and a new extension proposal is in preparation to request further operations beyond 2009.

1.1.1 Cluster Mission A key aspect of the Cluster mission is the controllable flight formation. At two locations along the satellite trajectories, the four spacecraft are aligned at the tips of a regular tetrahedron that is the optimal configuration for the measurements of spatial gradients. In other positions the tetrahedron is less regular and near the perigee the satellites fly in a string-of-pearls formation, and there are two positions on each orbit where the four satellites are coplanar. The size of the tetrahedron formation was changed systematically, covering various distances between 100 and 10,000 km during the years 2001–2005. This strategy was necessary to cover the different characteristic spatial and temporal scales of the fundamental plasma physics processes in the different regions of the dynamic magnetosphere. During the mission extension, which started in January 2006, Cluster has been in a multi-scale phase where three satellites fly in a 10,000 km triangular formation and the fourth satellite (Cluster 4) is perpendicular to the plane and is separated from Cluster 3 by a distance varying between 40 km and a few 1,000 km.

1.1.2 CAA Concept The key scientific rationales for the Cluster Active Archive are (1) to maximize the scientific return from the Cluster mission by making all high-resolution Cluster data available to the world-wide scientific community, (2) to ensure that the unique Cluster observations are preserved in a stable, long-term archive for scientific analysis beyond the end of the mission, and (3) to provide this archive as a major contribution by ESA and the Cluster science community to the International Living With a Star programme. The CAA database and services are established and maintained by the European Space Agency (ESA) at ESTEC. The aims of the CAA are to ensure that the CAA contains all of the Cluster high resolution data; the data archived should be of the best quality achievable within the limited resources available; the data should be suitable for science use and publication by the world-wide scientific community;

1

Cluster Active Archive: Overview

5

and the CAA will provide user friendly services for searching and accessing these data and ancillary products such as orbit information. The implementation of the CAA has two important aspects: (1) In view of the shortage of manpower in most institutes processing Cluster data, ESA-supported manpower has been deployed in institutes where the relevant expertise exists, to assist in the preparation, validation, and documentation of the high-resolution data to be deposited in the archive, and (2) each PI team matches the level of ESA supported manpower with an equal level of effort from their own National Funding Agency.

1.1.3 CAA Standards The CAA standards form the basis of a consistent approach to the formatting and description of the digital data products being handled by the CAA. This is vital for the development of standard data handling tools and to maximize the ease of use of numerous products that are provided by many different groups. The lack of standardization in the past has proved a major impediment to the full exploitation of data, and every effort is being made to avoid this in the case of the CAA. In the early phase of the CAA two important working groups were established for standardization: one for the definition of the data format [1], and the other for the creation of a metadata model and dictionary [3].

1.2 Infrastructure The CAA project, managed by ESA, draws upon expertise from within ESA, the Cluster instrument teams and other groups. This combination encompasses the full range of skills and experience in handling the diverse and complex Cluster products that is required to achieve the CAA goals.

1.2.1 Organization The structure of the project team in summer 2008 is presented in Fig. 1.1. The core team consists of a project manager, technical manager, system engineer, and two system/software developers. The project has received support from the agency in the form of a Young Graduate Trainee (YGT) for 2 years. The Cluster project scientist and his deputy provide additional scientific support. The remaining CAA resources are deployed within the instrument teams. This is a key aspect of the CAA approach and is intended to ensure timely delivery of the best quality data from all the Cluster instruments. Some important supporting data are received from

6

H. Laakso et al.

Fig. 1.1 CAA management structure

the Cluster Joint Science Operations Centre (JSOC) that provides the CAA with data catalogues of orbital parameters of the predicted scientific events. The European Space Operations Centre (ESOC) provides the CAA with a wide range of spacecraft operations related information, which could be helpful as background information for science data processing; for further information, see Volpp and Sieg [Chapter 13].

1.2.2 System Architecture A key requirement of the CAA system architecture is that the system be scalable as the amount of data increases over time, affecting all hardware components, including servers, storage, backup system, and network. The current hardware components are sketched in Fig. 1.2. The total volume of compressed data products to be handled by the CAA is expected to exceed 50 TB, built up over the course of the mission. The CAA will keep all data available online to ensure the users have rapid access to the data they request. To provide the necessary flexibility and performance, the data storage is implemented via a storage area network (SAN). The SAN uses Fibre Channel network technology, which is well suited to the small, closed environment of the CAA system. This SAN is local to the CAA system, and is not part of a wider departmental SAN infrastructure. Storage is implemented on 7 RAID (Redundant Array of Independent Disks) arrays, each connected to the CAA servers via a dedicated fibre-channel infrastructure that includes fibre-channel “fabric” switches. These RAID arrays use highly cost-effective SATA (Serial Advanced Technology Attachment) disks, up to 250 GB each, but logical volumes are created by a hardware controller on each device, to create much larger volumes. The volumes are generally built as RAID 5, the most recent as RAID 6, in most cases with a hot spare. Each RAID array is divided

Cluster Active Archive: Overview

Fig. 1.2 Sketch of the CAA system in autumn 2008. The system has seven servers and most disks are directly mounted to the main data processing server (CAA-2), and the data catalogue disks are NFS mounted to other servers. CAA-5 is the access point to the users and data providers

1 7

8

H. Laakso et al.

into volumes of approximately 2 TB, in order to be manageable for backups. There are currently 21 volumes in use, giving approximately 42 TB of data connected to the server. In addition, several servers have significant internal storage, bringing the total data storage to approximately 45TB. The main data processing and database handling operations are undertaken by two (Ubuntu) Linux servers (caa-2 and caa-3 in Fig. 1.2), both having two Quadcore Intel Xeon 2.33 GHz processors. The systems share data using NFS (Network File System) and access rights using NIS (Network Information Service). Additional servers (containing 8 Intel Xeon processors and 4 Pentium processors) host the main web interface and development activities, and data backup services. The recent addition of a dedicated web development server allows development in parallel to operational systems. The CAA backup solution is based on an Overland Neo 4100 system with 2 internal drives and 24 TB tape capacity (58 tapes online from a pool over several hundred, each tape holding between 200 and 400-GB) based on LTO3 (Linear Tape Open) technology. The library is connected via fast SCSI to an Intel based Linux server, running the commercially licenced BakBone Netvault software. This system handles incremental backups, full backups, and periodic data archiving requirements. A number of network issues had to be taken into account in implementing the CAA at ESA. The CAA machines are inside a firewall protected part of ESA grid that provides high speed access to the international academic networks. Significant volumes of data are transferred from the instrument teams and ESOC to the CAA, and later the users request these data, all of which require good network connectivity between the CAA and the public internet. For this network, a dedicated inter-connect between systems has been implemented, before that is connected to the world. Interconnectivity of the servers is essential as the tasks are distributed to various servers. For this purpose, a dedicated local network was implemented in order to allow secure local communications between systems, without loading the DMZ (demilitarized zone) or exposing internal data such as backups, potentially to the world. The application software at the CAA has been partially developed by the CAA core team and partially in the PI institutes or other involved institutes, all of which needs to be supported by the CAA system. For this reason Linux was selected as the CAA operating system, as many instrument teams use Linux in one form or another. The services and data sets available from the CAA have grown and as a consequence have increased the requirements on the system performance. The CAA processing system handles various tasks, such as data management; data delivery; data processing for those instruments that provide software rather than pre-processed products; data ingestion into the CAA database; and data manipulation and visualization.

1

Cluster Active Archive: Overview

9

1.2.3 System Reviews Formal project reviews are held annually to verify that the key milestones have been met, that the technical aspects of the project are sound, and that the activities are progressing on schedule and within budget. The following project reviews have been held: System Specification Review in November 2003, to approve the user require-

ments, to review conceptual design, to review proposed PI inputs, to agree the archive implementation plan, and to release instrument team contracts Implementation Review in May 2005, to approve the archiving plans of the experiments and the technical implementation of the overall system First to second Operations Review in May 2006 and 2007 to review the status of data delivery and CAA services for years 2001–2002 and 2003–2004, respectively Third Operations Review in May 2008, to review the first part of the CAA activity (data availability from years 2001–2005) Fourth Operations Review in May 2009, to review the CAA activity for the first part of the second mission extension (data availability from years 2006–2007)

In future the operations reviews will be held annually as before. In addition in February-March 2009 an extensive peer review process was carried out to evaluate the usability and quality of the CAA database and its services to the users. More than 200 recommendations were obtained that are now under implementation.

1.2.4 User Access Statistics The CAA opened its services to the science community in February 2006. The evolution of the number of registered users and their use of CAA is shown in Fig. 1.3. At the time of opening, nearly 100 people had registered to support beta testing of the system. The top panel of the figure shows there has been a consistent rise in the number of registered users, approximately 20–30 new users per month, as shown in the middle panel, until there are currently about 800 registered users. The monthly number of user logins to the CAA website is shown in the second panel from bottom. The current average rate is around 450–500 logins per month. There is a clear increase by approximately 50% over the last year, which is in line with the increasing number of active users shown in the second panel from top that shows the number of those users who have logged in the CAA, either via web or command line interface. The bottom panel in Fig. 1.3 shows the monthly download volume. This is not expected to increase drastically because many institutes have implemented centralized data servers for the CAA data so that files are downloaded only once unless a new version of data file is ingested into the CAA (which can be automatically

10

H. Laakso et al.

Fig. 1.3 User statistics of the CAA. The panels from top to bottom are: the total number of registered users, the number of active users per month (i.e. users who have logged in the system), the number of new registered users per month, the total number of logins per month through the web-site, and the total monthly download rate in GigaBytes

checked via a command-line request). The download minimum in December 2007 was caused by a system failure whose consequences required several weeks for recovery and the statistics of download volume was lost. Within spring and summer 2008 the CAA will received plenty of new data as well as many of the old datasets have been recalibrated and will be ingested into the system which will likely raise the downloading rate.

1

Cluster Active Archive: Overview

11

1.2.5 User Feedback The CAA is a new type of data archiving activity for the Sun–Earth science community which, for the first time, provides this community with access to all calibrated full resolution data from a space mission. Therefore it is planned that both the data and the services of the CAA will significantly improve over time. Here the scientific community can make important contributions by using the data and the system and providing feedback to the CAA suggesting where the improvements, changes and additional features are desirable. To facilitate this feedback, the CAA has opened a user feedback area at http://caa.estec.esa.int/caa/feedback intro.xml where users can find two kinds of forms for reporting problems or sending suggestions. The first form is technical where users can report on missing data, speed of the system, and so on. The second form is for scientific issues where users can tell the CAA team what they want from the CAA system.

1.3 Standards The Cluster Exchange Format (CEF) is a self-describing ASCII format consisting of a metadata header followed by data represented as a comma-separated list of values. This was recommended by the Cluster Science Data System (CSDS) Archive Task Group for the exchange of science data between instrument teams of the Cluster mission. Adoption of a single common format allows the delivery of data products to users throughout the whole science community, including future scientists without the need for specific access software. The CEF format provides storage of science products in a robust and easily accessible form. The complexity of some high-resolution data products led the CAA design team to produce an extended version 2 of the CEF as the format to be used in the CAA [1]. Earlier versions of this syntax are referred to as CEF 1, and the syntax specified for the CAA is designated CEF 2. The data description or metadata consists of (1) a data model that is a logical framework into which the metadata information is placed, and (2) a data dictionary that is an ensemble of words used to populate the data model. The data model and data dictionary define the semantics used to describe the data. The development of the standard CAA data dictionary [3] was a major activity during implementation phase since it is a vital input to the development of the instrument team products. However, the dictionary has been updated several times during the operations phase.

1.3.1 CEF Data Format Files in the CEF syntax must have names with the extension .cef to assist identification. An exclamation mark is used as a comment marker, and all input to the right

12

H. Laakso et al.

of this marker up to the new line character is ignored on input. A header must be included, preceding the data records, and provide sufficient metadata to fully describe the data and their formatting. Use of detached headers is supported to minimize duplication. The file syntax and header content of a CEF file is specified in Allen et al. [1]. Entries in a record are comma separated and white space surrounding delimiters is ignored. Missing parameter values must be padded by a fill value that must be specified in the metadata. All data files are tabular and homogeneous – they have a sequence of records each with the same variables in the same order. Each record is ended by END OF RECORD MARKER, which by default is a new line character (nn) but can be set to another character in order to allow multiline records. The END OF RECORD MARKER, tab characters, carriage return (nr), new line and white space characters at the start and end of records and surrounding delimiters should be ignored when read. Thus, data may safely be formatted with white space and end of line markers for human (text editor) readability, allowing for easier exchange between platforms where the end of line marker is variously nrnn under DOS, nn under Unix and nr under Mac OS. For time series the records are ordered on the monotonic increasing time variable, and the time tag must be the first entry in a record. The other entries in the record are associated with that time via the metadata. Time tags are usually centered within the sampling interval, and the spacing between time tags is often the same as that sampling interval. Otherwise, Delta PLUS and Delta MINUS metadata describe the sampling or integration interval corresponding to the data and the location of the time tag within that interval. Time is represented as a text string in the strict ISO 8601 ASCII calendar date time format: yyyy-mm-ddTHH:MM:ss.wwwZ where the trailing ‘Z’ is UTC designator. For example, 2001-02-01T01:23:00.000Z corresponds to 01 February 2001 at 01:23:00 UT. This format permits any number of digits after the decimal point in the seconds field, e.g., the example above is a valid time string to millisecond accuracy. This format is robust for arbitrary timing accuracy, but software developers must take care to ensure that their software retains the accuracy. Multi-dimensional variables follow the natural ordering of the C Programming Language (i.e., the last index varies the fastest), and one entry is required for each element. For a velocity vector one has three entries and for a 2 by 2 array four entries, while a 16 by 8 by 24 array (e.g. for a distribution function) has 3,072 entries. It must be recognized that some data products may result in extremely long record lines. The rigid record structure to the data is intended to allow simple generic software to handle all instrument files.

1.3.2 CEF Metadata Metadata must contain all the information required to read and interpret the data (syntactic description), and to understand what the resulting numerical values

1

Cluster Active Archive: Overview

13

Table 1.1 Hierarchical levels for metadata Level Description Mission Information relevant to the whole mission Observatory One of four spacecraft (Cluster-1, Cluster-2, Cluster-3, and Cluster-4) or Cluster-Multi that contains datasets that contain data from more than one observatory Experiment Each spacecraft has 11 experiments, each identified by its Principal Investigator, plus the auxiliary data Instrument Some experiments consists of more than one physical instrument, each of which produces its own datasets; for two of these, CIS and STAFF, for the purpose of archiving it has been found convenient to identify two instruments. Thus CAA contains datasets from 4(11 C 2) D 52 instruments Dataset Each instrument produces one or more datasets; this level of metadata is common to the whole of each dataset Parameter A dataset contains one or more parameters, each of which has its own metadata and is also common to the entire dataset File Each dataset is composed of files, the number of which will grow with time during the lifetime of the CAA

represent (semantic description), including how the data were obtained. The purpose of the CAA Metadata Dictionary [3] is to describe fully the required CAA metadata information and to explain how that information must be formatted so as to be exploitable by the generic CAA software. Table 1.1 shows different concepts in the CAA that are hierarchically organized. An example of CEF metadata is given in Appendix. Three forms of metadata are recognized: File metadata are used to provide information unique to the file such as filename,

time coverage and provenance. Global metadata is applicable to the entire contents of a dataset, and fulfils the

same purpose as the global attributes in a CDF file. Support metadata (called variable metadata in the CEF) describes an individual

parameter in a dataset, and is equivalent to the CDF variable attributes. Support metadata blocks must appear in the CEF header in the same order as the corresponding parameters appear in the data records. Apart from this, there is no required order for the global metadata or parameters blocks within any metadata block although usually the global metadata are followed by the parameter definitions. Apart from quoted text data and variable names, metadata information is case insensitive. The open format also allows for the inclusion of any extra metadata deemed desirable by the generating team. Lines of metadata may be continued using ‘n’ as a continuation marker, following one of the commas separating a list of values. A metadatum consists of a text string of the form “KEYWORD D value”. The keyword values can be of three types: (1) a single data value (which may be numerical, logical, enumerated text string, formatted text string, free text string, ISO time code), (2) a comma delimited sequence of data values or (3) the name of a data variable in the same file carrying the requisite information.

14

H. Laakso et al.

1.4 Software System The CAA is built around a MySQL database and XML-based data management system, which provides a robust and extensible framework for handling the data products delivered to the CAA. There are two routes of accessing the data products: web GUI and command-line interfaces that are described in the following section.

1.4.1 User Interface The access point to the CAA database and services is the website at http://caa.estec. esa.int/ (Fig. 1.4). Part of the information on the website is available to all visitors, but the access to the main data and graphical products requires a login into the system. Therefore users must first register to the system. The registration procedure is fully automatic so that the user defines the password (which can be changed later if wished) but the user-id is determined by the system and is sent to the e-mail address given during the registration process. The registration allows the CAA to gather usage statistics (see Section 1.2.4) and more importantly to monitor the system performance in order to improve it, when and where necessary. In addition, the users can save their configurations of datasets or plots for using them during future requests. Furthermore, occasionally poor-quality data products may have been delivered and ingested into the system, and in such cases the CAA can find from the usage database who had requested those data and these users can then be informed about the presence of better quality products.

1.4.1.1 Web Interface Figure 1.5 shows the front page of the CAA website after a user has successfully logged into the system. The user can now find USER AREA in the left-hand menu bar that currently holds user-defined lists of profiles that are useful when searching or plotting certain variables regularly. In “Modify User Details” one can modify user details such as password or contact e-mail address. Currently the following data services are provided (see the left-hand menu bar): CAA Data Download Area: allows users to perform data requests, using filters

given in the entry page. CAA Graphical Products: allows users to display some of variables. CAA Quick-look Plots: allows users to view pre-generated quick-look plots that

are based on non-validated data sets processed by instrument teams or CSDS. CAA Inventory Plots: allows users to view the completeness of the database.

1

Cluster Active Archive: Overview

15

Fig. 1.4 The front page of the CAA website (http://caa.estec.esa.int/) in early 2009

There are currently a little more than 200 datasets from each spacecraft in the CAA data repository. By setting values for Start Time, Stop Time, Observatory, Experiment, Measurement Type, and Instrument Type, one can filter datasets to be shown on the GUI, thus speeding identification of the datasets needed by a user. Some, all or none of these selections may be set. For instance, if only Start and Stop Time are selected, then all the datasets that exist for this interval will be shown. If Time fields are left empty, this defaults to the time range of the entire mission.

16

Fig. 1.5 The front page of the CAA website after login into the system

H. Laakso et al.

1

Cluster Active Archive: Overview

17

For Observatory, the option Cluster Multi refers to the products that contain data from more than one spacecraft. Measurement Type is a general description of the type of measurement. Instrument Type supplies information about the nature of the instrument used to obtain the data. Often the same physical parameter can be obtained from more than one type of instrument. For example, electron density can be derived from particle detector, Langmuir probe, wave spectral or active resonance sounder data. After the selection, available datasets are listed by Observatory (see Fig. 1.6). Clicking “Expand/Collapse All” will toggle the expansion/collapse of the results. When expanded, the datasets available are listed. The following details are provided for each dataset: Dataset Name: with check box beside it to indicate if the user wishes to download

this dataset. Notice that this Dataset Name needs to be known for the use of command-line interface (see next section). Result Timespan: tells the timespan where this dataset is available on the CAA. Note that this does not tell anything about the coverage of the dataset which can be examined via inventory plots. Dataset Details this link opens up a new window with the dataset metadata details, e.g., variables available in the dataset. Users must select each dataset they wish to download by clicking on the checkbox. Currently there is a limit of about 1 GB on the amount of data one can download at one time. This limit is necessary because the web request is executed immediately and larger requests may time-out the Internet connection before completion. The requests of larger amounts of data can be done via command-line interface (see Section 1.4.1.2), which supports batch operation and currently has a limit of 50 GB. After selecting the datasets, possibly modifying the time interval and then clicking on “Search”, the available datasets will be listed, including an estimated total size of uncompressed files given in brackets. There is also again a link to the dataset information. Here the users must select each dataset they wish to download, by clicking on the checkbox, and then click “Download”. In addition the user may also choose to download the files in a single file (default) or in shorter intervals (e.g. hourly or daily files). Next the files are combined and compressed into a single zip file. Once this process has completed, the new page shows a link “Search results ready for download”. Selecting this link, the user will be prompted for their local directory where the file will be stored in the .zip format.

1.4.1.2 Command-Line Interface During year 2007 a command-line tool was introduced which allows users to bypass the graphical interface to gain faster access to the data products. This facility is utilized to make automated enquiries about the arrival of new files on the CAA database and also for data analysis and visualization software tool to gain direct

18

H. Laakso et al.

Fig. 1.6 Results of searching datasets. For each experiment the datasets are grouped into smaller sets helping the faster identification of the right dataset

1

Cluster Active Archive: Overview

19

access to the CAA database. For example, when new files have been ingested into the system, a user can run a script that will automatically transfer the appropriate files from the CAA to the users’ own local systems. The data download volume via the command-line interface is three times greater than via the web forms interface, and it supports batch operation so that large jobs can be queued for offline processing. McCaffrey [4] defines the format, syntax and mechanism of this tool that allows users to submit requests from a command line using the GNU wget tool [GNUWGET]. The non-CEF files delivered (e.g., for graphical products) are direct copies of the original files, whereas the CEF files delivered cover exactly the time span specified in the request. If the requested parameters are correct, and there are data available, the request may not be executed immediately if there are many other jobs currently executing or if the request is large. When a request is scheduled, the user will receive a file called CAA Info.log.X that indicates a URL where the completed file can be retrieved. The following example URL is a request of three CEF products, covering an interval of four days to be split to daily files: http://caa.estec.esa.int/caa query/?uname=uname&pwd=pwd&dataset id=C1 C P EDI EGD,C1 CP EFW L3 P,C1 CP FGM FULL&time range=2001-02-01T00: 00:00Z/2001-02-05T00:00:00Z &file interval=1day where C1 CP EDI EGD, C1 CP EFW L3 P, and C1 CP FGM FULL are the datasets for electron gyroperiod, 4-s averages of spacecraft potential and fullresolution magnetic field, respectively. Notice that CEF is a default format, but one can also request data in CDF format. The job size is over the ‘immediate execute’ limit, and will therefore be scheduled. When complete, the software creates a zip file and alerts the user via email that the file is available for download (if the parameter ‘notify’ is not specified, a notification email is sent as default).

1.4.2 Visualization The CAA provides three visualization utilities. The main graphics tool is “CAA Graphical Products” that displays variables using the calibrated CAA data products. The “CAA Quicklook Plots” are produced outside the CAA and may be based on quick calibration routines. “CAA Inventory Plots” provide a view on the coverage of the CAA data products.

1.4.2.1 CAA Graphics At http://caa.estec.esa.int/caa/graphics.xml one can display some variables that are either variables in datasets (e.g. electric or magnetic field components) or results of some data manipulation (e.g. omni-directional energy spectrograms for electrons

20

H. Laakso et al.

Fig. 1.7 Selecting variables for graphical display. At the moment there are about 100 variables for each spacecraft that can be displayed. Selected variables can be saved as a profile for future use

and ions). Figure 1.7 shows the variables of Whisper available for graphics. The list of selected variables are displayed in a window where the order of the variables can be changed, and if the set of variables is frequently used, one can save it as a profile for future use. The graphics are based on IDL software libraries that are used to read, manipulate and plot the variables. These scripts are made available for download from the CAA website (see software in the menu bar). Figure 1.8 shows an example of the plot for a 4-h interval on 17 March 2001. Under “Plot Type”, one can find four types of plots, including pre-generated 1h, 6 h and 1 day plots that are created when new data files are ingested into the system. Selecting “On demand”, one can specify any interval (up to one orbit) for which the selected panels will be displayed.

1

Cluster Active Archive: Overview

Fig. 1.8 Example of graphical display for a 4-h interval on 17 March 2001

21

22

H. Laakso et al.

1.4.2.2 Quick-Look Plots The quick-look plots, produced by the PI teams or the CSDS, may be based on simple calibration routines and are generally not suitable for scientific analysis or publication without prior discussion with the producer. For CSDSweb plots, please note that the plot availability on the CAA is delayed. The complete set of plots is available at the CSDSwebsite (http://www.cluster.rl.ac.uk/csdsweb-cgi/ csdsweb pick). The quick-look plot utility also allows browsing of the pre-generated WBD (Cluster Wide-Band Plasma Wave Investigation Experiment) science plots. The WBD plots can also be retrieved through “Search”. The overview plots can be downloaded in both postscript and png format under Cluster-Multi Observatory whereas the high-resolution (30 s) WBD gif plots are produced for each spacecraft individually and can be downloaded under each spacecraft. 1.4.2.3 Pre-generated Inventory Plots The inventory plots give an overview of the coverage of the database for a number of datasets for each instrument and spacecraft. During the ingestion process, each data file is examined and any time step longer than 5 min is recorded as a data gap and is shown in the inventory plot. The data used by the inventory plots are generated subsequent to ingestion, and the inventory plots are produced later usually at the end of the day, and so there may be a delay between ingestion (i.e. the time when the data become available for downloading or plotting) and an update of an inventory plot. The following three colours are used to indicate the data content: Blue: the data exist (notice that the data can be fill values) Purple: there is a data gap White: no data exists in the database

In addition the top two panels tell the presence of predicted events (top panel) and the spacecraft mode (second from top) where blue is used for normal mode and red for burst mode (higher data rates).

1.4.2.4 User-Defined Inventory Plots Users can also produce their own inventory plots over intervals that are longer than 6 h in the data search area. Selecting one or more datasets and clicking then on “INVENTORY”, one gets an inventory plot that shows the data coverage for the selected time interval. Notice that these plots are generated on the fly using the current ingestion status. So some data may appear on these that are not yet on the pre-generated inventory plots.

1

Cluster Active Archive: Overview

23

1.4.3 Support Tools There are some software tools available for the users under “Software” in the lefthand menu bar. The software, called Qtran (Queen Mary File Translator), can translate files between CDF and CEF. Notice that if a user requests to download a CEF file in CDF format, this is the tool used to perform the conversion. The same software is also able to read other record-oriented ASCII data files and may help with translation from ASCII into the recommended exchange format. The QSAS (Queen Mary Science Analysis System) software [Allen, Chapter 14] can manipulate CEF files and visualize multi-instrument and multi-spacecraft products. Both the Qtran and QSAS software are maintained at the Imperial College of London. The CAAtools package provides a set of tools for checking the products against the metadata dictionary and CEF specification. The CEF product validation part of the package is made available for download, and the library is also used by the CAA Matlab toolbox for reading CEF datasets. For Matlab users there are two toolboxes for reading, writing, manipulating and plotting CEF files. The function cefRead can be applied directly to gzipped CEF files. Documentation and the Matlab source files are provided under Software. For IDL the CAA has developed two packages that simplify the process of reading and plotting data. The cef read procedure provides a single command to load a CEF file into IDL. Options include to load only a sub-set of variables or to restrict the reading to a set of time intervals. The software will also handle conversion of the time information into MJD (Modified Julian Date) or CDF epoch formats. The documentation provides the full details of the utility as well as several examples of different reading options and a full example of generating multi-panel plots of simple scalar and vector parameters. For users not wishing to do their own plotting in IDL an additional package provides an extensive set of options to generate high quality plots of CEF products including scalars, vectors and multi-dimensional arrays. The software makes use of the cef read software described above and is based on the software used within the CAA to provide pre-generated and on-demand graphical products.

1.5 Science Data Products The CAA holds three types of science data files: calibrated and uncalibrated data files from the instrument teams that are available in CEF format; raw data files received from the ESOC; CSDS primary and summary parameters (PP and SP, respectively) data files. The CSDS files are preliminary survey data and not calibrated to the level done within the CAA and are so not suitable for publication without discussion with the relevant instrument team.

24

H. Laakso et al.

1.5.1 CAA Products The key CAA science data products are the datasets that have been calibrated and validated within the CAA. Each team performs detailed calibration of the data to the level suitable for optimum scientific return. However, all data cannot be validated manually, and thus the data files can still contain features that represent instrumental effects rather than physical processes. It is therefore important that the users read all documentation and caveat information related to the data and contact the CAA for specific questions or problems. At the moment there are over 200 different datasets from each spacecraft, and even in this phase of the project new datasets can still appear into the system. In summer 2008 approximately the first 5 years (2001–2005) of observations have been ingested, and the following 4 years (2006–2009) will become available before summer 2010. In addition, large-scale reprocessing of some products is planned to improve quality and consistency, which highlights the benefit of the cross calibration working group, which has helped to identify and solve calibration issues. The total number of CEF and other format data files that are currently active (i.e. latest version) is in excess of 3.5 million with a total uncompressed volume equivalent to over 40 TB. The pre-generated plot panels amount to more than twice this number of files although they require much less disk space. In addition there remains a large volume of raw and some Level-1 data that are not yet available via the CAA catalogue system but that are being added as time and resources permit.

1.5.2 Cluster Raw Data System For the second extension of the Cluster mission, starting in January 2006, there was a change in the distribution of the raw data. Earlier, the raw data collected by ESOC were transferred to CDROM (Raw Data Media – RDM) and mailed to about 70 recipients. Now the data are transferred over the network from ESOC to the CAA from where they can be accessed by the users. The CAA is also working to provide a full online archive of the RDM from the entire mission by copying the existing CDs. Notice that these data primarily useful to the instrument teams that are able to decommutate the data content and produce the instrument products in physical units. Currently the CAA does not provide access to the PI software. For automated access of the RDM data, one can use HTTP GET method either within a web browser address bar or using a command line based http browser (e.g. wget) tool: http://caa.estec.esa.int/cgi-bin/caa rdm?key1=value1&key2=value2&. . . Both the keyword and the value are treated as case insensitive. The files delivered are direct copies of the contents of the original RDM volume. The details of the utility including the keywords and their values are explained by Perry [5].

1

Cluster Active Archive: Overview

25

1.5.3 CSDS Products The raw Cluster data files are transferred from ESOC to the national Data Centres (DCs), previously on CD-ROMs, and since January 2006 through the network via CAA. The national DCs process the raw data using PI software and produce physically meaningful parameters (fields, particle fluxes, temperatures, spectral densities, etc.) for those experiments for which the DC is responsible. The results are stored as CDF files. These files are then exchanged with the other national DCs so that each one has a full set, which is then made available to the scientific users in that country. The interfacing between the users and the DC, as well as among the DCs themselves, is the CSDS Data Management System (DMS), a software package that has been developed especially for this purpose. A number of JSOC (Joint Science Operations Centre) products (e.g. predicted orbit and event files, scientific events catalogue) are also made available to the local users via the national DC. The key CSDS datasets are Prime Parameter Data Base consists of 65 parameters from all four spacecraft

averaged over one spin (4 s). Summary Parameter Data Base consists of 86 parameters from only one space-

craft and averaged over 1 min; auxiliary parameters such as position and spin orientation are also included. Event Catalogues provide data concerning predicted and observed scientific events for the Cluster mission. CSDS auxiliary data contains such information as tetrahedron quantities, orbit number, position, velocity and attitude of spacecraft. All the vector quantities are given in the GSE coordinate system. The rotation angle between the GSE and GSM is also given, enabling a straightforward and unambiguous conversion. CSDS products that can be found at http://sci2.estec.esa.nl/cluster/csds/ ring.html were not initially part of the CAA. However, now these products are being transcribed into CEF format files and ingested in the CAA and can be used in combination with CAA products.

1.6 Cross-calibration Activity The CAA cross-calibration working group is actively supported by the instrument teams with two to three annual workshops; for the list of the past workshops, see http://caa.estec.esa.int/caa/cross-cal.xml. This activity is invaluable for assessing and improving the quality of the datasets. Although the active participants in the working group come mainly from the instrument teams, anyone interested in the calibration issues of various measurements are welcome to participate in these workshops.

26

H. Laakso et al.

The key issues in the meetings have been: To improve the calibration of dc magnetic field measurements much better than

1 nT To determine the total plasma density using various measurements techniques To calibrate the dc electric field measurements using and comparing various data

sources To determine the E B-drift velocity using and comparing various data sources To compare energy fluxes of electrons and ions between different experiments To compare power spectral densities of wave measurements, including both elec-

tric and magnetic field wave components

1.7 Future Planning The CAA development will continue to enhance the core functionality of the system, to reprocess existing data files (2001–2005) and to expand the coverage of the datasets to the next years (2006–2009). There will be further updates to both the web and command line data request mechanism to support the selection and delivery of the non-CEF formatted data products. These products include pre-generated graphical products supplied by the instruments teams, calibration information and reformatted raw data. The activities planned for the period leading up to the next operations review in May 2009 are as follows: Routine operations, data delivery and maintenance of the existing data access and

value-added services Production and delivery of data files from years 2006–2007 before May 2009 Monitoring system of the file ingestion processes in order to display the status

of ingestion as well as to monitor the performance of the system in the validation/ingestion procedures Development of advanced documentation service for searching and browsing the Cluster documents, i.e., a utility to search for specific reports (e.g. ESOC spacecraft anomaly information) Improvements to user interface based on user feedback and problem reports. ı Additional search and filtering options for dataset selection ı Provision of the detailed data coverage information as a CAA product (to be used by external tool developers) ı Updates to the command line interface in response to requests from tool developers ı Provision of a simple and easy to use interface Value-added software development activities

ı Adding new plotting options to the CAA graphics menu ı Provide a user selectable option to automatically include referenced products such as caveat information and documentation

1

Cluster Active Archive: Overview

27

In addition some progress has also been made on the translation of CAA metadata into Space Physics Archive Search and Extract (SPASE). Automated translation of simple products is straightforward but differences between the CAA and SPASE parameter specification increase the difficulty of translation as the dimensionality and complexity of products increase. Discussions are underway with the NASA Virtual Magnetospheric Observatory at UCLA to improve compatibility.

Appendix: CEF Metadata The users are strongly advised to become familiar with the CEF metadata dictionary given by Harvey et al. [3] in order to get a full benefit of the metadata provided with the numerical data. This appendix points out some of the metadata details that users should be aware of. We have included an example of metadata from the FGM SPIN resolution data file on 1 January 2003, 0–1 UT, which is given in the end of this appendix. The user should be aware that the delivered files to the users and the actual source files in the CAA database are not identical. The source files can be of any length and most of the metadata are usually given in detached header files. When the data are requested by the user, CAA software tools are used to create a file of the desired interval from the source files and the metadata are included in the file. The FILE NAME starts with the dataset name and the requested interval followed by the version number that tells the latest ingestion date of all input data files used for the requested file, which is 2006 August 15 in the case of the example file. The top-level metadata provides some general level information about the mission and the experiment: MISSION: all MISSION metadata are the same for all datasets and are provided

by a detached MISSION header file. OBSERVATORY: There are four OBSERVATORY header files, one for each

Cluster spacecraft. These metadata are the same for all datasets from the same spacecraft. EXPERIMENT and INSTRUMENT: these metadata provide some details of the experiment and are identical to all the datasets of one experiment on one spacecraft. INSTRUMENT CAVEATS: the instrument level general caveats are given here, the relevant dataset is given here. This may include references to external files. The important part of metadata are given at the dataset level and are global to the all files of the given dataset. Some of the useful keywords are: DATASET ID is a unique identifier of the dataset, which is also used in the be-

ginning of the filename. This ID is needed when requesting data via the command line interface. DATASET TITLE is a concise description of the dataset. DATASET DESCRIPTION provides a detailed description of the dataset.

28

H. Laakso et al.

TIME RESOLUTION is a time interval between two data points, expressed

in seconds. If this is not constant, the two further keywords, MIN TIME RESOLUTION and MAX TIME RESOLUTION, must be used. PROCESSING LEVEL tells the level to which the data have been processed. Possible values are: Raw, Uncalibrated, Calibrated, Derived, and Auxiliary. DATASET CAVEATS provides miscellaneous information about the dataset. This may include references to external files. LOGICAL FILE ID is the same filename. VERSION NUMBER is the date of the latest ingested file used to create the requested file. DATASET VERSION lists all the individual files used to produce the requested file. FILE CAVEATS provides specific caveats of individual files. This may include references to external files. FILE TYPE is normally CEF but for plots and some other files it can be something else. In principle any type of file including binary files can be included in the CAA database as long as a proper detached header is provided. FILE TIME SPAN tells the requested time interval.

Each parameter is described with a number of useful keywords where the most important ones are: CATDESC provides a concise description of the parameter. COORDINATE SYSTEM provides an acronym of a coordinate system (for

vectors, tensors or their components). Allowed acronyms are provided by Harvey et al. [3]. DELTA PLUS (DELTA MINUS) provides the number of units (of the parameter) to be added to (or be subtracted from) the nominal value to obtain the upper (or lower) limit of the parameter interval within which the data was acquired. This is used in particular with time tags to provide the time interval where the data value is valid. DEPEND 0 ties explicitly the parameters to the independent variable parameter(s) on which it depends. FIELDNAM describes the parameter and which can be used, for example, to give a title for a plot (use LABLAXIS for the axis label). This could be the same as CATDESC. Note, however, that while FIELDNAM is optional, CATDESC is not. FILLVAL is used to replace bad or missing data. FRAME VELOCITY is important to the physical parameters whose values depend upon the motion of the coordinate system in which they are measured. Three possible values are accepted: Observatory (default); Inertial (i.e. GEI); EarthCorotating. LABEL i is used for multi-dimensional parameters to label individual components (as LABLAXIS is not sufficient to describe the variables).

1

Cluster Active Archive: Overview

29

LABLAXIS can be used to label the y-axis of a plot or to provide a column

heading for a data listing. For more than one-dimensional parameters, the labels can be given in LABEL i. QUALITY provides a quality of the parameter, with the following values: 0 – Not applicable; 1 – Known problems, use at your own risk; 2 – Survey data, not for publication; 3 – Good for publication, subject to PI approval; 4 – Excellent data which has received special treatment. REPRESENTATION is used to provide essential information describing vectors and tensors; for details, see Harvey et al. [3]. SI CONVERSION factor is required to take the archived value to the corresponding value in a standard SI unit. For instance, “1:0E 9 > T” is used to transform the magnetic field data stored in nanotesla (nT) into tesla (T). SIGNIFICANT DIGITS provides the number of decimal digits required to preserve the precision of the parameter and can have a value of any positive integer. SIZES provides the dimensions of the array that is required for any parameter represented by more than one component (vectors, tensors, spectral arrays, etc.). For example, SIZES D 3 is for a vector of 3 components. TENSOR ORDER: 0 for scalars (default value), 1 for vectors (e.g., magnetic field or velocity), 2 for tensors (e.g., plasma pressure). UNITS: this provides the unit of the parameter that can be indicated on the axes of a plot. If parameter has no unit, this must be specified by UNITS D “unitless”. VALUE TYPE provides identification of the value type (essential for ASCII conversion) with possible values: CHAR; DOUBLE; FLOAT; INT; ISO TIME; ISO TIME RANGE.

The following example shows the metadata for the FGM SPIN resolution data file on 1 January 2003, 0–1 UT, requested on 17 February 2009: FILE NAME = “C1 CP FGM SPIN 20030101 000000 20030101 010000 V060815.cef” FILE FORMAT VERSION = “CEF-2.0” END OF RECORD MARKER = “$” START META = MISSION ENTRY = “Cluster” END META = MISSION START META = MISSION TIME SPAN VALUE TYPE = ISO TIME RANGE ENTRY = 2000-08-16T12:39:00Z/2009-12-31T23:59:59Z END META = MISSION TIME SPAN START META = MISSION AGENCY ENTRY = “ESA” END META = MISSION AGENCY START META = MISSION DESCRIPTION ENTRY = “The aim of the Cluster mission is to study small-scale structures of the” ENTRY = “magnetosphere and its environment in three dimensions. To achieve this,” ENTRY = “Cluster is constituted of four identical spacecraft that will flight in a”

30

H. Laakso et al.

ENTRY = “tetrahedral configuration. The separation distances between the spacecraft” ENTRY = “will be varied between 40 km and 10 000 km, according to the key” ENTRY = “scientific regions.” END META = MISSION DESCRIPTION START META = MISSION KEY PERSONNEL ENTRY = “Philippe Escoubet>[email protected] >Cluster Project Scientist” END META = MISSION KEY PERSONNEL START META = MISSION REFERENCES ENTRY = “The Cluster and Phoenix Missions>Cluster project and” ENTRY = “instrument teams>Space Sci. Rev. 79, Nos. 1-2, 1997” END META = MISSION REFERENCES START META = MISSION REGION ENTRY = “Solar Wind” ENTRY = “Bow Shock” ENTRY = “Magnetosheath” ENTRY = “Magnetopause” ENTRY = “Magnetosphere” ENTRY = “Magnetotail” ENTRY = “Polar Cap” ENTRY = “Auroral Region” ENTRY = “Cusp” ENTRY = “Radiation Belt” ENTRY = “Plasmasphere” END META = MISSION REGION START META = MISSION CAVEATS ENTRY = “*CL” END META = MISSION CAVEATS START META = OBSERVATORY ENTRY = “Cluster-1” END META = OBSERVATORY START META = OBSERVATORY CAVEATS ENTRY = “*C1 CQ” END META = OBSERVATORY CAVEATS START META = OBSERVATORY DESCRIPTION ENTRY = “Cluster-1 (Rumba)” ENTRY = “Launched: 09 Aug 2000” ENTRY = “ESA Number: 1” ENTRY = “COSPAR ID: 2000-045A” ENTRY = “USSPACECOM catalogue number 26463” ENTRY = “CSDS Code: C1” ENTRY = “ESOC FD code: S1” ENTRY = “ESA Flight Model Number: FM5” END META = OBSERVATORY DESCRIPTION

1

Cluster Active Archive: Overview

START META = OBSERVATORY TIME SPAN VALUE TYPE = ISO TIME RANGE ENTRY = 2000-07-16T12:39:00Z/2009-12-31T23:59:59Z END META = OBSERVATORY TIME SPAN START META = OBSERVATORY REGION ENTRY = “Solar Wind” ENTRY = “Bow Shock” ENTRY = “Magnetosheath” ENTRY = “Magnetopause” ENTRY = “Magnetosphere” ENTRY = “Magnetotail” ENTRY = “Polar Cap” ENTRY = “Auroral Region” ENTRY = “Cusp” ENTRY = “Radiation Belt” ENTRY = “Plasmasphere” END META = OBSERVATORY REGION START META = EXPERIMENT ENTRY = “FGM” END META = EXPERIMENT START META = EXPERIMENT DESCRIPTION ENTRY = “Fluxgate Magnetometer” END META = EXPERIMENT DESCRIPTION START META = INVESTIGATOR COORDINATES ENTRY = “Andre Balogh>PI>[email protected]” ENTRY = “Elizabeth Lucek>PI>[email protected]” END META = INVESTIGATOR COORDINATES START META = EXPERIMENT REFERENCES ENTRY = “*CL CD CAA FGM ICD 0001 V0 1.pdf” ENTRY = “*CL CD FGM USERMAN.pdf” ENTRY = “http://www.sp.ph.ic.ac.uk/Cluster/” END META = EXPERIMENT REFERENCES START META = EXPERIMENT KEY PERSONNEL ENTRY = “Andre Balogh>PI>[email protected]” ENTRY = “Elizabeth Lucek>PI>[email protected]” END META = EXPERIMENT KEY PERSONNEL START META = EXPERIMENT CAVEATS ENTRY = “*CL CQ FGM CAVF.txt” END META = EXPERIMENT CAVEATS

31

32

H. Laakso et al.

START META = INSTRUMENT NAME ENTRY = “FGM1” END META = INSTRUMENT NAME START META = INSTRUMENT DESCRIPTION ENTRY = “FGM Experiment on Cluster C1” END META = INSTRUMENT DESCRIPTION START META = INSTRUMENT TYPE ENTRY = “Flux Feedback” END META = INSTRUMENT TYPE START META = MEASUREMENT TYPE ENTRY = “Magnetic Field” END META = MEASUREMENT TYPE START META = INSTRUMENT CAVEATS ENTRY = “*C1 CQ FGM CAVF” END META = INSTRUMENT CAVEATS START META = DATASET ID ENTRY = “C1 CP FGM SPIN” END META = DATASET ID START META = DATA TYPE ENTRY = “CP>CAA Parameter” END META = DATA TYPE START META = DATASET TITLE ENTRY = “Magnetic field, spin resolution” END META = DATASET TITLE START META = DATASET DESCRIPTION ENTRY = “This dataset contains spin resolution measurements of the magnetic field vector” ENTRY = “ from the FGM experiment on the Cluster C1 spacecraft” END META = DATASET DESCRIPTION START META = CONTACT COORDINATES ENTRY = “Elizabeth Lucek>PI>[email protected]” END META = CONTACT COORDINATES START META = TIME RESOLUTION ENTRY = 4 END META = TIME RESOLUTION START META = MIN TIME RESOLUTION ENTRY = 4.412 END META = MIN TIME RESOLUTION

1

Cluster Active Archive: Overview

33

START META = MAX TIME RESOLUTION ENTRY = 3.636 END META = MAX TIME RESOLUTION START META = PROCESSING LEVEL ENTRY = “Calibrated” END META = PROCESSING LEVEL START META = ACKNOWLEDGEMENT ENTRY = “Please acknowledge the FGM team and ESA Cluster Active Archive” ENTRY = “ in any publication based upon use of this data” END META = ACKNOWLEDGEMENT START META = DATASET CAVEATS ENTRY = “*C1 CQ FGM CAVF” END META = DATASET CAVEATS START META = LOGICAL FILE ID ENTRY = “C1 CP FGM SPIN 20030101 000000 20030101 010000 V060815” END META = LOGICAL FILE ID START META = VERSION NUMBER ENTRY = 060815 END META = VERSION NUMBER START META = DATASET VERSION ENTRY = “Merged file, the dataset version for each segment follows: ” ENTRY = “2003-01-01T00:00:00Z/2003-01-01T01:00:00Z, C1 CP FGM SPIN 20021231 092502 20030102 182906 V01” ENTRY = “ 01” END META = DATASET VERSION START META = FILE CAVEATS ENTRY = “CAA Merged File - $Id: cefmerge.c,v 1.23 2008/08/06 14:56:43 cperry Exp cperry $” ENTRY = “The file caveats for each segment follows:-” ENTRY = “2003-01-01T00:00:00Z/2003-01-01T01:00:00Z C1 CP FGM SPIN 20021231 092502 20030102 182906 V01” ENTRY = “Calibration parameters have been estimated using the Fourier analysis” ENTRY = “ method, solar wind analysis and the range jump correction method.” END META = FILE CAVEATS START META = FILE TYPE ENTRY = “cef” END META = FILE TYPE START META = METADATA TYPE ENTRY = “CAA” END META = METADATA TYPE

34

H. Laakso et al.

START META = METADATA VERSION ENTRY = “2.0” END META = METADATA VERSION START META = FILE TIME SPAN VALUE TYPE = ISO TIME RANGE ENTRY = 2003-01-01T00:00:00.000Z/2003-01-01T01:00:00.000Z END META = FILE TIME SPAN START META = GENERATION DATE VALUE TYPE = ISO TIME ENTRY = 2009-02-17T15:49:51Z END META = GENERATION DATE START VARIABLE = time tags C1 CP FGM SPIN PARAMETER TYPE = “Support Data” CATDESC = “Interval centred time tag” UNITS = “s” SI CONVERSION = “1.0 > s” SIZES = 1 VALUE TYPE = ISO TIME SIGNIFICANT DIGITS = 24 FILLVAL = 9999-12-31T23:59:59Z FIELDNAM = “Universal Time” LABLAXIS = “UT” DELTA PLUS = half interval C1 CP FGM SPIN DELTA MINUS = half interval C1 CP FGM SPIN END VARIABLE = time tags C1 CP FGM SPIN START VARIABLE = half interval C1 CP FGM SPIN PARAMETER TYPE = “Support Data” CATDESC = “Half averaging interval length” UNITS = “s” SI CONVERSION = “1.0>s” SIZES = 1 VALUE TYPE = FLOAT SIGNIFICANT DIGITS = 6 FILLVAL = -1e30 FIELDNAM = “Half width of averaging interval” LABLAXIS = “s” END VARIABLE = half interval C1 CP FGM SPIN START VARIABLE = B vec xyz PARAMETER TYPE = ENTITY = PROPERTY = FLUCTUATIONS =

gse C1 CP FGM SPIN “Data” “Magnetic Field” “Vector” “Waveform”

1

Cluster Active Archive: Overview

CATDESC = UNITS = SI CONVERSION = TENSOR ORDER = COORDINATE SYSTEM = REPRESENTATION 1 = SIZES = VALUE TYPE = SIGNIFICANT DIGITS = FILLVAL = QUALITY = FIELDNAM = LABLAXIS = DEPEND 0 = LABEL 1 = END VARIABLE = B vec xyz gse

“Cluster C1, Magnetic Field Vector, spin resolution in GSE” “nT” “1.0E-9>T” “1” “GSE>Geocentric Solar Ecliptic” “x”, “y”, “z” 3 FLOAT 6 -1e30 3 “Cluster C1, Magnetic Field Vector, spin resolution in GSE” “Mag Field” time tags C1 CP FGM SPIN “Bx”, “By”, “Bz” C1 CP FGM SPIN

START VARIABLE = B mag C1 CP FGM SPIN PARAMETER TYPE = “Data” ENTITY = “Magnetic Field” PROPERTY = “Magnitude” FLUCTUATIONS = “Waveform” CATDESC = “Cluster C1, Magnetic Field Magnitude, spin resolution” UNITS = “nT” SI CONVERSION = “1.0E-9>T” TENSOR ORDER = “0” SIZES = 1 VALUE TYPE = FLOAT SIGNIFICANT DIGITS = 6 FILLVAL = -1.0E30 QUALITY = 3 FIELDNAM = “Cluster C1, Magnetic Field Magnitude, spin resolution” LABLAXIS = “B” DEPEND 0 = time tags C1 CP FGM SPIN END VARIABLE = B mag C1 CP FGM SPIN START VARIABLE = sc pos PARAMETER TYPE ENTITY PROPERTY CATDESC UNITS SI CONVERSION TENSOR ORDER COORDINATE SYSTEM REPRESENTATION 1 SIZES VALUE TYPE SIGNIFICANT DIGITS FILLVAL

xyz = = = = = = = = = = = = =

35

gse C1 CP FGM SPIN “Data” “Other1” “Vector” “Position of Cluster C1 in GSE” “km” “1.0E3>m” “1” “GSE>Geocentric Solar Ecliptic” “x”, “y”, “z” 3 FLOAT 7 -1.0E30

36 QUALITY FIELDNAM LABLAXIS DEPEND 0 LABEL 1 END VARIABLE = sc pos

H. Laakso et al. = = = = = xyz

0 “Position of Cluster C1 in GSE” “Position” time tags C1 CP FGM SPIN “x”, “y”, “z” gse C1 CP FGM SPIN

START VARIABLE = range C1 CP FGM SPIN PARAMETER TYPE = “Support Data” CATDESC = “Cluster C1, FGM instrument range, defined on spin resolution time line” TENSOR ORDER = “0” UNITS = “Unitless” SIZES = 1 VALUE TYPE = INT SIGNIFICANT DIGITS = 1 FILLVAL = -9 FIELDNAM = “Cluster C1, FGM Instrument Range, on spin resolution time line” LABLAXIS = “Range” DEPEND 0 = time tags C1 CP FGM SPIN END VARIABLE = range C1 CP FGM SPIN START VARIABLE = tm C1 CP FGM SPIN PARAMETER TYPE = “Support Data” CATDESC = “Cluster C1, FGM telemtry mode (burst mode/normal mode) on spin resolution time line” UNITS = “Unitless” TENSOR ORDER = “0” SIZES = 1 VALUE TYPE = INT SIGNIFICANT DIGITS = 2 FILLVAL = -99 FIELDNAM = “Cluster C1, telemetry mode on spin resolution time line” LABLAXIS = “TM” DEPEND 0 = time tags C1 CP FGM SPIN END VARIABLE = tm C1 CP FGM SPIN DATA UNTIL = EOF 2003-01-01T00:00:02.187Z,2,-15.086,7.672,8.769,19.062,31105.1,91115.5,33976.5,2,22 $ 2003-01-01T00:00:06.197Z,2,-15.017,9.020,8.156,19.323,31108.8,91119.2,33972.9,2,22 $ 2003-01-01T00:00:10.208Z,2,-14.396,9.873,5.241,18.226,31112.4,91123.0,33969.3,2,22 $ 2003-01-01T00:00:14.218Z,2,-11.938,14.285,2.235,18.750,31116.1,91126.8,33965.7,2,22 $ 2003-01-01T00:00:18.229Z,2,-11.995,15.514,5.203,20.289,31119.8,91130.6,33962.1,2,22 $ 2003-01-01T00:00:22.240Z,2,-11.672,15.160,5.497,19.907,31123.5,91134.4,33958.5,2,22 $ 2003-01-01T00:00:26.250Z,2,-11.143,13.097,7.096,18.602,31127.2,91138.1,33954.9,2,22 $ 2003-01-01T00:00:30.261Z,2,-12.215,12.659,6.489,18.751,31130.8,91141.9,33951.3,2,22 $

1

Cluster Active Archive: Overview

37

References 1. Allen A., S.J. Schwartz, C. Harvey, C. Perry, C. Huc, P. Robert, Cluster Exchange Format – Data File Syntax, DS-QMW-TN-0010, v2.4, 04 March 2008. URL: http://caa.estec.esa.int/ caa/documentation.xml 2. Escoubet, C. P., M. Fehringer, and M. Goldstein, The Cluster mission, Ann Geophys., 19, 1197, 2001. 3. Harvey C.C., A.J. Allen, F. D´eriot, C. Huc, M. Nonon-Latapie, C.H. Perry, S.J. Schwartz, T. Erikssony and S. McCaffrey, Cluster Metadata Dictionary, CAA-CDPP-TN-0002, v2.3, 03 March 2008. URL: http://caa.estec.esa.int/caa/documentation.xml 4. McCaffrey, S., Command Line Query Interface, Doc: CAA-EST-CMDLINE-0001, v2.0, 06 April 2007. URL: http://caa.estec.esa.int/documents/CAA-EST-CMDLINE-001.pdf 5. Perry, C., Raw Data Network Delivery Interface Document, Doc CAA-EST-ID-0001, v1.2, 28 November 2005. URL: http://caa.estec.esa.int/documents/ CAA-EST-CMDLINE-001.pdf

Chapter 2

ASPOC Data Products in the Cluster Active Archive K. Torkar and H. Jeszenszky

Abstract The instrument ASPOC (Active Spacecraft Potential Control) controls the electric potential of the Cluster spacecraft by means of an ion beam. This modification of the charge balance improves the plasma measurements on board. Beneficial effects have also been observed for the electric field measurements by double probes (EFW). Comprehensive knowledge about the status of the instruments including the ion beam current is necessary to correctly interpret spacecraft potential data. This paper provides an overview of the instrument modes and the parameters contained in the ASPOC data sets in the Cluster Active Archive. A few examples illustrate the relation between instrument modes and data.

2.1 Introduction The ASPOC (Active Spacecraft Potential Control) instruments for the Cluster mission have been described by Torkar et al. [6]. The primary objective of ASPOC is to insure the effective and complete measurement of the ambient plasma distribution functions down to low energy. With the exception of very rare eclipse time intervals the Cluster spacecraft are illuminated by the Sun, and photo-electrons are emitted from the surface. Throughout the Cluster orbit, i.e. outside the plasmasphere, emitted photo-electrons will dominate over collected plasma electrons. A balance between escaping and collected electrons is achieved at a positive potential forcing low energy photoelectrons to orbit back to the spacecraft, allowing only a fraction of the photoelectrons at higher energies to escape and be in current balance with collected plasma electrons. Ions can be neglected because of their low mobility compared to the electrons. The resulting uncontrolled potentials may reach C50 V and more in regions of the magnetosphere with tenuous plasma. In order to control the electrical potential of the spacecraft, the ASPOC instruments emit a beam of positive indium ions (115 amu) at energies of about 6–9 keV K. Torkar () and H. Jeszenszky Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, 8042 Graz, Austria e-mail: [email protected]

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 2, c Springer Science+Business Media B.V. 2010

39

40

K. Torkar and H. Jeszenszky

and currents up to about 30 A. Typical currents applied in the mission were between 10 and 20 A. The emission of positive charges from the spacecraft balances the excess of charge accumulating on the vehicle from interactions with the environment. By adjusting the positive emission current, the spacecraft potential can thus be adjusted to single-digit positive voltages. As a result of the equilibrium between the relevant currents of photo-electrons generated at the spacecraft surface, plasma electrons from the environment, and the ASPOC ion beam current, the spacecraft potential will be clamped to a potential determined by a current balance between ASPOC-generated ions on the one hand and escaping photoelectrons minus collected ambient electrons on the other hand. In a tenuous plasma the latter term can be neglected. The resulting improvement of both particle and electric field measurements is most effective in low plasma density environments and boundaries thereof. In these environments the equilibrium potential of the spacecraft is not only strongly fluctuating together with the variations of plasma density, but also the absolute value of the potential easily reaches tens of volts positive. As a consequence, the bulk of photo-electrons will return to the surface and the plasma electron detectors, where the resulting high count rates let the micro-channel plates age at a faster rate. The energies and trajectories of plasma electrons and ions are modified in the potential well around the spacecraft, thereby complicating the interpretation of the distribution functions and the calculation of plasma moments. The improvement of the electron measurements by a controlled spacecraft potential is illustrated in Fig. 2.1 adapted from Torkar et al. [7]. It shows energy-time spectrograms of field-aligned electrons measured by the PEACE-LEEA sensors (for an instrument description see Szita et al. [5]) of Spacecraft 1 (panel a) and 4 (panel b) over a 2-h time interval starting at 06 UT on May 21, 2002. The electron features in the 10–100 eV decade clearly visible on spacecraft 4 – where ASPOC reduced the potential to 8 V – are not discernible on spacecraft 1, where the uncontrolled potential peaked at 47 V. The spacecraft potential is measured by the EFW instrument (Gustafsson et al., [2, 3]). Its probes are controlled to stay within 1–2 V relative to the ambient plasma and thus provide a reference for the measurement. A reduction of the spacecraft potential also helps to avoid wake artefacts in electric field measurements by the double probes, which may occur in cold, streaming plasma, e.g., polar wind [1]. Under similar conditions, the ion spectrometer CIS [4] could detect cold ion beams (HC and OC ) only while ASPOC was operating. As the photo-emission is fairly stable, the potential of a spacecraft (without control) reflects the variation of the plasma electron current and thereby images the electron density and temperature. As the temperature effect is relatively small, the uncontrolled spacecraft potential is often used as an estimator for plasma density. However, the introduction of an artificial ion current for control purposes changes the relationship dramatically. Under the influence of a constant artificial ion beam current, the controlled spacecraft behaves like a strongly biased plasma probe with rather small residual variations of the potential. As illustrated in Fig. 2.2, a good correlation between the uncontrolled potential (spacecraft 1, black line, scale on the left hand side) and the controlled potential (here represented by spacecraft 3, green

2

ASPOC Data Products in the Cluster Active Archive

a

41

C1 2002141 0600 PA 180

Energy [eV]

1000

100

10

06

b

07

08 UT

C4 2002141 0600 PA 180

Energy [eV]

10–4

100

10–5

10–6

10

06

07

Energy Flux [ergs/(cm 2 s sr eV)]

1000

08 UT

Fig. 2.1 Energy-time spectrograms of field-aligned electrons from the PEACE-LEEA sensors of spacecraft 1 (panel a) and spacecraft 4 (panel b) plotted as function of time on May 21, 2002, from 06 to 08 UT

line and dots) may still exist. The potential of spacecraft 3 is shown in two different ways. The full, green line refers to the common scale with spacecraft 1 on the left hand side and demonstrates the significant reduction of the potential. The same data set is also shown by the green dots for which the right hand scale is valid. The good correlation between the residual variations of the controlled potential on spacecraft 3 with the uncontrolled values on spacecraft 1 is obvious. Thus, by inserting the ASPOC ion current into the current balance equation for the controlled spacecraft one can deduce the uncontrolled potential and use it for density estimations as usual.

42

K. Torkar and H. Jeszenszky

Fig. 2.2 Spacecraft potential (raw potential between the probes and the spacecraft body) measured by EFW (Electric Fields and Waves, [2]) on spacecraft 1 (black, scale at the left) and 3 (both the green line with scale at the left and green dots with scale on the right) on April 30, 2002, from 10:20 to 12:00 UT

2.2 Instrument Overview 2.2.1 Ion Emitters The ion emitters are liquid metal ion sources using indium as charge material. A needle is mounted in a reservoir with indium, which has to be elevated to operational temperature (well above the melting point of indium at 156ıC by an attached small heater. The melting process is initiated by command and takes about 12–20 min. When the specified temperature has been reached, high voltage is turned on and ramped up until the field emission process around the liquid indium at the tip ignites and an ion beam is generated. As soon as the ion beam appears, the high voltage supply is switched into a current controlled mode where the emission current is being set, while the extraction voltage adjusts itself to the conditions of the emitter. The emitters are operated one at a time. For lifetime and redundancy, each instrument holds eight emitters, arranged in two “modules” with four emitters each. All emitters within a module are connected to a single high voltage circuit. The selection of the active emitter is made by heating one of them (the tip curvature of cold emitters with solid indium is not suitable for field emission).

2

ASPOC Data Products in the Cluster Active Archive

43

2.2.2 Currents in the Emitter System The behaviour of the ion beam current is better understood with some knowledge of the currents flowing in the emitter system. At first, there is the current of the emitted ion beam. This current is referred to as ion current, beam current, or emission current. It continues through the ambient plasma and is closed by other currents to the spacecraft surface, particularly the photoemission current. ASPOC ions returning to the spacecraft can be totally neglected even if the ion beam is emitted perpendicular to the magnetic field. Secondly, there is the total current delivered by the high voltage supply to the emitter. This current includes the beam current and internal loss currents (e.g. the current to the extraction and beam focusing electrodes), and may therefore be larger than the beam current. There are two major types of loss currents: currents from ions hitting the extractor electrode, and leakage currents through insulators. The percentage of extractor electrode currents with regard to the total current is small (0–20%) for small to medium currents and may increase from 30% to 50% for currents of about 50 A. The typical emission currents applied on Cluster range between 10 and 16 A. They set an upper limit to the spacecraft potential on the order of 6–8 V. The emission current of an emitter may be increased to maximum current over a short period (between 30 s and a few minutes) as a precaution to remove any contamination from the emitter, thereby ensuring that the operating voltage remains within operational limits. This procedure is called “cleaning”.

2.2.3 Instrument Modes The most widely used active mode of the instrument is the “constant total current mode” (ITOT). It sets a constant output current of the high voltage unit, which includes any losses inside the lens system. Experience has shown that the resulting emission of an almost constant ion current fulfils all requirements for spacecraft potential control in the magnetosphere and the solar wind even without on-board feedback from measurements of the spacecraft potential. In “constant ion current mode” (IION) the processor of the instrument reads the monitor of the outgoing beam current and adjusts the output current of the high voltage supply to compensate for any losses in the system. This mode has been used during commissioning only. In the two available, so-called feedback modes, a measurement of the spacecraft potential is supplied to ASPOC by either the electric field experiment (EFW) or the electron analyser (PEACE) and this information is then used to adjust the beam current in order to maintain a constant value of the potential in a closedloop scheme. This mode is called feedback mode with EFW (mode FEFW) or feedback mode with PEACE (mode FPEA). The measurements of the spacecraft potential are updated once every spin and sent to ASPOC via dedicated serial, digital inter-experiment links. The mode FEFW has been tested successfully during commissioning. Unfortunately, active soundings of the WHISPER instrument cause

44

K. Torkar and H. Jeszenszky

spikes in the potential information transmitted on board. The control software could not handle these spikes, so that this feedback mode could not be used in the nominal mission. The feedback mode with PEACE suffered from the difficulty to provide a reliable potential estimator from the electron spectra onboard, and was not used in the mission at all. In standby mode (STDB) both the emitters and their heaters are turned off. The standby mode is also the safe mode of the instrument, to which it returns autonomously under certain error conditions. The transition into standby mode also clears all error flags and the emitter selection, and disables high voltage and the heaters. In order to reduce the time before emission starts, a “hot standby” mode (HOTS) keeps the indium in a liquid state. This mode can be used to interrupt the ion emission by command, without change of modes or emitters before and after the break. The re-ignition time is reduced to the time required to sweep the high voltage (less than a minute). A “test and commissioning” mode (T&C) varies the total ion current in steps of 8 or 16 s with 2 or 4 A current increments. This mode has been used occasionally to establish the current-voltage characteristics of the spacecraft. A technical mode (TECH) is available for low-level commanding during commissioning and re-commissioning of emitters. Start-up (STUP) is a state of the instrument at the beginning of an active mode when the emitter is being warmed up and ion emission has not yet started. Depending on ambient temperature and emitter condition it takes about 12–20 min to reach a temperature inside the emitters which is sufficient to ignite the ion beam. Within this period the “instrument mode” reported in telemetry is already the commanded target mode, although there is no ion emission yet.

2.3 ASPOC Datasets in the Cluster Active Archive The ASPOC Datasets in the Cluster Active Archive are listed in Table 2.1. All data sets are Cluster Exchange Format (CEF) files, except the textual caveat description. The caveat information consists of an automatically generated part (CEF) and a manually generated part (text) added after inspection of the data produced in the batch process. The automatically generated caveats shall help the user to assess the quality. The production software derives a quality parameter from the total current consumed by the ion emitter and the emitted ion current. In the typical mode of operation a constant current is applied to the ion emitters. If internal loss currents can be neglected, the emitted ion current is also constant and the resulting spacecraft potential has the best stability. In the presence of loss currents, the emitted ion current and with it the spacecraft potential will start to fluctuate. Table 2.2 summarises the definition of the quality levels. There are new entries whenever the instrument mode changes, or when the commanded emission current changes.

2

ASPOC Data Products in the Cluster Active Archive

45

Table 2.1 ASPOC data sets Product Description IONC ASPOC ion beam current at a sampling interval of 0.5 s IONS Snapshots (samples of 8.25 s duration collected every 128.8 s) of the ion beam current with a sampling interval of 0:033 s. Suitable to check current fluctuations in the beam and to resolve the current profile during the ignition or shut-down of an ion emitter STAT Comprehensive status information of the instrument at 5.15 s resolution (10.3 s for some parameters) Operating mode, including emitter start-up status and cleaning: In both of these states the beam current deviates significantly from the nominal current Emitter identification and emitter module identification: Each instrument carries eight emitters bundled into two modules with a common high voltage system within a module. Both emitters and modules may exhibit individual characteristics and performance, which may have an effect on the value and temporal behaviour of the beam current Anomaly flags: These flags give the reason for an anomaly, for example failure of ignition of an emitter Ion beam energy: The voltage applied to the emitter is equivalent to the energy of the ions in the beam Total ion source current: The current delivered by the high voltage supply into the emitter will be higher than the emitted current if internal losses are present Heater current: The current drawn by the heater element associated with an emitter Heater voltage: The voltage applied to the heater element associated with an emitter. Voltage and current together can be used to calculate not only the electrical power but also the temperature of the heater element. From this temperature, the temperature of the emitter tip can be derived Temperature of the ion source module: This parameter contains the ambient temperature of the emitter module. The temperature of the emitter tip shows a strong dependence on the heater temperature, but also some influence of this ion source module temperature Raw spacecraft potential received on board from EFW: This parameter mirrors the spacecraft potential data delivered by the instrument EFW to ASPOC on board the spacecraft. It differs from the EFW data product in the archive with respect to timing CMDH

CAVEATS

Command history with time stamps according to on-board execution time by the instrument. The execution delay may reach 1 min. Included parameters are the command code, the command counter, command mnemonic, description, and parameter (if any) Information on the status and quality of the ion emission (see the description of quality levels). Time spans correlate with commands to change the operational mode or the value of the current

ACTIVE

Mean quality level during time span Minimum quality level during time span Maximum quality level during time span Comment; a textual description associated with the average quality level value

Contains the time spans when ASPOC is switched on and is emitting a significant ion current (>1 A). Useful for quick checks whether ASPOC was operational

46

K. Torkar and H. Jeszenszky Table 2.2 Quality level definition Quality flag Condition 1 Ibeam =Itotal > 0:97 2 0:92 < Ibeam =Itotal 0:97 3 0:75 < Ibeam =Itotal 0:92 4 0:30 < Ibeam =Itotal 0:75 5 0:00 < Ibeam =Itotal 0:30 8 Itotal 20A and Ibeam =Itotal 0:50 9 Ibeam D 0:0

Ion emission quality definition Excellent Good, almost completely stable Moderate fluctuations Substantial fluctuations Severe variations Cleaning No emission

2.4 Examples and Applications No specific software is required to use the data sets. However, visualisation software in IDL language originally developed for the validation of the data products is available. It allows to browse through a set of CEF files, either uncompressed or in gzip. Check boxes allow to select certain parameters. Plot options and scales can be set automatically or interactively. The plots presented in this section have been generated by this tool. Figure 2.3 shows beam current and voltage in the top and bottom panel, respectively, over a 12-h time interval for spacecraft 3. The colour bar at the bottom indicates the quality flag, which is one (excellent) for this operational interval of about 7 h. The commands to enter into active mode have been sent shortly before 04 UT (red vertical lines). It takes about 15 min (interval indicated by the yellow status bar) until the emitter is hot and ion emission starts. The time interval contained in the respective entry in the ACTIVE data set would only cover the time interval of the green status bar. Figure 2.4 shows (from top to bottom) for the same time interval the beam current, total current, and spacecraft potential (as transmitted on board from EFW to ASPOC). Noteworthy is the “bite-out” of the beam current at 18:30 UT, due to a suddenly appearing and later disappearing leakage current inside the emitter. Although the beam current decreases by about 30%, the spacecraft potential hardly shows any reaction. This shows that variations of the beam current should be taken into account for accurate studies related to spacecraft potential, but do not generally deteriorate the control of the spacecraft potential. Finally, Fig. 2.5 illustrates an application of the beam current snapshot parameter (IONS) at 33 ms resolution. The plotted interval covers 15 s around a beam turn-on. The top panel contains the standard beam current parameter (IION) which cannot fully resolve the variation of the beam current, whereas the snapshot reflects the variations very accurately, which is confirmed by the analogous variation of the spacecraft potential (bottom panel) taken from the EFW data set C3 CP EFW L2 P.

2

ASPOC Data Products in the Cluster Active Archive

47

Satellite #3 - 20040728

30

I_Beam [uA]

NOE

40

CLN

50

0 10

4 2

UT

0 00 2004-07-28

Energy [kV]

Q=2 Q=1

8

OFF

Q=3

Q=4

10

6

#051: ZEASTDBS (0x1000)

#046: ZEAITOTS (0x1100)

Q=5

20

02

04

06

08

10

12 2004-07-28

Fig. 2.3 ASPOC data set visualisation with beam current, beam energy, data quality, and commands

Fig. 2.4 ASPOC data set visualisation with beam current, total current, and spacecraft potential

48

K. Torkar and H. Jeszenszky

Fig. 2.5 Example of ASPOC snapshot data set with corresponding variation of the spacecraft potential

2.5 Status of the Data Sets As the ASPOC instruments were finally shut down at the end of June, 2008, the data set in the Cluster Active Archive can be considered complete. Unless there are future outside requests for additions or modifications, the data set is also the final one. Files have been generated whenever the instruments were powered on, even if no ion emission took place. This is the case for many data sets in the extended mission due to exhaustion of the ion emitter reservoirs. Acknowledgements The authors would like to thank many teams for their valuable efforts: the CAA team for building the archive, the ASPOC team for building the instruments, the EFW and PEACE teams for good co-operation during the project, and ESOC and JSOC for operational support.

References 1. Engwall, E., A. I. Eriksson, M. Andr´e, I. Dandouras, G. Paschmann, J. Quinn, K. Torkar, Low-energy (order 10 eV) ion flow in the magnetotail lobes inferred from spacecraft wake observations, Geophys. Res. Lett 33, L06110, doi:10.1029/2005GL025179, 2006.

2

ASPOC Data Products in the Cluster Active Archive

49

2. Gustafsson, G, R. Bostrom, G. Holmgren, A. Lundgren, K. Stasiewicz, L. Ahlen, F.S. Mozer, D. Pankow, P. Harvey, P. Berg, R. Ulrich, A. Pedersen, R. Schmidt, A. Butler, A. Fransen, D. Klinge, C.-G. Falthammar, P.-A. Linqvist, S. Christenson, J. Holtet, B. Lybekk, T.A. Sten, P. Tanskanen, K. Lappalainen, J. Wygant, The electric field experiment for the Cluster mission, Space Sci. Rev. 79, 137–156, 1997. 3. Gustafsson, G., M. Andr´e, T. Carozzi, A.I. Eriksson, C.-G. F¨alhammar, R. Grard, G. Holmgren, J.A. Holtet, N. Ivchenko, T. Karlsson, Y. Khotyaintsev, S. Klimov, H. Laakso, P.-A. Lindqvist, B. Lybekk, G. Marklund, F. Mozer, K. Mursula, A. Pedersen,B. Popielawska, S. Savin, K. Stasiewicz, P. Tanskanen, A. Vaivads, J-E.Wahlund, First results of electric field and density observations by Cluster EFW based on initial months of operation, Ann. Geophys. 19, 1219– 1240, 2001. 4. R`eme, H., Aoustin, C., Bosqued, J. M., Dandouras, I, et al., First multispacecraft ion measurements in and near the Earth’s magnetosphere with the identical Cluster ion spectrometry (CIS) experiment, Ann. Geophys. 19, 1303–1354, 2001. 5. Szita, S., A. N. Fazakerley, P. J. Carter, et al., Cluster PEACE observations of electrons of spacecraft origin, Ann. Geophys 19, 1721–1730, 2001. 6. Torkar K., W. Riedler, C.P. Escoubet M. Fehringer, R. Schmidt, R.J.L. Grard, H. Arends, F. R¨udenauer, W. Steiger, B.T. Narheim, K. Svenes, R. Torbert, M. Andr´e, A. Fazakerley, R. Goldstein, R.C. Olsen, A. Pedersen, E. Whipple, H. Zhao, Active spacecraft potential control for Cluster – implementation and first results, Ann. Geophys 19, 1289–1302, 2001. 7. Torkar K., K.R. Svenes, A. Fazakerley, S. Szita, H. R`eme, I. Dandouras, M. Fehringer, C.P. Escoubet, M. Andr´e, Improvement of plasma measurements onboard Cluster due to spacecraft potential control, Adv. Space Res 36, doi:10.1016/j.asr.2005.01.109, 2005.

Chapter 3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA) I. Dandouras, A. Barthe, E. Penou, S. Brunato, H. R`eme, L.M. Kistler, M.B. Bavassano-Cattaneo, A. Blagau, and the CIS Team

Abstract The Cluster Active Archive (CAA) aims at preserving the four Cluster spacecraft data, so that they are usable in the long-term by the scientific community as well as by the instrument team PIs and Co-Is. This implies that the data are filed together with the descriptive and documentary elements making it possible to select and interpret them. The CIS (Cluster Ion Spectrometry) experiment is a comprehensive ionic plasma spectrometry package onboard the four Cluster spacecraft, capable of obtaining full three-dimensional ion distributions (about 0–40 keV/e) with a time resolution of one spacecraft spin (4 s) and with mass-per-charge composition determination. The CIS package consists of two different instruments, a Hot Ion Analyser (HIA) and a time-of-flight ion Composition Distribution Function (CODIF) analyser, plus a sophisticated dual-processor based instrument control and

I. Dandouras (), A. Barthe, E. Penou, S. Brunato, and H. R`eme Universit´e de Toulouse, Centre d’Etude Spatiale des Rayonnements, B.P. 44346, 31028 Toulouse, France and CNRS, UMR 5187, B.P. 44346, 31028 Toulouse, France e-mail: [email protected]; [email protected]; [email protected] A. Barthe AKKA Technologies, 6 Rue R. Camboulives, 31036 Toulouse, France e-mail: [email protected] S. Brunato Noveltis, 2 Avenue de l’Europe, 31520 Ramonville-Saint-Agne, France e-mail: [email protected] L.M. Kistler Space Science Center, University of New Hampshire, Durham, NH 03824, USA e-mail: [email protected] M.B. Bavassano-Cattaneo Istituto di Fisica dello Spazio Interplanetario, INAF, Via del Fosso del Cavaliere 100, 00133 Roma, Italy e-mail: [email protected] A. Blagau Institute of Space Sciences, P.O. Box MG-23, RO 77125 Bucharest-Magurele, Romania e-mail: [email protected]

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 3, c Springer Science+Business Media B.V. 2010

51

52

I. Dandouras et al.

data processing system (DPS). For the archival of the CIS data a multi-level approach has been adopted. The CAA archival includes processed raw data (Level 1 data), moments of the ion distribution functions (Level 2 data), and calibrated highresolution data in a variety of physical units (Level 3 data). The latter are 3-D ion distribution functions. In addition, a software package has been developed to allow the CAA user to interactively calculate partial or total moments of the ion distributions. The CIS data archive includes also experiment documentation, graphical products for browsing through the data, and data caveats. Given the complexity of an ion spectrometer, and the variety of its operational modes, each one being optimised for a different magnetospheric region or measurement objective, consultation of the data caveats by the end user will always be a necessary step in the data analysis.

3.1 Introduction The Cluster Active Archive (CAA) aims at preserving the four Cluster spacecraft data, so that they are usable in the long-term by the world-wide scientific community as well as by the instrument team PIs and Co-Is. Its purpose is to maximise the scientific return from the mission, and to ensure that the unique data set returned by the Cluster mission is preserved in a stable, long-term archive for scientific analysis beyond the end of the mission. This implies that the instrument data, properly calibrated, are filed together with the descriptive and documentary elements making it possible to select and interpret them [8]. The CAA home page is at http://caa.estec.esa.int/.

3.2 The CIS Experiment The CIS (Cluster Ion Spectrometry) experiment is a comprehensive ionic plasma spectrometry package onboard the four Cluster spacecraft, capable of obtaining full three-dimensional ion distributions (about 0–40 keV/e) with a time resolution of one spacecraft spin (4 s) and with mass-per-charge composition determination [10]. The prime scientific objective of the CIS experiment is the study of the dynamics of magnetised plasma structures in and in the vicinity of the Earth’s magnetosphere, with the determination, as accurately as possible, of the local orientation and the state of motion of the plasma structures required for macrophysics and microphysics studies. The four Cluster spacecraft encounter ionic plasma of vastly diverse characteristics in the course of 1 year (Fig. 3.1). In order to study all the plasma regions with the fluxes shown in Fig. 3.1, the CIS experiment needs therefore to be a highly versatile and reliable ionic plasma experiment, capable of measuring both the cold and hot ions of Maxwellian and non-Maxwellian populations (for example, beams) from the solar wind, the magnetosheath, and the magnetosphere (including the upper ionosphere), with sufficient angular, energy and mass resolutions to accomplish the scientific objectives.

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

53

IONS 1010 SW

CIS 2 / HIA LOW SIDE

DIFFRENTIAL DIRECTIONAL ENERGY FLUX (cm2.s.sr)–1

109

MP

108

CIS 1/ CODIF LOW SIDE

MSH

UPW 0+ 107

CIS 1/ CODIF HIGH SIDE PM

106

PS 105

LOBE

CIS 2/ HIA HIGH SIDE

MSPH

104

CIS 1/RPA

LOBE 103

1

10

100 1000 ENERGY (eV)

10000

Fig. 3.1 Representative ion fluxes encountered along the Cluster orbit in the solar wind (SW), the magnetopause (MP), the magnetosheath (MSH), the plasma mantle (PM), the plasma sheet (PS), the lobe and upwelling ions (UPW). The range of the different sensitivities of CIS1/CODIF (low side, high side and RPA) and CIS2/HIA (low side and high side) are shown with different colours

To satisfy all these criteria, which cannot be met with a single instrument, the CIS package consists of two different instruments: A time-of-flight ion Composition and Distribution Function analyser (CODIF,

or CIS-1) A Hot Ion Analyser (HIA, or CIS-2)

In addition, each of the instruments, in order to be able to cover a dynamic range of about 6 orders of magnitude in particle fluxes, provides two different geometric factors: a high-sensitivity side (or HS) and a low-sensitivity side (or LS).

54

I. Dandouras et al.

The CIS experiment includes also a sophisticated dual-processor based instrument control and data processing system (DPS). The CIS home page is at http://cluster.cesr.fr:8000/.

3.2.1 The CODIF (CIS-1) Instrument The CODIF instrument is a high-sensitivity mass-resolving spectrometer with an instantaneous 360ı 8ı field of view to measure full 3-D distribution functions of the major ion species (in as much as they contribute significantly to the total mass density of the plasma), within one spin period of the spacecraft. Typically these include HC , HeC , HeCC and OC , with energies from 25 eV=e to 40 keV/e and with medium .22:5ı / angular resolution. The CODIF instrument combines ion energy per charge selection, by deflection in a rotationally symmetric toroidal electrostatic analyser, with a subsequent time-of-flight analysis after post-acceleration to 15 keV=e. Ions are selected as a function of their E=q (energy per charge) ratio, by sweeping the high voltage applied between the two toroidal hemispheres. The full energy sweep with 31 contiguous energy channels is performed 32 times per spin. In the time-of-flight (TOF) section the velocity of the incoming ions is measured, which allows then the calculation of their m=q (mass per charge) ratio. Microchannel plate (MCP) electron multipliers are used to detect both the ions and the secondary electrons, which are emitted from a carbon foil at the entry of the TOF section, during the passage of the ions. These secondary electrons give the “start” signal, for the time-of-flight measurement, and the position information (elevation angle of the incoming ion, provided by the MCP sectoring in anodes). In order to cover populations ranging from magnetosheath protons to tail lobe ions, a dynamic range of more than 105 is required. CODIF therefore consists of two sections, each with a 180ı field of view, with geometry factors differing by a factor of 100. This way, one of the sections will have counting rates which are statistically meaningful and which at the same time can be handled by the time-offlight electronics. However, intense ion fluxes can in some cases saturate the CODIF instrument (particularly if data are acquired from the high sensitivity side), but these fluxes are measured with HIA. The operation of the high-sensitivity side (“high-G”, or “HS”) and of the low-sensitivity side (“low-g”, or “LS”) on CODIF is mutually exclusive, and only one of the two sides can be selected at a time to supply data. The high-sensitivity side is divided into eight sectors (or anodes), 22:5ı each, and the low-sensitivity side into six 22:5ı sectors, i.e. there are two “blind” anodes. With an additional Retarding Potential Analyser (RPA) device in the aperture system of the CODIF sensor, and with pre-acceleration for the energies below 25 eV/e, the range is extended to energies as low as almost the spacecraft potential. The retarding potential analyser operates only in the RPA mode, and provides an energy range between about 0.7 and 25 eV/e (with respect to the spacecraft potential). The operation on CODIF of the RPA mode and of the normal magnetospheric modes (which provide a 25 eV/e to 40 keV/e energy range) is mutually exclusive.

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

55

During the RPA operation, however, ion measurements above 25 eV/e are provided by the HIA instrument. For the RPA aperture system all anodes have the same geometric factor (on both sides), although the instrument electronics allow the operation of only one side at a time.

3.2.2 The HIA (CIS-2) Instrument The HIA instrument is an ion energy spectrometer, capable of obtaining full three-dimensional ion distributions with good angular and time resolution (one spacecraft spin). HIA combines the selection of incoming ions, according to the ion energy per charge by electrostatic deflection in a quadrispherical analyser, with a fast imaging particle detection system. This particle imaging is based on microchannel plate (MCP) electron multipliers and position encoding discrete anodes. As for CODIF, ions are selected as a function of their E=q (energy per charge) ratio, by sweeping the high voltage applied between the two hemispheres. In order to cover populations ranging from solar wind and magnetosheath ions to tail lobe ions, a dynamic range of more than 105 is required. HIA therefore consists of two 180ı field-of-view sections, with two different sensitivities (with a ratio 20), corresponding respectively to the high-sensitivity (“high-G”, or “HS”) and to the low-sensitivity (“low-g”, or “LS”) side. The “low g” side allows detection of the solar wind and the required high angular resolution is achieved through the use of 8 sectors (or MCP anodes), 5:625ı each, the remaining 8 sectors having 11:25ı resolution. The 180ı “high G” side is divided into 16 sectors, 11:25ı each. For each sensitivity side a full 4 steradian scan, consisting of 32 energy sweeps, is completed every spin of the spacecraft, i.e. every 4 s, giving a full three-dimensional distribution of ions in the energy range 5 eV=e–32 keV=e. Each sweep consists of 62 contiguous energy channels.

3.2.3 The CIS Operational Modes The CIS instruments have a large amount of flexibility, either in the selection of the operational mode or in the reduction of the data necessary to fit the available telemetry bandwidth. CIS can thus operate in any combination of the six Spacecraft Telemetry Modes (three Normal Telemetry Modes, NM1-NM3, and three Burst Telemetry Modes, BM1-BM3) and the 16 CIS Operational Modes. These 16 modes (Table 3.1) can be grouped into solar wind tracking modes, solar wind study modes with the priority on the upstreaming ions, magnetospheric modes, magnetosheath modes, an RPA mode and a calibration and test mode. These modes correspond to different energy sweeping schemes and different combinations of telemetry products transmitted. Onboard calculated moments are always transmitted to the telemetry with a high-time resolution (1 spin). A combination of 2-D and 3-D ion distribution functions, plus other telemetry products, are transmitted in parallel to onboard calculated

56

I. Dandouras et al. Table 3.1 CIS operational modes CIS mode Mode name 0 SW-1 Solar wind/SW tracking – mode 1 1 SW-2 Solar wind/3D upstreaming ions – mode 2 2 SW-3 Solar wind/SW tracking – mode 3 3 SW-4 Solar wind/3D upstreaming ions – mode 4 4 SW-C1 Solar wind/SW tracking – data compression – mode 1 5 SW-C2 Solar wind/3D upstreaming ions – data compression-mode 2 6 RPA RPA mode 7 PROM PROM operation 8 MAG-1 Magnetosphere – mode 1 9 MAG-2 Magnetosphere – mode 2 10 MAG-3 Magnetosphere – mode 3 11 MAG-4 Magnetosphere/Magnetosheath – mode 1 12 MAG-5 Magnetosheath – mode 2 13 MAG-C1 Magnetosphere – data compression – mode 1 14 MAG-C2 Magnetosheath – data compression – mode 2 15 CAL Calibration/test mode

moments, with a mode-dependent and product-dependent time resolution. The table in Fig. 3.2 gives a typical example of CODIF telemetry products transmitted during the NM1 and the BM1 spacecraft telemetry modes, for the 16 CIS modes, and the corresponding time resolution for each product, in number of spins. The actual tables are supplied in the CIS calibration files (cf. Section 3.3.5). Mode change is performed by time-tagged commands, according to the plasma populations anticipated along the Cluster orbit. Although during the first 2 years of operations various CIS mode selection rules have been in use, since 2003 the most commonly used modes are mode 8 (MAG-1) and 13 (MAG-C1) in the magnetosphere, mode 12 (MAG-5) and 14 (MAG-C2) in the magnetosheath, and mode 3 (SW-4) and 5 (SW-C2) in the solar wind. Mode 10 (MAG-3) is also used in the magnetosphere for providing full resolution HIA data at the expense of CODIF data, as e.g. onboard sc1 since the end of 2004 (cf. Section 3.2.4). The RPA mode is operated about once per month, either on all spacecraft or on selected spacecraft. Magnetospheric modes stay relatively simple, i.e. the full energy-angle ranges are systematically covered. For HIA the different telemetry products (including moments) are deduced from the 62E 128 energy solid angle count rate matrices accumulated on the “high G” section. For CODIF they are deduced from the sensitivity side selected by command. Solar wind tracking modes allow a precise and fast measurement (4 s) of the ion flow parameters (HC , HeCC ). For that, in the solar wind, for HIA the energy sweep range is automatically reduced when the field-of-view of the “low g” section is facing the 45ı sector centred in the solar wind direction. This energy sweep range is adapted every spin (Modes 0, 2, 4), centred on the main solar wind velocity by using a criterion based on the HC thermal and bulk velocities computed during the previous spin (solar wind beam tracking). This allows a higher energy resolution for the solar wind beam data.

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

Fig. 3.2 CODIF Telemetry Products transmitted in various modes and corresponding time resolution, in number of spins

3 57

58

I. Dandouras et al.

Fig. 3.3 Schematic of the CIS-2 (HIA) energy-sweep scheme, as a function of the instrument field-of-view orientation in the spin plane: solar wind modes (modes 0–5). Note that, at any given spin phase, the high-sensitivity side field-of-view is looking at the diametrically opposite direction with respect to the low-sensitivity side field-of-view, indicated in the schematic

During solar wind modes with the priority on the upstreaming ions (Modes 1, 3, 5) solar wind beam tracking is performed by HIA only once every 16 spins (exact value can vary depending on mode number). During the remaining 15/16 spins a broader energy sweep is used for the solar wind detection by the “low g” section, allowing at the same time the detection of upstreaming ions by the “high G” section, which is then looking in the anti-sunward direction. Detailed 3-D distributions for the solar wind (“low g” section) are then transmitted to the telemetry only during solar wind beam tracking (once every 16 spins), but onboard calculated solar-wind moments are transmitted every spin. Moreover, detailed 3-D distributions from the “high G” section (e.g. for upstreaming ions and/or for interplanetary disturbances) are included in the basic products transmitted to the telemetry. Outside this 45ı sector, the full energy sweep range is used (solar wind modes and solar wind modes with the priority on the upstreaming ions). However, when the field-of-view of the “high G” section is facing the 45ı sector centred in the solar wind direction (Fig. 3.3), the energy sweep stops (and “freezes”) above the solar wind alpha particles energy, to avoid a quick degradation of the MCPs by the intense solar wind beam (modes 0–5). For CODIF, the reduced energy-sweep principle is also used during modes 0–5, when the field-of-view of the “high G” section is facing the 45ı sector centred in the solar wind direction. Complete energy sweeping is used in the remaining part of the spin (Fig. 3.4). Magnetosheath modes are like magnetospheric modes. However, starting from 1 November 2003, the energy sweep scheme has been redefined for CODIF. With the new CODIF Magnetosheath Modes (modes 12 and 14), the energy sweeps are

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA) log E

3

0

11

15

59

Spin# n

31

10 Sweep

1

Solar Wind Alphas Protons

0.1 HS 0.01

180 0

11

0 15

31

10 Solar Wind

Sweep

1

Alphas Protons

0.1 LS 0.01 Azimuth

0

180

Fig. 3.4 Schematic of the CIS-1 (CODIF) energy-sweep scheme, for solar wind modes, as a function of the spin phase: high-sensitivity side field-of-view orientation (upper panel), and lowsensitivity side field-of-view orientation (bottom panel). The energy “freeze”, during the reduced energy sweeps, is at about 2 keV

truncated and “frozen” at about 2 keV for 16 out of 32 sweeps (when the highsensitivity side faces the magnetosheath mainstream flow), to save the MCP lifetime.

3.2.4 CIS Instrument Status The overall experiment performance, after about 8 years of operation in space, is good. The particle detection efficiency degradation, due to the MCP gain fatigue mechanism [9], is moderate for the HIA instrument. Actual efficiencies (first half 2008) are reduced to about 77% of the initial HIA detection efficiencies on Cluster sc1 (spacecraft 1), and are almost the same as the initial ones for HIA on sc3. However, the physical degradation of the MCPs is in reality higher, but it is in some extent compensated by applying increased high voltages on the MCP plates, to restore secondary electron gain in the MCP channels despite the physical degradation of their emissive surfaces. For CODIF the efficiency degradation is more pronounced, due to the operational principle of this instrument, requiring the detection of both the ion (“stop” timeof-flight signal) and of the electrons emitted by the carbon foil (“start” time-offlight signal), plus a “position” signal, to validate the detection of an ion. Actual efficiencies for CODIF are now reduced, on the average, to less than about 10% of the initial detection efficiencies on Cluster sc4, and to less than about 5% on sc3

60

I. Dandouras et al. Table 3.2 CIS instrument status SC 1 SC 2 CODIF Operations until Not operational October 2004 HIA

Normal operations

SC 3 One deficient MCP quadrant (cf. Section 3.3.8) Normal operations

SC 4 Normal operations

Switched OFF

(cf. Section 3.3.5). In spite of these values, the CODIF performance and ability to still collect data of quality adequate for addressing most of its scientific objectives is remarkable. When processing the raw instrument data to convert them into higher-level data in physical units, these efficiency degradations are compensated by the calibration values used, supplied in the calibration files (cf. Section 3.3.5). However, these calibrations cannot correct for the degraded signal-to-noise ratio induced by the reduced particle counting statistics, particularly when measurements are performed in lowdensity plasmas. The CODIF instrument onboard Cluster sc1 is switched off since 25 October 2004, due to an MCP high voltage anomaly. Onboard Cluster sc3, the CODIF instrument presents one deficient MCP quadrant (southward looking directions on the high-sensitivity side, cf. Section 3.3.8). The HIA instrument is switched off on Cluster sc4, but could be switched on if necessary. On this spacecraft, due to an electrostatic analyser high voltage issue, the HIA energy sweep range is limited between 5 and 400 eV=e. The CIS experiment is not operational on Cluster sc2. Table 3.2 summarises the CIS instrument status, at the time of this writing (2008).

3.3 CIS-CAA Archival Products For the archival of the CIS data a multi-level approach has been adopted [2, 4]. The CAA archival includes processed raw data (Level 1 data), moments of the distribution functions (Level 2 data), and calibrated high-resolution data in a variety of physical units (Level 3 data). Furthermore, the calibration files and high-level processing software are also archived. This approach provides “ready to use” high-resolution data products, archived in ASCII (CEF-2 format), which are expected to have a long lifetime. However, should a user wish to re-analyse the raw data and the instrument calibrations, he will also be able to do it. The CIS data archive includes also experiment documentation, graphical products for browsing through the data, and data caveats. In order to help the user, a CIS Data Selection Guide has been also provided as appendix in the CIS-CAA ICD and in the User Guide to the CIS Measurements in the CAA.

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

61

3.3.1 CIS Level 1 Data CIS Level 1 data are decommutated, decompressed, and time-tagged telemetry data. They are in raw instrument units (no corrections for MCP efficiencies etc.), represent the complete CIS Telemetry, and are the input for all higher-level processing. CIS Level 1 data, together with the CIS calibration files, are in particular the input files for the “cl” software (cf. Section 3.3.6). These data are organised in one file per telemetry product – spacecraft – day, and each file is a time-series of equal-length records. Each data record is complete: time tag, product type, mode info, etc. Data are in IEEE integers or floats.

3.3.2 CIS Level 2 Data CIS Level 2 data are moments of the particle distribution functions: ion density, velocity (in the GSE reference frame), temperature (parallel and perpendicular components, obtained by diagonalising the pressure tensor). For CODIF, which has a higher upper energy limit than HIA, the pressure is also supplied in addition. They correspond to the Prime Parameters from the Cluster Science Data System [1]: onboard calculated moments, then reprocessed on ground (total efficiency calibration adjustments, coordinate transformations etc.). Onboard calculated moments provide 1-spin time resolution, and are calculated from the full angular and energy resolution 3-D ion distributions. In addition to the onboard calculated moments, CIS Level 2 data include also CODIF moments calculated on the ground from the 3-D ion distributions. These provide better calibration adjustments (per anode efficiency) and are thus more accurate, but have a reduced time and energy resolution. HIA on the ground calculated moments are not supplied, because the same anode calibrations would be used as for the onboard calculated ones, but with a degraded time, energy and angular resolution (no added value). A software package has also been developed and is available for download at the CAA web site, allowing the user to interactively calculate partial (or total) moments of the ion distributions, for selected energy and solid angle ranges: CODIF (all 4 ion species) and HIA data (cf. Section 3.3.6). Onboard calculated moments and on the ground calculated moments from the 3-D ion distributions are thus complementary. A data selection guide, to help the CAA user select the most appropriate dataset, is given in Appendix 1 of the CAACIS Interface Control Document [2] and in the User Guide to the CIS Measurements in the CAA. The way moments are calculated from the 3-D ion distribution functions is described in Appendix 2 of the CAA-CIS Interface Control Document.

62

I. Dandouras et al.

3.3.3 CIS Level 3 Data CIS Level 3 data are processed high-resolution data: 3-D ion distributions. They are produced by correcting the Level 1 data for detector efficiencies, geometric factors and other information available from the calibration tables, and give measurements in several physical units (in separate files):

(Differential) Particle flux (Differential) Particle energy flux Particle phase space density Corrected-for-efficiency particle count rate Raw particle counts

ions cm2 s1 sr1 keV1 keV cm2 s1 sr1 keV1 ions s3 km6 ions s1 Number of ions per counter bin

These Level 3 archival files are constructed by joining files from similar telemetry data products (same ion species, different angular, energy or time resolution). Figure 3.5 shows how these files are organised as a function of:

The instrument (HIA or CODIF) The operational mode (magnetospheric, solar wind, or RPA) The instrument sensitivity side (HS or LS) The ion species (for CODIF only)

The 3-D distributions are in the ISR2 pseudo-GSE reference frame (X sunward, Z is the spacecraft axis and northward pointing). They give the complete 16 azimuth 8 elevation D 128 solid angles matrix, and remove the reduction to a total of 88 solid angles in the telemetry products, where adjacent solid angles near the polar directions have been binned together (for some CIS telemetry products). This is achieved by sub-diving each of these binned-together solid angles into the original 2 or 4 solid angles, and distributing the corresponding particle counts in an equal way. Solid angles, defined in the distributions, correspond to particle arrival directions (direction of travel of the particle). HS CODIF

LS

3 types x 4 ion species x 5 physical units = 60 datasets

RPA CIS HS_MAG HIA

HS_SW

3 types x 5 physical units = 15 datasets

LS_SW

Fig. 3.5 CIS level 3 data archival scheme for the 3-D ion distributions

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

63

In addition to the 3-D distributions, pitch-angle distributions (PADs) are also provided. These are calculated on the ground from the 3-D distributions and the magnetic field direction (spacecraft reference frame), are given in particle flux units as a function of the particle energy and pitch-angle (2-D distributions), and their resolution is in 16 equal angular sectors covering the pitch angle range from 0ı to 180ı . Besides these 2-D and 3-D distributions, some uncalibrated CODIF data sets are also included in the CIS Level 3 data: files with onboard selected ion detection events, MCP monitor rates, and low-angular resolution distributions in 64 m=q ranges. These data sets are useful for monitoring instrument performance against eventual background and ion mass species separation, or for searching for minority ion species. Level 3 archival files are labelled with metadata, i.e. information that describes the dataset and its contents, so as to be easily understandable by the CAA user. These metadata follow the specifications given in the Cluster Metadata Dictionary [6], which has been produced for the CAA project by the CAA Metadata Working Group and the CDPP, the French Plasma Physics Data Centre (http://cdpp.cesr.fr/).

3.3.4 CIS Graphical Data Products The CIS graphical data archived are 6-h energy-time ion spectrograms. They are preformatted displays: PNG graphic files embedded in HTML pages, and they are given at two levels of resolution: browsing and detailed. These spectrograms are supplied separately for each spacecraft. Figure 3.6 shows an example of a CIS energy-time spectrogram. It includes, from top to bottom, HIA data from the low-sensitivity side (ions in the 45ı 45ı sector centred in the solar wind direction), HIA data from the high-sensitivity side (separately for four azimuthal sectors: ions arriving in the 90ı 180ı sector with a field-of-view pointing in the sun, dusk, tail, and dawn direction respectively), omnidirectional CODIF data for HC , HeC and OC ions respectively, ion velocity components measured by HIA, and ion densities measured by HIA and by CODIF. Instrument operational mode and spacecraft telemetry mode data are also included, as well as spacecraft coordinates. As can be seen in the HIA mode panel, at the top, the instrument was initially in a solar wind – upstreaming ions mode (mode 3). The solar wind narrow-energy beam is clearly identified in the low-sensitivity side HIA spectrogram, while at the same time the high-sensitivity side HIA spectrogram for ions arriving from the sunward looking direction does not detect the solar wind beam. This is due to the energy-sweep scheme adopted during these modes, that stops (and “freezes”) above the solar wind alpha particles energy, as explained in Section 3.2.3. Some high-energy upstreaming ions are however visible at that time, in the high-sensitivity side HIA spectrograms, as well as in the CODIF HC spectrogram, which was then operating at the high-sensitivity side. Note that CODIF also does not detect the solar wind beam, for the same reason: the energy-sweep scheme adopted during these modes (cf. Fig. 3.4), but solar wind beam data are supplied by

I. Dandouras et al.

5.9 4.3 2.7 Log c/s 6.8

HIA HIA lons

6.0 5.3

HIA lons

4.6

A

3.9

HIA lons

3.2

A

E (eV) SW–E E (eV) E (eV) E (eV) E (eV) E (eV) v (km/s) E (eV)

n (cm–3) n (cm–3)

BM3 NM1

A

Fig. 3.6 CIS energy-time ion spectrogram. See text for details

13/Feb/2001

2.5

HIA lons BM3

1.8

NM1 15kv 12kv

PACC

SAMBA (SC 3)

A

CIS

16 HIA Mode 8 0 10000 1000 100 10000 1000 100 10 10000 1000 100 10 10000 1000 100 10 10000 1000 100 10 16 CODIF Mode 8 0 High CODIF G Low 10000 1000 100 10 10000 1000 100 10 10000 1000 100 10 400 200 0 –200 –400 –600 100.00 10.00 1.00 0.10 0.01 100.0000 10.0000 1.0000 0.1000 0.0100 0.0010 0.0001 12:00 13:00 14:00 15:00 16:00 14.09 12.38 11.39 13.28 14.83 XGSE 1.42 0.62 0.21 1.02 1.81 YGSE –5.72 –6.78 –5.30 –6.12 –6.47 ZGSE 15.28 13.25 15.85 14.66 13.98 DIST Produced by CESR, Printing date: 16/Oct/2007

A

64

Ä

Log c/s 6.7 5.8 4.9 Ä 4.0 CODIF He+ 3.1 Ä 2.2 CODIF 1.3 O+ VxH1_HIA_GSM VyH1_HIA_GSM

CODIF H+

VzH1_HIA_GSM NH1_HIA NHe2_HIA NH1_CODIF N01_CODIF NHe1_CODIF NHe2_CODIF 17:00 18:00 9.10 10.30 –0.59 –0.19 –7.02 –7.20 12.47 11.62 vue_web_new.cl

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

65

HIA. At 12:40 UT the instrument, while in the solar wind, was switched into mode 12 (magnetosheath mode). Both instruments switched then to an omnidirectional full energy sweep scheme (cf. Section 3.2.3), and the low-sensitivity side of HIA ceased collecting data. The solar wind beam became then detectable in the sunward looking direction spectrogram from the HIA high-sensitivity side (but with a lower energy resolution), and in the CODIF data, collected then from the high-sensitivity instrument side, but with a strong saturation in the solar wind beam energies. Note, however, the ability of the CODIF high-sensitivity side to detect the relatively weak fluxes of the upstreaming ion populations. The energy flux of these upstreaming ions is a factor of 2;000 lower than the energy flux of the solar wind beam ions, as this beam population is measured by the low-sensitivity side of HIA. At 14:40 UT the spacecraft entered in the magnetosheath, as revealed by the broader energy distribution of the detected ions (bow shock heating) and by the anisotropic flow shown in the four directional HIA spectrograms. These energy-time ion spectrograms are also available through the CIS home page: http://cluster.cesr.fr:8000/.

3.3.5 CIS Calibration Files CIS calibration files are ASCII files, self-documented (including comments), machine-readable and human-readable. They include parameters such as: Geometric factors MCP ion detection efficiencies: for each of the four main ions species, separately

for each anode, and including the polynomial coefficients for the efficiency energy dependence An overall instrument efficiency (called also absolute efficiency), for each of the four main ions species and separately for each instrument sensitivity side CODIF post-acceleration high voltage values Energy sweep tables for the various modes During how many spacecraft spins each 3-D ion distribution is accumulated, for the various modes

The way these calibration factors are used to convert instrument raw particle counts to higher-level data in physical units, such as particle flux or particle energy flux, and then to calculate the moments of the ion distribution functions, is described in Appendix 2 of the CAA-CIS Interface Control Document [2]. Both CODIF and HIA have been very well calibrated before launch, in vacuum test facilities [10]. However, due to the in-flight evolution of the MCP detection efficiencies as a function of time, the CIS calibration files are updated regularly. A calibrations catalogue file, which is provided with the calibration files, serves as a pointer to which calibration files to use for each data time period. This catalogue file evolves in an incremental way through the mission, to take into account the existence of new calibration files, which correspond to the instrument particle detection efficiency evolution, or other changes in the instrument.

66

I. Dandouras et al.

Fig. 3.7 CODIF high-sensitivity side absolute efficiency evolution for Cluster spacecraft 4, normalised to the efficiency at the start of the mission

1.2

Absolute Efficiency SC4/FM7

1

Efficiency

0.8

0.6

0.4

0.2

0 2002/1/1

2004/1/1 Day

2006/1/1

Calibration files include thus parameters that have been determined during ground calibrations in vacuum test facilities and are stable through the mission (e.g. instrument angle response, electrostatic analyser constant used in the calculation of the energy sweep tables, etc.), parameters that change gradually through the mission (e.g. particle detection efficiencies), and parameters that can be changed during the mission by command (e.g. upload of new spin accumulation tables for 3-D ion distributions in the various modes). All of them are equally important in converting raw data into physical units, and are used together. Figure 3.7 shows an example for the evolution, as a function of time, of the CODIF high-sensitivity side absolute efficiency for Cluster spacecraft 4, normalised to the efficiency at the start of the mission. Some sharp efficiency increases that are observed are due to an increased high voltage applied to the MCP, in order to compensate, as much as possible, for the efficiency decrease due to the well-known gain fatigue mechanism (cf. Section 3.2.4). However, as can be seen, the effects of such a high-voltage boost on the efficiency of the instrument are limited in time. Updating the calibration files for the detection efficiency evolution, with respect to the pre-launch calibrations, is a multi-step process, and is performed about once per year. The HIA ion density values are compared and cross-calibrated with the electron density values supplied by the Whisper sounder experiment [5]. This is performed in the magnetosheath for the high-sensitivity side and in the solar wind for the low-sensitivity side, i.e. in the plasma environment where each of the two sides has the optimum performance, and where the plasma energy spectrum is within the energy domain covered by the instrument. It should be noted also that the HIA anodes relative efficiencies are remarkably stable, i.e. the efficiency drift is very homogeneous between the anodes (11:25ı or 5:625ı sectors).

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

67

The CODIF calibrations updating is a more complex process. It involves the determination of the start-MCP efficiency, the stop-MCP efficiency, the fraction of coincidences between the “start” time-of-flight signal and the “stop” time-of-flight signal that also have a “single position” signal, allowing thus to calculate the total efficiency. In addition, the efficiencies of the individual anodes (22:5ı sectors) have to be cross-calibrated, since they slowly drift relative to each other, and this is performed by using time periods when the ion distributions are expected to be gyrotropic. CODIF calibrations involve also separate efficiencies determination for HC and OC ions. The resulting CODIF HC measurements are finally cross-checked with the HIA measurements, for periods when the plasma is composed mainly from HC ions. CODIF data are also cross-checked with the high-energy ion data supplied by the RAPID experiment [12], for periods where energetic plasma is present in the energy range of both instruments.

3.3.6 CIS Data Processing Software The CIS data processing software to be archived is composed of two packages: The CIS 3D MOM software, specially developed for the CAA, allows the user

to read the CIS Level 3 files, available at the CAA, and interactively calculate partial or total moments of the ion distributions, for selected energy and solid angle ranges. The input is 3-D ion distribution functions, in CEF-2 format and in corrected-for-efficiency particle count rate units, and the output is ion density, bulk velocity, pressure tensor and temperature, also in CEF-2 format and in the ISR2 pseudo-GSE reference frame. It is written in C and it can be used for either CODIF or HIA data. The cl software, written in IDL, and developed initially for the CIS team. It reads Level 1 CIS data and calibration files; is interactive, and can generate a large variety of high-resolution graphics (spectrograms, distribution functions, PADs . . . ) in several physical units. It can also export the results as ASCII (CEF) data files. In addition to the CIS data, this software can also read generic CDF and CEF data files, for correlation studies. These software packages are available “as documents”, for installation and execution on the end user’s machine.

3.3.7 CIS Documentation The CIS documentation for the CAA includes: The CIS-CAA Interface Control Document (ICD), including the appendices

“CIS Data Selection Guide” and “CIS: Particle Counts to Flux and to Moments of the Ion Distribution Function” [2]

68

I. Dandouras et al.

A description of the instrument [10] The User Guide to the CIS Measurements in the CAA Calibration documentation The present paper

3.3.8 CIS Data Caveats CIS caveat files include both general caveats for the CIS data, and caveats for specific data intervals. Given the complexity of an ion spectrometer, and the variety of its operational modes, each one being optimised for a different plasma region or measurement objective, consultation of the data caveats by the end user will always be a necessary step in the data analysis (independently of the data level). The CIS caveats are also available at the CIS home page, at http://cluster.cesr.fr: 8000/index.php?page=caveats&langue=en. Following are some of the most important caveats, which the user of the CIS data should be aware of. The calculated density values are in reality ion partial density values in the energy

domain covered by the instrument, typically 25 eV/e to 40 keV/e for CODIF and 5 eV/e to 32 keV/e for HIA. In the presence of cold plasma at energies below the instrument energy threshold, or of hot plasma at energies above the instrument upper energy limit, this partial density is evidently lower than the total plasma density. In the presence of hot plasma at energies above the instrument upper energy limit the result is also underestimated ion bulk velocity, temperature and pressure values. The instrument energy domain is always defined with respect to the spacecraft potential. Spacecraft charging to a positive floating potential, as can be the case in low density plasmas when the ASPOC ion emitter for spacecraft potential control is not operating [11], repels the low-energy ions which in these cases cannot be detected by CIS (cf. for example Section 6.10 of R`eme et al. [10], and Section 3.4 of Dandouras et al. [3]). This effect results in a further increase of the difference between the partial density, measured by CIS, and the total plasma density. In addition to the finite energy range, the accuracy of the computed moments is also affected by the finite energy and angular resolution of the two instruments. This is in particular the case for narrow-energy distributions (cold plasmas or cold beams), and/or narrow solid angle beams. CODIF is more sensitive to this effect, due to its coarser energy and angular resolution. Adequate counting statistics are essential for reliable results. For density calculations, in the case of isotropic plasma distributions, HIA “hits” a lower limit density value of the order of 0:01–0:02 cm3 . CODIF continues to supply density values down to about 0:001–0:002 cm3 , although with very poor counting statistics in these cases, which do not allow a reliable calculation of the higherorder moments.

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

69

Ion measurements are sensitive to possible detector saturation in the presence

of high ion fluxes (instrument dead time effects). The most striking example is the strongly underestimated solar wind density and magnetosheath density by the CODIF high-sensitivity side, while the instrument is operating in a magnetospheric mode. However, these saturation effects are phase space dependent: the detector can be saturated only in a limited energy range and for particles arriving in a given solid angle, where the highest counting rates occur (e.g. solar wind beam), but at the same time can supply reliable measurements in the remaining phase space, as are for example the upstreaming ion populations discussed in Section 3.3.4 (CODIF data spectrogram). Note that HIA data acquired in a magnetospheric mode can also suffer from partial detector saturation when in the solar wind: slightly underestimated densities, by a factor depending on the solar wind density and temperature [7]. Eventual instrument background counts due to penetrating particles, from the radiation belts around perigee passes or during SEP (Solar Energetic Particles) events, can result in an overestimation of the measured fluxes, and the resulting density values. This effect is stronger for HIA, whereas for CODIF it is mitigated by the time-of-flight technique, which requires both a “start” signal and a “stop” signal, in the correct time-of-flight window, to validate the detection of an ion. The entire velocity phase space, corresponding to the instrument energy domain, is not covered during all modes (cf. for example the spectrogram discussed in Section 3.3.4). This is the case during solar wind modes, and during some magnetosheath modes (CODIF data acquired after 1 November 2003 during modes 12 and 14). The result is: ı Incorrect HIA moment values in the magnetosheath when the instrument is in a solar wind mode (modes 0–5) ı Incorrect CODIF moment values when data come from the high-sensitivity side and the instrument is in a solar wind mode (modes 0–5) ı Incorrect CODIF moment values when data come from the high-sensitivity side and the instrument is in mode 12 or 14, for data acquired after 1 November 2003 CIS mode information is supplied in the spectrograms (cf. Section 3.3.4) and in the Level 3 data (cis mode variable). CODIF data in the solar wind, if acquired from the low-sensitivity side but with

the instrument operating in a magnetospheric mode, can suffer from cross-talk with the high counting rates of the high-sensitivity side, which is then saturated by the solar wind. The result is then overestimated densities and underestimated velocities. The Vz term of the ion bulk velocity is very sensitive to the anode crosscalibrations, and in particular to those of the anodes looking in directions away from the spacecraft spin plane (polar anodes). In some cases the efficiency calibration coefficients cannot completely compensate for strongly asymmetrically decreased efficiencies of such polar anodes, which results in a residual offset of Vz . This is in particular the case of CODIF onboard Cluster sc3, where one

70

I. Dandouras et al.

MCP quadrant (southward looking directions on the high-sensitivity side) suffers from a strongly decreased particle detection efficiency. This asymmetric efficiency degradation amplified with time, and for data acquired after 23 February 2003 it cannot anymore be corrected. As a result the Vz velocity component is highly unreliable (strong offset), and the density suffers from the fact that the instrument under-detects the ions in one hemisphere of the phase space. CODIF mass-separated data can suffer from spillover between neighboring mass channels. HeCC data, in particular, are usually contaminated by HC ions, resulting in over-estimated HeCC densities. CODIF OC data can be contaminated by penetrating particles in the radiation belts. They can also be slightly contaminated by HC ions in very high HC flux plasmas, as for example in the magnetosheath. The latter is due to occasionally two uncorrelated HC ions, one generating only a “start” time-of-flight signal and the other generating only a “stop” time-of-flight signal, and the time difference between the two being that of an OC ion. Presence of ions other than HC in the HIA data results in underestimated densities and in overestimated temperatures and pressures. During short eclipses the absence of a Sun reference pulse completely desynchronises the HIA data, which become useless, whereas the CODIF data are acquired using an extrapolated Sun reference pulse, and are thus still usable. During long eclipses CIS is off. Determinations of the evolution of the MCPs detection efficiency are statistical (based on the analysis of a number of data intervals). Detector ageing is accompanied by a degradation of the signal-to-noise ratio. Operation incidents can occasionally result, for short periods, in missing data products, in incorrect MCP high voltage and/or discriminator settings, which reduce the accuracy of the collected data, etc. These are flagged in the caveats for specific data intervals.

3.4 Conclusion The CIS experiment, due to its unique features, is a versatile instrument package, capable of obtaining full three-dimensional ion distributions with a time resolution of one spacecraft spin and with mass-per-charge composition determination. The quality of the data supplied by CIS has resulted, until now, in the publication of more than 300 scientific papers in refereed journals, based on the analysis of these data. These papers, and the results they present, constitute the ultimate validation of the data supplied by the two complementary instruments composing the CIS experiment, CODIF and HIA. The Cluster Active Archive will preserve these data in the long-term, so that they are usable by the scientific community. For the CIS data archived there a multilevel approach has been adopted, including both processed raw data and calibrated

3

Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA)

71

high-resolution data in a variety of physical units. The CIS data archive includes also experiment documentation, graphical products for browsing through the data, and data caveats. Given the complexity of an ion spectrometer, and the variety of its operational modes, each one being optimised for a different plasma region or measurement objective, the user has to be careful in selecting for his analysis the appropriate data set and instrument mode. In order to help the user in this task, a User Guide to the CIS Measurements in the CAA has been provided. Consultation of the data caveats by the user will also always be a necessary step in the data analysis. Acknowledgements The CIS Team is grateful to ESA, for the support supplied in operating the instrument and in preparing the CIS data for the CAA. CIS archival activities at CESR are jointly funded by ESA, by CNES and by CNRS. Activities in the USA are funded by NASA.

References 1. Daly, P.W. et al.: Users Guide to the Cluster Science Data System. DS–MPA–TN–0015 (2002) 2. Dandouras, I., A. Barthe: Cluster Active Archive: Interface Control Document for CIS. CAACIS-ICD-0001 (2007). http://caa.estec.esa.int/caa/instr doc.xml 3. Dandouras, I., V. Pierrard, J. Goldstein, C. Vallat, G. K. Parks, H. R`eme, C. Gouillart, F. Sevestre, M. McCarthy, L. M. Kistler, B. Klecker, A. Korth, M. B. Bavassano-Cattaneo, P. Escoubet, A. Masson: Multipoint observations of ionic structures in the Plasmasphere by CLUSTER-CIS and comparisons with IMAGE-EUV observations and with Model Simulations. In: AGU Monograph: Inner Magnetosphere Interactions: New Perspectives from Imaging, 159, 23–53, 10.1029/159GM03 (2005) 4. Dandouras, I., A. Barthe, E. Penou, H. R`eme, S. McCaffrey, C. Vallat, L.M. Kistler, and the CIS Team: Archival of the Cluster Ion Spectrometry (CIS) Data in the Cluster Active Archive (CAA). In: Proceeding of the Cluster and Double Star Symposium-5th Anniversary of Cluster in Space, ESA SP-598, Noordwjik (2006) 5. D´ecr´eau, P.M.E., P. Fergeau, V. Krasnoselskikh, E. Le Guirriec, M. L´evˆeque, Ph. Martin, O. Randriamboarison, J.L. Rauch, F.X. Sen´e, H.C. S´eran, J.G. Trotignon, P. Canu, N. Cornilleau, H. de F´eraudy, H. Alleyne, K. Yearby, P.B. M¨ogensen, G. Gustafsson, M. Andr´e, D.C. Gurnett, F. Darrouzet, J. Lemaire, C.C. Harvey, P. Travnicek: Early results from the Whisper instrument on Cluster: an overview. Ann. Geophys., 19, 1241 (2001) 6. Harvey, C.C., A.J. Allen, F. D´eriot, C. Huc, M. Nonon-Latapie, C.H. Perry, S.J. Schwartz, T. Eriksson, S. McCaffrey: Cluster Metadata Dictionary. CAA-CDPP-TN-0002 (2008) 7. Martz, C., J.A. Sauvaud, H. R`eme : Accuracy of ion distribution measurements and related parameters using the Cluster CIS experiment. In: Spatio-Temporal Analysis for Resolving Plasma Turbulence (START), ESA WPP-047, 229 (1993) 8. Perry, C., T. Eriksson, P. Escoubet, S. Esson, H. Laakso, S. McCaffrey, T. Sanderson, H. Bowen, A. Allen, C. Harvey: The ESA Cluster Active Archive. In: Proceeding of the Cluster and Double Star Symposium-5th Anniversary of Cluster in Space, ESA SP-598, Noordwjik (2006) 9. Prince, R.H., J.A. Cross: Gain Fatigue Mechanism in Channel Electron Multipliers. Rev. Sci. Instrum., 42, 66; DOI:10.1063/1.1684879 (1971) 10. R`eme, H., C. Aoustin, J.M. Bosqued, I. Dandouras, B. Lavraud, J.A. Sauvaud, A. Barthe, J. Bouyssou, Th. Camus, O. Coeur-Joly, A. Cros, J. Cuvilo, F. Ducay, Y. Garbarowitz, J.L. Medale, E. Penou, H. Perrier, D. Romefort, J. Rouzaud, C. Vallat, D. Alcayd´e, C. Jacquey, C. Mazelle, C. dˆaUston, E. M¨obius, L.M. Kistler, K. Crocker, M. Granoff, C. Mouikis,

72

I. Dandouras et al.

M. Popecki, M. Vosbury, B. Klecker, D. Hovestadt, H. Kucharek, E. Kuenneth, G. Paschmann, M. Scholer, N. Sckopke, E. Seidenschwang, C.W. Carlson, D.W. Curtis, C. Ingraham, R.P. Lin, J.P. McFadden, G.K. Parks, T. Phan, V. Formisano, E. Amata, M.B. Bavassano-Cattaneo, P. Baldetti, R. Bruno, G. Chionchio, A. Di Lellis, M.F. Marcucci, G. Pallocchia, A. Korth, P.W. Daly, B. Graeve, H. Rosenbauer, V. Vasyliunas, M. McCarthy, M. Wilber, L. Eliasson, R. Lundin, S. Olsen, E.G. Shelley, S. Fuselier, A.G. Ghielmetti, W. Lennartsson, C.P. Escoubet, H. Balsiger, R. Friedel, J-B. Cao, R. A. Kovrazhkin, I. Papamastorakis, R. Pellat, J. Scudder, B. Sonnerup: First multispacecraft ion measurements in and near the Earth’s magnetosphere with the identical Cluster ion spectrometry (CIS) experiment. Ann. Geophys., 19, 1303 (2001) 11. Torkar, K., W. Riedler, C.P. Escoubet, M. Fehringer, R. Schmidt, R.J.L. Grard, H. Arends, F. R¨udenauer, W. Steiger, B.T. Narheim, K. Svenes, R. Torbert, M. Andr´e, A. Fazakerley, R. Goldstein, R.C. Olsen, A. Pedersen, E. Whipple, H. Zhao: Active spacecraft potential control for Cluster implementation and first results. Ann. Geophys., 19, 1289 (2001) 12. Wilken, B., P.W. Daly, U. Mall, K. Aarsnes, D.N. Baker, R.D. Belian, J.B. Blake, H. Borg, J. B¨uchner, M. Carter, J.F. Fennell, R. Friedel, T.A. Fritz, F. Gliem, M. Grande, K. Kecskemety, G. Kettmann, A. Korth, S. Livi, S. McKenna-Lawlor, K. Mursula, B. Nikutowski, C.H. Perry, Z.Y. Pu, J. Roeder, G.D. Reeves, E.T. Sarris, I. Sandahl, F. Søraas, J. Woch, Q.-G. Zong: First results from the RAPID imaging energetic particle spectrometer on board Cluster. Ann. Geophys., 19, 1355 (2001)

Chapter 4

Digital Wave Processor Products in the Cluster Active Archive K.H. Yearby, H.St.C. Alleyne, S.N. Walker, I. Bates, M.P. Gough, A. Buckley, and T.D. Carozzi

Abstract The Digital Wave Processor (DWP) is the central control and data processing instrument for the Cluster Wave Experiment Consortium. DWP products in the Cluster Active Archive (CAA) provide a mainly supporting function for the rest of the consortium. This includes a time correction dataset which allows the standard timing accuracy of 2 ms to be improved to around 20 s, and experiment command and status datasets which show what commands have been sent to the experiments, and the resulting status. DWP also contains a particle correlator experiment that computes the auto-correlation of electron counts received by the PEACE electron experiment via an inter-experiment link.

4.1 Introduction The DWP instrument is the central control and data processing unit for the Wave Experiment Consortium (WEC) which comprises the EFW, STAFF, WHISPER and WIDEBAND instruments. Pedersen et al. [8] give an overview of the WEC, while Woolliscroft et al. [9] describe the DWP instrument. A block diagram of the consortium is shown in Fig. 4.1). For nominal operations, all commanding and telemetry acquisition for these instruments is routed via DWP, except for WIDEBAND telemetry to NASA’s Deep Space Network (DSN) and the Czech Republic’s Panska Ves station. Time synchronisation of WEC modes is controlled by DWP, and the WEC science telemetry is dynamically allocated between the WEC instruments according to the requirements of each WEC mode. Housekeeping data for all WEC instruments is acquired by DWP. In addition DWP contains a particle correlator experiment which computes K.H. Yearby (), H.St.C. Alleyne, S.N. Walker, and I. Bates Department of Automatic Control and Systems Engineering, University of Sheffield, Mappin Street, Sheffield, S1 3JD, UK e-mail: [email protected]; [email protected]; [email protected] M.P. Gough, A. Buckley, and T.D. Carozzi Space Science Group, University of Sussex, Falmer, Brighton, BN1 9QH, UK e-mail: [email protected]; [email protected]; [email protected]

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 4, c Springer Science+Business Media B.V. 2010

73

74

K.H. Yearby et al.

STAFF Wave Analyser

Inter-experimental links to other instruments of the Cluster scientific payload Spacecraft telecommand interface

STAFF Spectrum Analyser

EFW

DWP

WBD Wideband data

WHISPER

Spacecraft data interface

Direct link for WBD data

WEC Power supply

Spacecraft power interface

Fig. 4.1 Block diagram of the WEC showing DWP in the centre of the consortium

the auto-correlation of electron counts received by the PEACE experiment [7]. The correlator was designed to detect modulations and short time particle bursts in the electron population as an indicator of wave-particle interactions. The DWP products in the CAA consist of the following datasets: Time correction data (C1 CP DWP TCOR). Enable the standard 2 ms timing ac-

curacy to be improved to 20 s Summary of the status of the Wave Experiment Consortium (C1 CP DWP

LOG). One record for each interval that the WEC operates in the same mode. A total of 64 parameters including: instrument modes, telemetry use statistics, errors and anomalies, summary of voltage and temperature housekeeping A listing of the command sequences uplinked to the Wave Experiment Consortium (CM CD DWP UT PIOR) High resolution particle correlator data (C1 CP DWP COR FX and C1 CP DWP COR ST) Documentation and software Their production and interpretation is outlined in the following sections.

4

Digital Wave Processor Products in the Cluster Active Archive

75

4.2 Time Correction Data 4.2.1 Introduction The STAFF and EFW experiments include magnetic and electric field wave data at frequencies up to 180 Hz in burst mode [2, 6]. Inter-spacecraft comparison of this data requires high timing accuracy – for example 1ı of phase at 180 Hz is 15 s. The standard timing accuracy of the Cluster mission is 2 ms, and in the worst case, this corresponds to a 4 ms error between spacecraft which is 260ı of phase at 180 Hz. In the normal bit rate, for which the maximum frequency is 10 Hz, the maximum phase error becomes 12ı . Fortunately a considerable improvement in timing accuracy is possible. To understand how this can be achieved it is first necessary to explain how the standard time stamping of Cluster data is performed by the European Space Operations Centre (ESOC). Most Cluster data is not acquired from the spacecraft in real time, but is recorded on an onboard solid state recorder. All data released to experimenters (both recorded onboard and acquired in real time) are time stamped using an onboard clock which is calibrated to UTC using a time correlation process. Until recently, the error of this onboard clock was allowed to increase to the ˙2 ms limit before a new calibration was applied. The time correlation process uses the real time data which is time stamped by the ground stations with the Earth Receive Time. By subtracting various delays, including the propagation time from spacecraft to ground, the Spacecraft Event Time (SCET) is obtained. The total error for real time data is estimated to be 11 s. During each real time pass, the difference (DIFF) between the SCET measured by the onboard clock, and that derived from the Earth Receive Time is monitored. When the difference exceeds 2 ms a new time correlation is performed. This process is described in detail in the ESOC report CL-OPS-TN-1004-OPS-OPC provided to the DWP team as a private communication, but this should be available from the CAA in due course. The DIFF measurements can be used to apply a correction to the onboard clock to bring its accuracy close to that of the real time data. The DIFF is slowly varying, so a simple linear interpolation can be used to estimate the DIFF between real time passes. An independent set of DIFF measurements are also available during real time data acquisition for the Wideband experiment [5] from the ground stations of DSN and Panska Ves. Detailed analysis of the time stamps has revealed that a second correction is needed. When data are recorded onboard, the onboard time counter recorded in the data packets is not precisely correlated with the timing reference pulse distributed to the experiments but may be subject to an offset which is fixed for each period of the same telemetry mode. This term is called OFFSET and specifies the difference between the time of the reference pulse and the onboard time stamp recorded in the data packets. It applies only to data recorded onboard, and is zero for real time data. OFFSET normally ranges ˙180 s for normal mode data, and ˙30 s in burst modes.

76

K.H. Yearby et al.

From 2007 November 23, a new procedure has been adopted at ESOC. A time correlation is performed during every nominal pass, instead of waiting until the DIFF reaches 2 ms. DIFF is now usually less than 20 s, but this is not guaranteed. While a correction for the OFFSET term is in principle still needed, for burst mode data (when the most accurate timing is needed), this is less than 30 s, so overall timing will usually be accurate to ˙50 s. This corresponds to 100 s between spacecraft or 6ı of phase at 180 Hz which may be good enough for many purposes. It will still be possible to use the time correction dataset when the highest possible accuracy is needed.

4.2.2 Production The DWP team use the time difference measured by ESOC and DSN to prepare the time correction dataset. Both ESOC and DSN timing are sometimes subject to errors due to ground station configuration problems, so these data are subject to a manual validation process, illustrated in Fig. 4.2. It is observed that most of the points fall close to a smooth curve which is taken to be the correct value. The result is a set of DIFF measurements valid at specific points in time (during ESA or DSN ground station passes). The OFFSET term is determined by analysis of the time stamps of the housekeeping telemetry. The onboard time (OBT) is first converted to its raw 56 bit binary format. The telemetry format generator issues the reference pulse every 5.15222168 s, which is exactly 86439936 OBT counts. Therefore OBT modulo 86439936 should be constant. In practice the OBT modulo changes at every telemetry mode change,

1000

DIFF [SCETrt - SCETobt] (us)

500

Cluster SC1 ESOC & WBD DIFF for January 2002 ESOC DIFF WBD diff Valid DIFF

0 –500 –1000 –1500 –2000 –2500 –3000 2002-01-01 2002-01-06 2002-01-11 2002-01-16 2002-01-21 2002-01-26 2002-01-31 Date

Fig. 4.2 Plot of the DIFF values measured by ESOC (C) and DSN ( ). The solid line represents the assumed correct value used to prepare the time corrections

4

Digital Wave Processor Products in the Cluster Active Archive

77

but tends to return to the same value for each period of real time telemetry. The deviation from the real time value is the OFFSET. Finally, the DIFF and OFFSET values are merged into the TCOR files. Linear interpolation of the original DIFF dataset is used to obtain values at the times of OFFSET changes.

4.2.3 Validation Validation of the TCOR dataset is performed by computing the difference between the time stamps of successive HK formats. The maximum, minimum, mean, and standard deviation from the nominal value are computed for each 24 h period. Values larger than about 3 s indicate an error in the production. These values are removed for the first issue of the dataset.

4.2.4 Caveats Users of the time correction dataset should be aware of its limitations. Most importantly, TCOR data is not available at all times. In the first issue of the dataset, any data that fails validation is simply deleted from the files. TCOR coverage is typically around 90%. Applications requiring accurate timing should confirm that TCOR data is available at the relevant time. Time corrections are provided at the start and end of each period of the same telemetry mode. Time corrections at other times may be obtained by interpolation subject to certain rules. The OFFSET term is constant throughout each period, and the same value will be written in the records at the start and end of the period. If the OFFSET values before and after the required time are different, or either has the fill value of –1e31, then OFFSET is not available for that period. No interpolation between different OFFSET values is allowed. The DIFF may be obtained by linear interpolation of the DIFF values immediately before and after the required time. However, if either DIFF has the fill value of –1e31, then DIFF is not available for that period. It is not allowed to interpolate over a fill value.

4.2.5 Application For WEC data, the time corrections may be applied using the latest version of the Telemetry Extraction and Decommutation (TED) package (see Section 4.5). This is done automatically once TED and the TCOR files are correctly installed. By default, TED 2.5 produces Decommutated Science Data (DSD) files with all time corrections applied. A new bit in the diagnostic word (HKTCOR CORRECTED) will be set when time corrections have been successfully applied. This bit should be checked by applications requiring accurate timing.

78

K.H. Yearby et al.

It is also in principle possible to apply time corrections to calibrated high level science data, provided that micro-second level precision has been maintained during the production of that data. The basic method is to obtain the DIFF and OFFSET at the required time, and add these to the data time tag.

4.3 Experiment Status and Commanding Datasets 4.3.1 Status Log (DWP LOG) The status log contains information regarding the overall status of the WEC instruments. The dataset contains one record for each interval that the WEC operates in the same mode. Each record has a total of 64 parameters including:

Time interval Instrument modes Telemetry use statistics Errors and anomalies Summary of voltage and temperature housekeeping

The dataset is produced by automatic analysis of WEC telemetry, and validated by visual inspection. Figure 4.3 shows a typical record from this dataset. Interval: Mode desc.: STAFF SA: Whisper:

2002-01-01T09:57:21Z/2002-01-01T23:34:45Z, (9519 formats) NBR 52s recurrence mode (SWECJ275) NM1, NM1b 4F/CB1050, 1A/AD12500E

TM mode: Model tag: Log event: DWP config: TM overflows: DWP errors: STA SA errors: Correlator:

NM CD09 MACRO FULL, FULL, FULL 0 0 0 ( 42 zeros) COR0, MAR

WEC current: STAFF SA +6V: STAFF SA +5V: STAFF SC +9V: STAFF SC +6V: Wideband +6V:

538.7869 6.0886 5.4880 9.1271 5.7319 6.0566

DWP temp: STAFF SA: Wideband: EFW TM: STAFF SA TM: Correl TM: Unused TM: Total TM:

± 15.2694 ± 0.0013 ± 0.0169 ± 0.0037 ± 0.0105 ± 0.0064

23.45 °C 24.11 °C 26.15 °C 1472 1640 135 81 5217

bps bps bps bps bps

WEC mode: NM52 Clock: 900.0208 Hz SEU count: 0 Checksum: 0000 AP overflows: 0 Whisper errors: 0 STA SC errors: 0 Fixed level: 15 DWP voltage: STAFF SA -6V:

4.8900 ± 0.0536 -5.9049 ± 0.0023

STAFF SC -9V: STAFF SC -6V:

-9.1652 ± 0.0117 -5.7730 ± 0.0310

Whisper temp: STAFF SC:

33.48 °C 20.45 °C

Whisper TM: STAFF SC TM: Wideband TM: TM overhead:

902 960 0 28

bps bps bps bps

Fig. 4.3 One record from the WEC status log dataset showing period from 09:57 to 23:34 on 1 January 2002 for spacecraft 1 (the spacecraft identification is contained within the metadata not shown in this figure)

4

Digital Wave Processor Products in the Cluster Active Archive

79

4.3.2 Commanding (DWP UT PIOR) This dataset is a record of the command sequences and parameters that have been used to control the WEC. A single dataset covers the four Cluster spacecraft, with a spacecraft mask field used to indicate which spacecraft each command sequence applies to. Normally, all four spacecraft are commanded in the same way, so this format is concise, and helps to highlight exceptions. Such exceptions include special commanding needed to allow for the partial failure of some electric field antennae.

4.4 Particle Correlator Datasets 4.4.1 Description The particle correlator experiment computes the Auto-Correlation Functions (ACF) of electron counts received by the PEACE HEEA sensor via an inter-experiment link. The ACFs are constructed in correlator electron energy bands which correspond to either two or four PEACE energy levels. Constraints of processing and WEC telemetry allows ACFs for only two of the possible 15 energy bands to be processed at any particular time. One of these energy bands is pre-selected while the other available band steps sequentially through the remaining 14 energies, once per spin. Individual ACFs are constructed within separate 1.111 ms DWP clock cycles. The time series of electron counts is of duration 732 s and comprises 61 “count bins” of 12 s duration. Zero to 31 lags are performed on each individual time series to produce a 32-point ACF. Individual ACFs based on separate 1.111 ms DWP clock cycles are summed lag-for-lag on board and the summed ACF is transmitted in telelmetry. This gives the particle correlator the following characteristics: Frequency range 1.4–41.6 KHz in 32 frequency bands and DC to 4 Hz based on successive ACF outputs Energy range 0.6 eV to 26 keV in 15 energy bands (PEACE mode dependent)

The correlator data consists of two identically formatted datasets, one for the fixed energy band (DWP COR FX), and the other for the stepped energy band (DWP COR ST). The full specification is given in Section 5.1 of the DWP CAA ICD [3]. The ACF amplitude is given in counts squared per unit time. The count rate obtained depends on both the particle flux (physical effect) and the geometric factor of the sensor and gain of the micro-channel plates (instrumental effects). It was decided that it was not feasible to provide CAA correlator ACF amplitudes recalibrated to take account of the varying HEEA sensor gain. However, all necessary supporting data (calibrated energy level, lag shift duration and look direction) are included. Results from this experiment were presented by Buckley et al. [1] and Gough et al. [4].

80

K.H. Yearby et al.

4.4.2 Caveats Some artifacts are illustrated in Fig. 4.4 which shows the validation plot for 23 April 2002, spacecraft 3. Correlator data is present only when the PEACE HEEA sensor is in use. In the figure, data is absent when PEACE is switched off around perigee (2–4 UTC). In addition there are some periods where, although HEEA is on, the particle count rates are so low that the ACFs are mostly zeros (12 UTC onwards). The energy range corresponds to that being scanned by the PEACE HEEA sensor at any particular time. The energy may be undefined if the PEACE and correlator sweep modes were not compatible (i.e. ACFs acquired at widely different energies have been summed together and no unique energy calibration is possible). The behaviour of the polar angle depends on how PEACE selects the polar zone for the correlator. Early in the mission, the polar zone was selected according to the magnetic field direction, so the polar angle tends to stay on a single value for long periods. Later, the zone stepped through the 12 possible zones, at one step per spin. The azimuth angle is averaged over a spin except in burst modes, where there may be 4, 8, or 16 summed ACFs output per spin. The period 8 to 12 UTC shows a sloping ACF which is due to steep edges in the energy spectrum causing a rapid change in count rate with time during the energy

40 20 0 40

ACF bin

20 0 30 20 10 0

Azim polar log ev

fix kHz

step kHz

Cluster/DSP DWP correlator summary plot: spacecraft: 3, date: 2002-04-23 File: C3_CP_DWP_COR_FX_20020423_V01, Valid records: 18236, invalid: 0 File: C3_CP_DWP_COR_ST_20020423_V01, Valid records: 18286, invalid: 0 Averaged spectrum (stepped, fixed) and ACF

4 2 –0 –0 360 180 –0

Averaged count rate, versus energy, polar angle and azimuth

0

2

4

6

8

10 12 Hour of day

14

16

18

20

22

24

Fig. 4.4 Correlator data validation plot for 23 April 2002, spacecraft 3. The top two panels show the Fourier transform of the ACF for the stepped and fixed energy bands, respectively. The next panel is the ACF itself, summed over fixed and stepped energies. The lower three panels show the count rate as a function of energy level, polar angle and azimuth angle (undefined in this case)

4

Digital Wave Processor Products in the Cluster Active Archive

81

sweep. This period also shows a correlation at 36 s period (variable, and different on each S/C) and associated high frequency interference line. This effect is present whenever the count rate is very high, and is believed to be caused by a modulation of the gain of the detectors.

4.5 Software The WEC telemetry extraction and decommutation (TED) software is archived for reference purposes only. No DWP software runs in the archive. The latest version is 2.5.0.108 and appears satisfactory for those using it. This version includes support for using the TCOR dataset to obtain 20 s timing accuracy. The TED software is the first stage in the processing of WEC science data by all WEC teams, with the exception of Wideband data telemetered directly to DSN or Panska Ves. It can be used in two ways: 1. As a stand alone application (levelone) reading the RDM, extracting the data for one or more WEC instruments, and writing it to decommutated science data (DSD) files 2. As a library, allowing application software to read the WEC RDM directly Several earlier versions of TED exist and may still be in use by WEC teams. Version 2.4.3 is satisfactory where the standard 2 ms timing accuracy is sufficient. It does not support the use of time correction files. TED version 3 was a major rewrite which unfortunately introduced several new problems, and was therefore abandoned. Version 2.5.0 was based on version 2.4.3 and correctly supported the time correction files. It contained one bug affecting only Wideband data obtained in the burst mode (BM2) which was corrected in version 2.5.0.108. Version 2.5.0.108 is available from the CAA on the DWP documentation page in file: http://caa.estec.esa.int/documents/teams/DWP/ted-2.5.0.108.tar.gz

References 1. Buckley, A.M., Gough, M.P., Alleyne, H., Yearby, K.H., Walker, S.N.: First measurements of electron modulations by the Particle Correlator experiments on Cluster. ESA Spec. Publ. 492, 19–26 (2001) 2. Cornilleau-Wehrlin, N., Chauveau, P., Louis, S., Meyer, A., Nappa, J.M., Perraut, S., Rezeau, L., Robert, P., Roux, A., de Villedary, C., de Conchy Y., Friel, L., Harvey, C.C., Hubert, D., Lacombe, C., Manning, R., Wouters, F., Lefeuvre, F., Parrot, M., Pinc¸on, J.L., Poirier, B., Kofman,W., and Louarn, Ph.: The CLUSTER Spatio-Temporal Analysis of Field Fluctuations (STAFF) Experiment. Space Sci. Rev. 79, 107–136 (1997) 3. DWP team: Cluster Active Archive: Interface Control Document for DWP. Cluster Active Archive. http://caa.estec.esa.int/caa/instr doc.xml (2005), Accessed 28 May 2008. 4. Gough, M.P., Buckley, A.M., Carozzi, T.D., and Beloff, N.: Experimental Studies of Wave Particle Interactions in Space Using Particle Correlators: Results & Future Developments. Adv. Space Res. 32, 407–416 (2003)

82

K.H. Yearby et al.

5. Gurnett, D.A., Huff, R.L., Kirchner, D.L.: The wide-band plasma wave investigation. Space Sci. Rev. 79, 195–208 (1997) 6. Gustafsson, G., Bostrom, R., Holback, B., Holmgren, G., Lundgren, A., Stasiewicz, K., Ahlen, L., Mozer, F.S., Pankow, D., Harvey, P., Berg, P., Ulrich, R., Pedersen, A., Schmidt, R., Butler, A., Fransen, A.W.C., Klinge, D., Thomsen, M., Falthammar, C.G., Lindqvist, P.A., Christenson, S., Holtet, J., Lybekk, B., Sten, T.A., Tanskanen, P., Lappalainen, K., Wygant, J.: The electric field and wave experiment for the Cluster mission. Space Sci. Rev. 79, 137–156 (1997) 7. Johnstone, A.D., Alsop, C., Burge, S., Carter, P.J., Coates, A.J., Coker, A.J., Fazakerley, A.N., Grande, M., Gowen, R.A., Gurgiolo, C., Hancock, B.K., Narheim, B., Preece, A., Sheather, P.H., Winningham, J.D., Woodliffe, R.D.: Peace: A Plasma Electron and Current Experiment. Space Sci. Rev. 79, 351–398 (1997) 8. Pedersen, A., Cornilleau-Wehrlin, N., de la Porte, B., Roux, A., Bouabdellah, A., D´ecr´eau, P.M.E., Lefeuvre, F., Sen´e, F.X., Gurnett, D., R. Huff, R., Gustafsson, G., Holmgren, G., Woolliscroft, L.J.C., Thompson, J.A., and Davies, P.H.N.: The Wave Experiment Consortium (WEC). Space Sci. Rev. 79, 93–106 (1997) 9. Woolliscroft, L.J.C., Alleyne, H.St.C., Dunford, C.M., Sumner, A., Thompson, J.A., Walker, S.N., Yearby, K.H., Buckley, A., Chapman, S., Gough, M.P., and the DWP Co-investigators: The digital wave processing experiment on Cluster. Space Sci. Rev. 79, 209–231 (1997)

Chapter 5

EDI Data Products in the Cluster Active Archive E. Georgescu, P. Puhl-Quinn, H. Vaith, M. Chutter, J. Quinn, G. Paschmann, and R. Torbert

Abstract The Electron Drift Instrument (EDI) contribution to the Cluster Active Archive (CAA) is described. Presented are descriptions of the EDI instrument, the various CAA/EDI data products, the CAA ingestion schedule and the current EDI status. An example of a science application is given for one of the main EDI data products available in the CAA.

5.1 Introduction The contribution of the Electron Drift Instrument (EDI) to the CAA is completely described by the Interface Control Document (ICD) available online [1]. The ICD contains, in addition to the description of the convention on data formats, metadata and delivery to the archive, a section on instrument description, dealing with the scientific objectives of the EDI experiment, a hardware overview, and the description of the data processing chain and of the data products. This paper presents the most interesting aspects of this document for the user of the archive. It briefly describes the EDI experiment, the data processing chain and the EDI data products in the archive and it gives an example of scientific use of one of the main data products. EDI delivers data products and documentation to the archive. The data range E. Georgescu () Max-Planck-Institute for Solar Research, Katlenburg-Lindau, Germany e-mail: [email protected] P. Puhl-Quinn, H. Vaith, M. Chutter, and R. Torbert University of New Hampshire, Durham, USA e-mail: [email protected]; [email protected]; [email protected]; [email protected] J. Quinn Boston University, Boston, MA, USA e-mail: [email protected] G. Paschmann Max-Planck-Institute for Extraterrestrial Physics, Garching, Germany e-mail: [email protected]

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 5, c Springer Science+Business Media B.V. 2010

83

84

E. Georgescu et al.

from the complete set of raw data to auxiliary and final products. The documentation refers to the already mentioned ICD, to the instrument user manual, instrument reference papers and software scripts.

5.2 Instrument Overview The EDI instrument and its modes of operation are described in detail by Paschmann et al. [2] and Quinn et al. [3]. EDI was designed to measure the electric field E? perpendicular to the ambient magnetic field B, when the gradient of B in the plane perpendicular to it can be neglected. This is done in an active mode using the electron drift technique. EDI also operates in a passive mode, whereby ambient electrons are detected.

5.2.1 Electric Field Measurement The aim is to determine the drift velocity of the injected electrons Vd induced by an electric field E? and/or a magnetic-field gradient rB? . Under the assumption that the rB? -drift can be ignored, the electric field can be computed according to: E? Š Vd B

(5.1)

The basis of the electron-drift technique is the injection of weak beams of electrons perpendicularly to the ambient magnetic field and their detection after one or more gyrations. In the presence of a drift velocity, the circular electron orbits are distorted into cycloids. Their shape depends on whether the beam is injected with a component parallel or anti-parallel to the drift velocity. The lengths of the two orbits and the electron travel times differ. To be able to realize both types of orbits simultaneously, EDI uses two guns and two detectors. Figure 5.1 shows examples of these two orbits in the plane perpendicular to B, which we refer to as the B? -plane. The instrument records the positions and firing directions of the guns and the times of flight of the electrons. The magnetic field information is received through the inter-experimental link from the FGM (flux-gate magnetometer). The drift velocity can be determined by two methods: triangulation and time-offlight. (a) The basis of the triangulation technique, explained in detail in [3], is the determination of the ‘drift-step’ vector d which is the displacement of the electrons after a gyro time Tg . The drift velocity Vd is: Vd D d=Tg

(5.2)

The location in the B? -plane, at which electrons reach the detector after one gyration, can be viewed as the ‘target’ for the electron beams.

5

EDI Data Products in the Cluster Active Archive

85

Fig. 5.1 Principle of operation for EDI

(b) The time-of-flight technique uses the difference of time-of-flight for the electrons emitted with velocity components parallel and anti-parallel to the drift velocity. The electrons emitted with their velocity directed with a component parallel to Vd have a time-of-flight T2 that is shorter than Tg while the electrons emitted in the opposite direction have a time-of-flight T1 that is longer than Tg : T1;2 D Tg .1 ˙ Vd =Ve /

(5.3)

where Ve is the electron velocity, which depends on the initial kinetic energy. From Eq. 5.3 it follows immediately that the difference between the two times-of-flight T D T1 –T2 provides a measure of the drift velocity magnitude, Vd Vd D T Ve =.2Tg /

(5.4)

while their sum is twice the gyro time: T1 C T2 D 2Tg

(5.5)

Equation 5.4 is the basic equation of the time-of-flight technique. Since Tg D 2me =.eB/

(5.6)

86

E. Georgescu et al.

where me and e are the mass and the charge of the electron, the time-of-flight measurements allow B to be determined as well. The triangulation and time-of-flight techniques complement each other ideally. While triangulation naturally becomes increasingly inaccurate if the target moves further and further away, the time-of-flight technique becomes more accurate because, according to Eq. 5.4, T increases with increasing drift step, and thus becomes easier to measure. EDI consists of two gun-detector units (GDUs) and a controller unit. The GDUs are mounted on opposite sides of the spacecraft and have oppositely directed fields of view. The guns are capable of firing in any direction within more than a hemisphere to accommodate arbitrary magnetic and electric field directions. Similarly, the detectors can detect beams coming from any selectable direction within more than a hemisphere. Electron energies can be switched between 0.5 and 1.0 keV. In order to measure the electron times-of-flight, as well as to distinguish beam electrons from the background of ambient electrons, the electron beams are amplitudemodulated with a pseudo-noise (PN) code. To find the beam directions that will hit the detector, EDI sweeps each beam in the plane perpendicular to B at a fixed angular rate (typically 0:2ı =ms) until a signal has been acquired by the detector. Once signal has been acquired, the beams are swept back and forth to stay on target; this operational mode is called windshield-wiper (WW). As the times when the beams hit their detectors are neither synchronized with the telemetry nor equidistant, EDI does not have a fixed time-resolution.

5.2.2 Ambient Electron Monitoring When the electron beams are off, the EDI detectors allow for ambient electron measurements with very high sensitivity and time resolution, albeit at fixed energies of 0.5 or 1.0 keV. In this mode the two detectors which are facing opposite hemispheres are looking strictly into opposite directions. Either both detectors are looking in the plane perpendicular to B (pitch angle of 90ı ), or one detector is looking along B while the other is looking antiparallel to B (pitch angles of 0ı and 180ı ). When looking in the plane perpendicular to B the two directions within the plane are determined by the cross product of the magnetic field vector B with a selectable but fixed vector P that is constant in the spinning spacecraft frame of reference. When looking parallel/antiparallel to B the two detectors switch roles every half spin of the spacecraft as the tip of the B vector spins outside the field of view of one detector and into the field of view of the other detector. In nominal telemetry mode (NM) there are 16 counts per detector and per second. Each count is accumulated over approximately 16 ms (1/64 s). In Burst mode (BM1) there are 128 counts per detector and per second, and the accumulation time is approximately 8 ms (1/128 s). Counts are accumulated synchronously by the two detectors and each counts pair is stored in telemetry with its corresponding look direction, expressed in spinning spacecraft coordinates (azimuth and polar angle). As the two look directions are antiparallel only one set of angles needs to be transmitted. As the spacecraft is spinning the

5

EDI Data Products in the Cluster Active Archive

87

look directions need to be maintained in order to point at a fixed pitch angle. The look directions reported in telemetry correspond to those used at the center of the respective sample accumulation time window. Full resolution AED (Ambient Electron Data) starting October 2004 have been delivered for ingestion in the CAA. Before this time, the detectors’ configuration for ambient operations and the correction procedures were evolving continuously and hence an ambient electron data product is not available.

5.2.3 Operations and Limitations EDI operations are limited by low magnetic fields and/or high background electron fluxes. Low magnetic field requires high beam currents to overcome the beam divergence along large gyro orbits, and to get sufficient signal-to-background ratio. However large beam currents, in conjunction with the beam-modulation and – coding, lead to interference with the electric wave measurements by the WHISPER instrument. Moreover, the smaller the magnetic field gets, the higher the requirement for very precise on-board magnetometer calibrations. Last but not least, rapid timevariations in magnetic and/or electric fields, as well as large fluxes of background electrons can also cause loss of track. Because of the noted interference with the WHISPER measurements, EDI beam operations have been subject to a 6-orbit cycle, where high beam currents (up to 300 nA) were allowed in only two out of six orbits, while currents were limited to about 100 nA in another orbit and no beam operation was allowed outside 4 h of perigee in the remaining three orbits. The code repetition frequency and beam current is delivered to the wave instruments as CRF (Code Repetition Frequency) auxiliary data product to help them recognize and correct for this interference.

5.3 Data Processing Chain The data processing chain consists of onboard, ground and science processing. The onboard processing writes to telemetry the measured quantities and instrument parameters. A schematic representation of the data processing chain is given in Fig. 5.2. The telemetry data consists of three types: (1) housekeeping (HK TM); (2) science normal mode (NM TM); and (3) science burst mode (BM TM). They are input into the EDI ground-processing chain after having a DDS-header (Data Disposition System, described in [5]) attached during the ground-processing performed at ESOC. The ground processing software consists of Merged Science File (MSF) production and the Pick-Library. The first merges the TM-packets and orders them in time and the second extracts the quantities needed by the science processing as input. The output of the science processing consists of time series of physical quantities contained in the science data products, as discussed in detail in the next section.

88

E. Georgescu et al.

EDI Data Processing Chain and Output ONBOARD SW

OUTPUT

HK TM

GROUND SW

(ESOC)

SCIENCE OUTPUT

SW

RDM

wwconv (c) caa_editof (IDL)

NM TM

OUTPUT EGD

qstat (c)

QSTAT

EDI_PISO (IDL)

PP PPplus MPD

MSF BM TM MSF (GCDC) FGM Data PICK LIBRARY (IDL) AEC

SW

Software TBA for documention (white on green)

DATA

SurveyPlot (IDL)

C3-3h Survey Plots

AE (IDL)

AED

Data product TBA = to be archived (white on blue)

Fig. 5.2 EDI data processing chain and output

The CRF and CLIST (CaveatLIST) auxiliary products used for cross-calibration with other instruments are not represented. They are produced from the MSF-files using the Pick-Library and an IDL main program that is archived for documentation. The software, archived for documentation only, is represented with white text on green background and the data products with white on blue (violet).

5.4 Instrument Data Products As we have learned from the instrument description section EDI operates in two modes: (1) an active one called WW (windshield wiper), or, EF (electric field), where electrons emitted by the guns are detected; and (2) a passive one called ambient electron mode (AE), where the guns are not emitting. Whenever EDI operates in its active mode, the electric field and the drift velocities of the electrons are measured. The data are analyzed with two techniques: triangulation and timeof-flight. The technique which gives better results based on some error criteria produces the winners, the other the losers. All results are assigned a quality flag (good/caution/bad).

5

EDI Data Products in the Cluster Active Archive

89

In addition to the main data products, raw and auxiliary data are being archived.

Raw data consist of merged science files (MSF); they are obtained by merging the housekeeping and science (burst (BM) and normal (NM) mode) telemetry files. They are in binary format and are input for the Pick-Library that is used by all science processing programs. Auxiliary are either data used internally by the main data production software (QSTAT) or data dedicated to intercalibration (EGD, CRF) or time-interval lists of the EDI operation modes (CLIST). – QSTAT contains a table of 24 rows (for each hour of the day) and 6 columns (for each 10 min of an hour) with detection efficiencies. The detection efficiency is defined as the percentage of the number of spin periods with at least 4 successful measurements per spin to the total number of spin periods. – AEC (ambient electron correction file) are used to correct raw, ambient electron counts for the internal, theta/phi dependence of the individual detectors. Transformation of the corrected counts to a particle flux is a non-trivial matter, and has only been performed (to-date) on a case-by-case basis. – CLIST files contain a listing of time intervals with the two possible operation modes of EDI: WW and AE. – EGD parameters are measurement time tags with microsecond precision, timeof-flight (TOF) and its error (sigma TOF) in microseconds, and the label of the used detector unit (DU). Using Eq. 5.6 the magnetic field magnitude can be determined and compared with FGM or WHISPER measurements for crosscalibration. An example of use of the EDI CAA auxiliary data is given in [4], where the QSTAT and EGD datasets were used to check and improve the calibration of the flux-gate magnetometer (FGM).

Main data products contain: – For the WW mode: time series of the three components of the electron drift velocity and of the three components of the electric field in Cartesian GSE coordinate system (corrected for spacecraft motion) with different qualities and time resolution (PP, PPP, MPD) – For the AE mode: time series of ambient electron counts for the three pitch angles .0ı ; 90ıo ; 180ıo /, the detector look direction in the spacecraft frame and in GSE and the status for AE mode – Overview plot of the main parameters for both modes for every 3 h for one reference spacecraft (usually C3)

The data products, their acronyms, contents, qualities, resolutions, time ranges and formats are summarized in Table 5.1. For the time ranges, following abbreviations have been used: d: day, mo: month, h: hour. Note that they are differences between the CAA acronyms and the group of letters appearing in the file and parameter names (e.g. the D for data was left away for MPD and AED).

90

E. Georgescu et al.

Table 5.1 EDI data products in the CAA CAA Data type Nr acronym Format (mode) 1 PP CDF MAIN

CEF

(WW)

2

PPP

CDF

MAIN (WW)

3

MPD

CDF CEF

MAIN (WW)

4

AED

CDF CEF

MAIN (AE)

5

3hSPLOT

PNG

MAIN Graphics (AE + WW)

6

EGD

CEF

Auxiliary (WW)

7

QSTAT

ASCII

Auxiliary (WW)

8

CLIST

CEF

Auxiliary (WW/AE)

9

AEC

ASCII

Auxiliary (AE)

10

CRF

CEF

Auxiliary (WW)

11

MSF

Binary

RAW

Name, content (C), quality (Q), time resolution (R) Prime Parameter C: electric field, drift velocity, status and errors Q: winners, good/caution R: spin (4 s) Prime Parameter plus C: electric field, drift velocity, status, errors, drift step (see Table 5.2) Q: winners + losers good/caution/bad R: spin (4 s) Merged Parameter Data C: electric field, drift velocity, status and errors R: 1–4 s Q: winners, good/caution/bad Ambient Electron Data C: ambient electron counts at 0ı ; 90ı ; 180ı pitch angles R: full 3-hourly Survey plots C: MPD, AED and CLIST parameters for one spacecraft (usually C3) Electron Gyrotime Data C: Electron time-of-flight, error and GDU-flag R: full Quality STATistics C: EDI detection efficiency (beam tracking performance), R: 10 min Caveat List C: Time intervals of EDI operation modes (WW/AE) AE Correction files Used to correct raw, ambient electron counts for theta/phi Code Repetition Frequency C: code repetition frequency, beam current index R: telemetry (5:12 s) Merged Science Files C: Merged HK + NM + BM

Time range 1d

1d

1d

1d

3h

1d

1d

1 mo

1d

1d

1d

5

EDI Data Products in the Cluster Active Archive

91

Table 5.2 EDI PP (MPD) and PPP parameters for CDF and CEF files (WW operation mode) CEF fPPjMPg CDF- fPPjMPg PP-Plus file parameters (CDF) parameters (winners and losers) C1 CP EDI fPPjMPg (“winners” only) time tags Epoch Epoch Half interval Status Status Status V ed xyz gse V ed xyz gse V ed xyz gse E xyz gse E xyz gse E xyz gse Reduced chi sq Reduced chi sq Reduced chi sq Drift step mag Drift step mag error inertial Drift step azi error inertial Nbeam Status loser V ed xyz gse loser E xyz gse loser Reduced chi sq loser Drift step mag loser Drift step mag error inertial loser Drift step azi error inertial loser Nbeam loser

Table 5.2 gives an overview of the WW-mode main data parameters. The first two columns describe the contents of the PP and MP files for both CEF and CDF formats. The third column describes the contents of the PP-Plus files, which are only available in CDF format. The PP-Plus files are meant for experts who want to compare the results of both “winner” and “loser” methods. The differences between the CEF and CDF files for PP or MPD are in time parameters. The CEF files have “time tags” variable instead of “Epoch”. The MP data have an additional half-interval parameter to account for the mixed resolution. The exact nature and quality of the data is indicated in the seven EDI status bytes described in detail in [5]. Status[0] byte is the data quality flag as defined for all Cluster experiments: 0, 1, and 2 for bad, use with caution and good data, respectively. The others define the percentage of 1 keV beams used in entire spin (Status[1]), the percentage of the highest quality beams used in entire spin (Status[2]), the chosen “winner” method among Triangulation/Time-offlight/simultaneous-time-of-flight with the ambiguity flag of the choice (Status[3]), percentage of Triangulation outliers (Status[4]), and the magnitude error (as a percentage) and azimuthal error (in degrees) on the drift step in the spacecraft frame (Status[5] and Status[6], respectively). The following caveats must be considered:

The measured electron drift velocity cannot always be separated into the electric-field and magnetic gradient induced parts; in such cases the computed electric fields are subject to contamination by magnetic gradient effects.

92

E. Georgescu et al. Table 5.3 Ambient electron data parameters CEF variable name CDF variable name ( C# CP EDI AE) Epoch C# EDI AE time tags C1 CP EDI AE C1 CP EDI AE C1 CP EDI AE AE status C# Status AE theta CL# Theta index AE phi D1 CL# Phi index AE counts CL# GDU1 PA 90 counts GDU1 PA 90 AE counts CL# GDU2 PA 90 counts GDU2 PA 90 AE counts CL# GDU1 PA 180 counts GDU1 PA 180 AE counts CL# GDU2 PA 0 counts GDU2 PA 0 AE counts CL# GDU1 PA 0 counts GDU1 PA 0 AE counts CL# GDU2 PA 180 counts GDU2 PA 180 AE D1 GSE CL# D1 xyz gse

The data quality is variable, and it is therefore mandatory that the status bytes be consulted before interpreting the data. The number of samples in the averages is variable.

Note that all EDI electric field and drift velocities delivered to the CAA, regardless of product, are in the inertial frame. Table 5.3 shows the AED parameters. These are raw data counts and in order to transform them in relative particle fluxes the theta and phi dependence must firstly be removed by using the AEC files. The AEC files contain a list of correction factors for the phi and theta angles of the look directions. The 6 “count” variables are signed 2-byte integers. The fill value for these counts is -32,768. The theta variable is the same for both EDI detectors and contains an index between 0 and 138; each step corresponds to 0:703125ı. The phi variable contains an index between 0 and 127 and each step corresponds to 2:8125ı . The reported phi is the azimuth look direction for detector 1. The phi for detector 2 is opposite, i.e. phi2 D 64 – phi1, and then adjusted so that it is between 0 and 127 if necessary. Only 2 bits of the status byte are used. Bit 0 indicates the energy: 1 is used for 1 keV, and 0 is used for 0.5 keV. Bit 1 indicates sub mode: if it is set, the data is from BM1 data; otherwise, the data is from NM123 or BM2. AE D1 GSE CL# (or D1 xyz gse C# CP EDI AE) is the detector 1 look direction in Cartesian GSE coordinates, where #=1, 2, 3 is the spacecraft id. The ambient electron counts must be divided by 2 in NM (normal mode) because the NM accumulation interval is twice that of BM1. The following caveats must be considered:

The raw-data counts have to be corrected for the angular dependence of the efficiency. The status byte and the AEC-files must be used as described above. Ambient electron data are available since October 2004 on C1, C2 and C3, see CLIST files for time intervals. The EDI instrument on C2 is working only in ambient mode.

5

EDI Data Products in the Cluster Active Archive

93

5.5 Example of Science Application An example of using EDI electric field data in the inner magnetosphere is described in this section. The primary electric field science product delivered to the CAA by the EDI team is the variable time-resolution Merged Parameter Data, or, MPD, as indicated in Table 5.1. This product is found at the CAA after selecting “EDI” as the “Experiment”, and then “Science” as the data category. The MPD data is available in either CDF or CEF formats, which is indicated in the product name as either C1 CE EDI MP CDF or C1 CP EDI MP, respectively (using spacecraft 1, or, C1, as an example). As noted earlier, the MPD data has a variable time-resolution of 1, 2 or 4 s. This differs from the PP dataset (i.e., C1 CP EDI PP, which is located under the “CSDS” data category) time-resolution of 4 s (spin-resolution). Another important difference between the MPD and PP datasets is that while the poorest quality data (i.e., quality D “BAD” data) has been filtered out of the PP dataset, it remains in the MPD dataset. The “BAD” data exists in the MPD dataset because it is useful for the advanced user who might be studying data quality itself. Therefore, it is necessary for occasional users to apply their own quality filter on the MPD dataset if desired. This can be done using the variable “Status C1 CP EDI MP”, or, Status, for short. As mentioned in the previous chapter the 0th byte of the Status is the quality flag and 0 indicates “BAD” data. It is recommended that the CAA user filter out the “BAD” data. Figure 5.3 shows MPD data from Cluster 1 on 08-Apr-2004. Plotted in the left-hand column (Panels a–c) is electric field data, which has the variable name “E xyz gse C1 CP EDI MP” in the MPD dataset. From top to bottom are the x, y and z components in GSE coordinates (Ex ; Ey and Ez , respectively). The “BAD” data has been filtered out. The MPD dataset contains electric fields and drift velocities in the inertial frame (i.e., the spacecraft motion has been removed). This hour of data, from 06:45 to 07:45 UT, encompasses perigee and the crossing of the magnetic equator in the pre-midnight MLT range. The two large electric field structures with peak magnitudes approaching 25 mV/m at approximately 06:58 and 07:28 UT were identified in [6] as subauroral ion drift (SAID) channels. This identification was facilitated by looking at the data in a more useful coordinate system than GSE. The electric field measured by EDI is only the perpendicular component, and therefore is only a two dimensional vector in a coordinate system with one of its axes along the magnetic field. This type of coordinate system is known as a “magnetic field-aligned” system, or, MFA system. Using the magnetic field data from the FGM experiment and the Cluster position data from the AUX data files, it is possible to construct an MFA system. In the inner magnetosphere, the magnetic field, B, and the position vector, P, form roughly a meridional plane. The MFA system constructed for this example has B lying along the z-axis while the x-axis lies in this meridional plane and points outward. The y-axis completes the coordinate system and points generally eastward. At the magnetic equator, for instance, the z-axis points northward (aligned), the y-axis points eastward (azimuthal) and the x-axis points outward (radial).

94

E. Georgescu et al. EDI Electric Field in GSE

Ez [mV/m]

c

Er [mV/m]

0 –10 –20

e

20 10 0 –10

Ea [mV/m]

Ey [mV/m]

b

20 10

–20

f

20 10 0 –10 –20

UT: 06:45 07:00 MLT: 22:00 21:55 R: 4.46 4.30 MLAT: –10.78 –2.81 08-Apr-2004

EDI Electric Field in MFA

d

E|| [mV/m]

Ex [mV/m]

a

20 10 0 –10 –20 20 10 0 –10 –20 20 10 0 –10 –20

07:15 21:50 4.18 5.72

07:30 21:43 4.10 14.66

07:45 21:36 4.06 23.84

06:45 22:00 4.46 –10.78

07:00 21:55 4.30 –2.81

07:15 21:50 4.18 5.72

07:30 21:43 4.10 14.66

07:45 21:36 4.06 23.84

Fig. 5.3 EDI electric field data from Cluster 1 on 08 April 2004. Plotted in the left-hand column are the x, y and z components of the perpendicular electric field in GSE coordinates. This is Merged Parameter Data (MPD), available at the CAA. Ephemeris data are tabulated underneath the UT labels. Plotted in the right-hand column are these same data in an alternative, magnetic field-aligned (MFA) coordinate system. From top to bottom are the radial, azimuthal and parallel components

The right-hand column of Fig. 5.3 (Panels d–f) shows the electric field data transformed into this MFA system. From top to bottom are the radial .Er /, azimuthal .Ea / and parallel Ejj components. By definition, the parallel component is identically zero (Panel f). In the MFA system, the electric field is close to being purely radially outward (Panel a). This is a defining characteristic of magnetospheric SAID electric fields. When mapped to ionospheric heights, the SAID field is directed poleward. The SAID electric field orientation and morphology produces a westward drift channel extending from the southern to northern ionosphere at subauroral latitudes. Subsequent SAID observations by Cluster have been reported in [7].

5.6 Summary of EDI Contribution EDI contributes to the Cluster Active Archive with data (raw, auxiliary and main products) and documentation. The main data products contain:

Time series of the following physical quantities (with errors and data qualities): (a) E D electric field vector in GSE, (b) V D drift velocity vector in GSE,

5

EDI Data Products in the Cluster Active Archive

95

(c) N D high resolution ambient electron counts at 0ı , 90ı and 180ı pitch angles, at fixed energy (0.5 or 1.0 keV). The time resolution is between 1 and 4 s except for (c). Graphics: Survey plots of the physical quantities and operation modes for one reference spacecraft (usually C3) with 3 h time range.

The auxiliary data contain time series of:

QSTAT: EDI detection efficiencies (10 min resolution) EGD: electron time-of-flight (full resolution, not regularly spaced) CLIST: listing of instrument operation mode time intervals CRF: code repetition frequencies, beam current indices, 5 s resolution AEC: correction tables for the AE mode

The documentation consists of:

ICD – a complete description of the EDI contribution to the CAA Instrument Reference Papers and User Manual Program Scripts

EDI data production and delivery are conforming to the ICD and is an ongoing activity. Currently all data for 2001–2005 were delivered and ingested. The data for 2006–2009 are going to be delivered until mid 2010. Note that all data files in the CAA contain in their headers caveats with general information for the user regarding the mission, the observatory, the instrument and the dataset. These state for instance that EDI didn’t work on Cluster 4 or that AED are available starting October 2004 and that C2 works only in ambient mode since April 2004. Acknowledgments The authors wish to thank the CAA team for their efforts in building the archive.

References 1. http://caa.estec.esa.int/documents/ICD/CAA EDI ICD V13.pdf 2. Paschmann G. et al., The Electron Drift Instrument on Cluster: overview of first results, Annales Geophysicae (2001) 19: 1273–1288 3. Quinn, J. M., Cluster EDI convection measurements across the high-latitude plasma sheet boundary at midnight, Annales Geophysicae (2001) 19: 1669–1681 4. Georgescu E. et al., Use of EDI Time-of-Flight Data for FGM Calibration Check on CLUSTER, Proceedings of the Cluster and Double Star Symposium 5th Anniversary of Cluster in Space, ESTEC, Noordwijk, ESA SP-598, 2005 5. “Users Guide to the Cluster Science Data System”, DS–MPA–TN–0015, issue3, 2008, ftp://ftp.estec.esa.nl/pub/csds/task for/users guide/csds guide 3 0.pdf 6. Puhl-Quinn, P. A., H. Matsui, E. Mishin, C. Mouikis, L. Kistler, Y. Khotyaintsev, P. M. E. Decreau, and E. Lucek (2007), Cluster and DMSP observations of SAID electric fields, J. Geophys. Res., 112, A05219, doi:10.1029/2006JA012065 7. Mishin, E. V. and P. A. Puhl-Quinn (2007), SAID: Plasmaspheric short-circuit of substorm injections, Geophys. Res. Lett., Vol. 34, L24101, doi: 10.1029/2007LG031925

Chapter 6

The EFW Data in the CAA Y. Khotyaintsev, P.-A. Lindqvist, A. Eriksson, and M. Andr´e

Abstract The Electric Field and Waves (EFW) instrument measures the 2D electric field (in the spacecraft spin plane) and spacecraft potential with sampling rates, on some occasions, up to 36,000 samples/s. We present a summary of the CAA data products produced from the EFW measurements. We briefly describe the production pipeline from the raw data to final data products and discuss the most common caveats.

6.1 Introduction The Cluster Active Archive (CAA) was created to archive all data from the Cluster mission. Emphasis is on providing the scientific community with calibrated high resolution science data. This paper describes the CAA products from the EFW instrument. It gives some brief information on the instrument, followed by a description of all EFW parameters available in CAA. Some important things to keep in mind when using the data are given in Section 6.5. For those interested, there is also a section describing some of the processing details of the electric field data.

6.2 Measured Quantities, Coordinate Systems A detailed description of the EFW instrument can be found in Gustafsson et al. [5]. We first briefly describe the raw quantities measured by the instrument which correspond to Level 1 products in the CAA. The detector of the instrument consists of four spherical sensors numbered 1 to 4 deployed orthogonally on 44 m long wire booms in the spin plane of the spacecraft. The potential drop between Y. Khotyaintsev (), A. Eriksson, and M. Andr´e Swedish Institute of Space Physics, Uppsala, Sweden e-mail: [email protected] P.-A. Lindqvist Royal Institute of Technology, Stockholm, Sweden

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 6, c Springer Science+Business Media B.V. 2010

97

98

Y. Khotyaintsev et al.

two sensors, separated by 88 m tip-tip (62 m in case of P32, see Section 6.5.2) is measured to provide an electric field measurement (CAA quantities P12/P32 and P34). The probe difference signals are normally routed through 10 Hz low-pass filters if sampled at 25 s1 , and through 180 Hz low-pass filters when sampled at 450 s1 (see Section 6.5 for the exceptions). The potential difference between each sensor and the spacecraft (CAA quantities P1, P2, P3 and P4) is measured separately with a sampling frequency of 5 s1 after routing through low-pass filters with a cut-off frequency of 10 Hz. The unfiltered analog output signals from the spherical sensor preamplifiers are also provided to the wave instruments (STAFF, WHISPER and WBD) for analysis of high frequency wave phenomena. The EFW instrument measures the electric field only in the spacecraft spin plane, therefore a spin-plane oriented coordinate system is best suited for scientific studies involving the electric field. The ISR2 (Inverted Spin Reference) system, also known as DSI (Despun System Inverted), is such a system. The X and Y axes are in the spin plane, with X pointing as near sunward as possible and Y perpendicular to the sunward direction, positive towards dusk. The Z-axis is along the (negative) spacecraft spin axis, towards the north ecliptic. The coordinate system is called “Inverted” because the actual spin axis of Cluster is pointing towards the south ecliptic. The difference between ISR2 (DSI) and the GSE (Geocentric Solar Ecliptic) is a tilt of 2ı to 7ı of the Z-axis performed in order to avoid shading of the EFW probes by the spacecraft.

6.3 Science Data Products As already mentioned the raw EFW data correspond to the Level 1 products in the CAA. They are archived for reference purpose only. In this section we discuss the science data represented by Level 2 (full resolution) and Level 3 (4 s resolution) products in the CAA. For a complete list of the EFW products see Lindqvist and Khotyaintsev [6]. The spacecraft potential (P) is here defined as the average of all available probes, measured relative to the spacecraft. It relates linearly to true spacecraft potential Vsc with respect to the surrounding plasmas, approximately as Vsc D 1:23 P C 0:7 V [1]. We here keep the opposite sign in order to get a quantity approximately logarithmically co-varying with the plasma density (see Section 6.5). If all four probes are available, the average is done over all 4 probes. If only two or three probes are available, the average is done over 2 probes (P1 and P2, or P3 and P4). If only one probe is available, this quantity is the value of that probe. The number of probes used for P is given in the Parameter caveats. The electric field vector (Ex and Ey / in the spin plane is represented by the quantity E. For Level 2 data it is computed using 2, 3 or 4 probes. When 4 probes are available, the computation of E is straightforward. When 3 probes are available, the computation is also mathematically straightforward, but the electric field has larger errors since the measured components are not orthogonal. When only 2 probes

6

The EFW Data in the CAA

99

are available, only one component of the electric field is measured in the spinning system, and thus E will have a strong spin modulation (see Section 6.5.2). For Level 3 data E is computed from a least-squares fit of a sine wave to one probe pair (P12 or P34) over 4 s (approximately one spin). The least-squares fit is normally done on P34, if available, otherwise on P12. The result is two components of the electric field (Ex and Ey / and a measure of the standard deviation of the raw data points from a sine wave (Sigma). The details of production are described in the next section. EINERT is the electric field vector in the spin plane in the inertial reference frame. It is computed from E by subtracting the spacecraft motion induced electric field vSC xB, where vSC is velocity of the spacecraft with respect to the center of the Earth. EGSE is the full 3-dimensional electric field vector in the GSE coordinate system. It is computed from EINERT by first computing the third (non-measured axial) component of the electric field using the zero parallel electric field assumption, E B D 0, and then transforming the full 3D vector into the GSE coordinate system. These computations are only done when the magnetic field direction is more than 15ı away from the spin plane and jBZ j is larger than 2 nT. Otherwise the error in the computed electric field component becomes too large. The plasma convection velocity VGSE is computed from EGSE and the magnetic field B as VGSE D .EGSE B/=B2 . This computation relies on the availability of EGSE, which in turn depends on the orientation and magnitude of B.

6.4 Production of the Electric Field Data Here we describe the processing chain of the electric field data. At the initial stage of production we remove intervals with: bad data due to issues with electronics, probe saturations due to low plasma density, and non-optimal bias current settings. Usually, only a few minutes of data are removed from each orbit for these reasons. However, large data gaps may occasionally occur. If the spacecraft is in the solar wind we attempt to correct for the wakes usually present in the raw data (see Section 6.5.3). In the presence of a constant ambient electric field, the raw data signal (probe potential difference) is a sine wave where the amplitude and phase give the electric field magnitude and direction. A least-squares fit to the raw data of the form y D A C B sin .!t/ C C cos .!t/ C D sin .2!t/ C E cos .2!t/ C : : : is done once every 4 s .2=! 4 s is the spacecraft spin period), and gives the following output: The sine and cosine terms, B and C (the electric field in ISR2). The raw data DC offset, A. Ideally, the DC level of the raw data should be zero,

however small differences between the probe surfaces and in the electronics create a DC offset in the raw data. If not corrected, it shows up as a signal at the spin frequency in the despun electric field.

100

Y. Khotyaintsev et al.

The standard deviation of the raw data from the fitted sine wave. Higher order terms, D, E, . . . , may be used for diagnostics of data quality.

The electric field (computed from B and C ) and the standard deviation are available in the CAA as Level 3 E. The raw data DC offset (A) is used for processing the Level 2 data. It is applied to P12/P32 and P34 prior to despinning. Variations in the electric field cause small changes of the DC offset, and therefore the DC offset is smoothed using a weighted average over seven spins [6]. Intervals containing WHISPER pulses are blanked prior to despinning (see Section 6.5.6). Another offset which is applied during despinning is the so called Delta offset, which is the difference between the electric fields in ISR2 obtained from the least-squares fit procedure for individual probe pairs, P12/P32 and P34. This offset comes from slow evolution of the probe characteristics with time. A value averaged over several orbits is used. At this stage both the Level 2 (from despinning) and Level 3 (from least-squares fit) electric fields contain systematic errors, namely an amplitude factor and a DC offset, which need to be corrected for. The ambient electric field is “short-circuited” by the presence of the spacecraft and wire booms. This is caused by the spacecraft potential, which is also the potential of the wire booms, extending out to a large distance from the spacecraft. On the basis of simulations and comparisons with other Cluster instruments it has been determined that the measured electric field magnitude needs to be multiplied by a factor of 1.1 to get the real electric field (see also appendix 4 in Lindqvist and Khotyaintsev [6], Cully et al. [1]). This value is routinely applied for all spacecraft and for the entire mission. However, this factor depends on plasma environment, and in some cases may deviate from the value used. In case of Cluster it can vary in a range from 1.0 to 1:2. The spacecraft, wire booms and probes emit photoelectrons, which create an excess of negative charge on the sunward side. This will be measured by the EFW instrument as a spurious sunward electric field, generally referred to as the sunward DC offset. The offset varies slowly with time and plasma conditions around the spacecraft, and is slightly different for the different spacecraft. Its magnitude is determined by comparisons with other measurements, performed by Electron Drift Instrument (EDI) and Cluster Ion Spectrometry experiment (CIS). The values of the offsets that have been subtracted from the data in the CAA are given in the Parameter caveats. The photoelectron asymmetry responsible for the sunward DC offset by definition gives an offset in the sunward direction only. However, results of comparisons with other instruments have at times shown an offset of a fraction of mV/m also in the duskward direction, which is not yet well understood. In most cases the duskward offset is negligible. Prior to delivery to the CAA we attempt to identify and blank intervals with suspected spurious fields in various magnetospheric regions (see Sections 6.5.4 and 6.5.5).

6

The EFW Data in the CAA

101

6.5 Caveats Much effort has been spent on calibrating the data and removing spurious effects to give a useful database for scientific analysis. In spite of this, there will always be data in the CAA which are not of optimum quality, and this section attempts to describe some of the potential problems in the data the CAA user must be aware of.

6.5.1 10 Hz Filter Failure on C2 The 10 Hz filter on probe 3 on C2 failed on 25 July 2001. As a workaround for this, the 180 Hz filter has been used for the difference signals sampled at 25 s1 (sc nominal mode), which has no effect on the 4 s resolution data or on the burst mode data. There is a marginal effect on the 25 s1 data in those situations with large amplitude electric field fluctuations present between 10 and 180 Hz, when these fluctuations alias into frequencies below 10 Hz. After the filter failure, probe 3 on C2 is also not used for computation of the spacecraft potential.

6.5.2 Probe 1 Failure on C1, C2 and C4 Normally, the spin plane electric field is computed using the orthogonal signals P12 and P34. However, a failure occurred on probe 1 on C1 (28 December 2001), on C3 (29 July 2002), and on C2 (13 May 2007). After this, it is no longer possible to measure P12, but a workaround was implemented in the flight software to use P32 instead, the so-called asymmetric mode. This was fully implemented on 29 September 2003 on C1 and C3, and on 24 November 2007 on C2. In the intermediate period (Jan 2002–Sep 2003 for SC1, Aug 2002–Sep 2003 for SC3, and May–Nov 2007 for SC2), full resolution data are available only from one probe pair, which means that they contain a strong spin modulation, and must be used for overview purposes only. The 4-second resolution electric field data are not affected. Data produced from P32 and P34 also have a lower quality with a significant error signal around 0.25 and 0.75 Hz. Figure 6.1 shows examples of the electric fields obtained in the cases of: (1) orthogonal signals P12 and P34, (2) P34 only, (3) the asymmetric configuration.

6.5.3 Solar Wind Wakes The streaming solar wind creates a negatively charged wake behind the spacecraft, in the anti-sunward direction. As the individual probes enter and exit this wake, there is a dip in the probe potential, and thus a spike in the raw data signal twice per spin

102

Y. Khotyaintsev et al.

Fig. 6.1 Examples of spectra and time series (X and Y components in ISR2, circles show 4 s data) of electric field obtained in different instrument configurations on Cluster 1. Panels show (top) normal configuration – both p12 and p34 are present, (middle) p34 only and (bottom) asymmetric configuration – p32 and p34

6

The EFW Data in the CAA

103

Fig. 6.2 Electric field measurements in the solar wind by Cluster 1. The two upper panels show spectra of Ex before and after wake correction. The bottom panel shows raw (blue) and corrected (green) data from p12. Here and in the other figures P12 and P34 are given in units of electric field, i.e. the measured potential difference divided by the boom length

on each probe pair (see blue curve on the bottom panel in Fig. 6.2). In the despun electric field data this shows up as a negative spike in the sunward component, Ex, four times per spin. The resulting spectrum of the electric field (top panel in Fig. 6.2) has a strong signal at 4x the spin frequency (1 Hz) and multiples thereof. An algorithm has been developed to correct for the solar wind wake in the electric field data before submission to the CAA [4]. While doing a good job (results of the correction shown in middle and bottom panels in Fig. 6.2), the algorithm is not always perfect, so the problem with solar wind wake should be kept in mind as soon as spikes at four times per spin period (signal at 1 Hz) are encountered in the data.

6.5.4 Cold Ion Drift Wakes In the low-density plasma encountered for example in the tail lobes, above the polar caps and in the low latitude dayside magnetosphere the spacecraft potential is

Raw p34 [mV/m]

104

Y. Khotyaintsev et al.

5 0 −5 04:20:30

04:20:40

Ex [mV/m]

0 −5 −10

Sc pot [−V]

Ey [mV/m]

−15 8 6 4 2 0 −2 −10 −20

ASPOC ON Removed in the CAA

−30 04:00

04:05

04:10

04:15 14-Feb-2001

04:20

04:25

04:30

Fig. 6.3 Electric field measurements in the polar cap by Cluster 3. The middle panels show Ex and Ey measured by EFW (blue) and EDI (black). Large deviations in Ex are caused by a wake. The upper panel shows 4 spin periods of non-sinusoidal raw p34 data inside the wake region. The bottom panel shows the spacecraft potential. ASPOC is turned on at 04:24 UT

usually of the order of several tens of volts. A cold ion population flowing in such plasma environments will see effectively much thicker booms (several meters instead of a few mm), creating a large negatively charged wake behind the booms in the direction of flow [2, 3]. Figure 6.3 shows an example of such a wake, defined by a large deviation of Ex measured by the EFW and EDI. When ASPOC is operating, the spacecraft potential is kept at a much lower value and the problem of wakes due to cold ion drift is much less severe. In comparison to the solar wind wake the ion drift wake is broader and more diffuse. For a relatively small wake the raw data are non-sinusoidal (see top panel in Fig. 6.3). For large wake they can become sinusoidal again and look very similar to those created by a real ambient electric field. The CAA production software attempts to detect such ion wakes by looking at a combination of parameters, such as spacecraft potential, magnetic field direction, and the relation between different electric field components. For small magnetic field elevations (B direction is within 15ı from the spin plane) we check for the ratio between the components of E along the projection of B into the spin plane and perpendicular to it. For larger elevations we look at the ratio between the measured

6

The EFW Data in the CAA

105

(spin plane) and unmeasured (perpendicular to the spin plane) components of E? , where E? is computed under assumption of zero parallel electric field, E B D 0. Higher ratios indicate a higher probability of the wake being present in the data. At present there is no algorithm to correct the data, and the bad intervals are removed from Level 2 and Level 3 data. Since it is sometimes difficult to discern between these wakes and a real electric field, analysis of the electric field should be done with caution in regions with possible cold ion drifts.

6.5.5 Plasmasphere

Raw p34 [mV/m]

During comparisons of electric field measurements made by EFW and EDI in the inner magnetosphere, it was found that EFW sometimes measures a spurious field of the order of 1–2 mV/m, mostly in the sunward direction (Fig. 6.4, see also Figure 1c in Puhl-Quinn et al. [8]). The raw data signal is often non-

2 1 0 −1 −2

Ey [mV/m]

Ex [mV/m]

12:35:30 1

12:35:35

12:35:40

1 0 −1

1 0 −1

EDI EFW

Sc pot [−V]

corotation −0.5 −1 −1.5

Removed in the CAA

−2 12:20

12:40

13:00 29−Sep−2002

13:20

13:40

Fig. 6.4 Electric field measurements in the plasmasphere by Cluster 1. The middle panels show Ex and Ey in ISR2 measured by EFW (blue) and EDI (black) as well as the co-rotation electric field (red). The upper panel shows 4 spin periods of non-sinusoidal raw p34 data. The bottom panel shows the spacecraft potential

106

Y. Khotyaintsev et al.

sinusoidal (top panel). This problem is not quantitatively understood, but apparently depends on simultaneously acting phenomena: (a) the dense plasmaspheric plasma can sometimes be so high as to give a random electron current to the probes almost as large as the applied bias current, driving the probe away from its vacuum potential with respect to the plasma; (b) the wake created by the spacecraft makes the random current and the resulting error dependent on the spin phase and hence gives an apparent DC field in a non-spinning reference frame; and (c) small differences in probe surfaces and bias circuitry will make this effect slightly different between the probes, further complicating the result. A first principles error treatment is therefore complex. Nevertheless, an empirical algorithm has been developed to detect the bad data. The algorithm uses a comparison between the measured electric field and the expected field if the ambient plasma were to co-rotate with Earth, and is applied only in regions of high density as indicated by the spacecraft potential (Sc Pot > 1:5 V). The problematic intervals are removed in the CAA data. Users studying the electric field in the inner magnetosphere should be aware of this problem, since the detection of bad data is not always perfect.

6.5.6 WHISPER Operations The WHISPER instrument regularly (often at a repetition period of 52 or 104 s in the normal mode) emits waves of certain frequencies to detect resonances in the ambient plasma. These active soundings are seen in the EFW data (the vertical stripes in the spectra in Fig. 6.2 are for example cause by WHISPER pulses) and are removed automatically in the CAA data by testing for a “sounder active” flag in the telemetry. There are some residual effects after removing the active sounding periods, mainly seen as small spikes in the Level 3 potential data (for details see appendix 5 in Lindqvist and Khotyaintsev [6].

6.5.7 ASPOC Operations The ASPOC instrument attempts to keep the spacecraft potential at a low value, primarily to enable low-energy ion and electron measurements by the particle instruments. In the absence of ASPOC the spacecraft potential often reaches several tens of V in the low-density plasmas encountered by Cluster. With ASPOC operating, the spacecraft potential is brought down to the order of 5–8 V. An example of ASPOC turning on is shown in the bottom panel of Fig. 6.3. As already mentioned, a positive side-effect of this is that the electric field measurements are most often improved since the wake effects associated with large spacecraft potentials in the polar cap are drastically reduced. However, while ASPOC alleviates spacecraft cold plasma wakes, there are also cases when the EFW data are adversely affected by

6

The EFW Data in the CAA

107

ASPOC. First, with ASPOC on, the sunward offset in the magnetosphere is more variable than with ASPOC off. This is not yet understood. Second, with ASPOC on, the noise level sometimes increases significantly at low frequencies. As ASPOC never has been operational on all four s/c, a general advice is that when using EFW data from a s/c with ASPOC on, always look out for ASPOC-related effects on the EFW data by comparing to the measurements from a nearby Cluster s/c with ASPOC off. The spacecraft potential is often used as a proxy for ambient plasma density variations [7]. The data quantity in the CAA is the probe potential measured with respect to the spacecraft potential, which is most often a negative quantity. More negative probe potential corresponds to more positive spacecraft potential which also usually corresponds to lower ambient plasma density. Though the conversion of the spacecraft potential to density can be used at reduced accuracy also when ASPOC is on [9], it works best when ASPOC is off and the spacecraft potential is allowed to float freely. Any use of the spacecraft potential to determine plasma density must take into account whether ASPOC was operating or not. ASPOC is not operational on C1.

6.5.8 EDI Operations EDI measures the ambient electric field by emitting a beam of electrons and detecting the drift step as the electrons gyrate around the ambient magnetic field. The emitted beam current contributes to increasing the spacecraft potential, particularly in a low density plasma. During some periods in the beginning of the mission, the emitted EDI beam current was larger than expected, in particular on C2, so there is a tendency for the spacecraft potential to be larger than on the other spacecraft. It also affects the EFW electric field measurements in that there is a larger risk of saturating the measurements, which happens when the spacecraft potential reaches 70 V. However, after April, 2004, the EDI guns on C2 were turned off in favor of running the instrument in an alternative, “ambient” mode and they no longer cause problems.

6.6 Summary We presented a summary of EFW products in the CAA. We stress that a spin-plane oriented coordinate system (for example ISR2) is best suited for scientific studies involving the electric field. The uncertainty of the electric field measurement is generally less than the greater of 1 mV/m or 10% of the measured value. Accuracies better than 1 mV/m can be achieved by detailed analysis of calibration factors and errors in specific events. We have also listed the most important caveats a user of the electric field data must be aware of.

108

Y. Khotyaintsev et al.

References 1. Cully C. M., R. E. Ergun, and A. I. Eriksson, Electrostatic structure around spacecraft in tenuous plasmas, J. Geophys. Res. 112, A09211 (2007). doi:10.1029/2007JA012269 2. Engwall, E., A. I. Eriksson, and J. Forest, Wake formation behind positively charged spacecraft in flowing tenuous plasmas, Phys. Plasmas 13, 062904 (2006). doi: 10.1063/1.2199207 3. Eriksson, A. I., M. Andr´e, B. Klecker, H. Laakso, P.-A. Lindqvist, F. Mozer, G. Paschmann, A. Pedersen, J. Quinn, R. Torbert, K. Torkar, and H. Vaith, Electric field measurements on Cluster: comparing the double-probe and electron drift techniques, Ann. Geophys. 24, 275–289 (2006) 4. Eriksson, A. I., Y. Khotyaintsev, and P.-A. Lindqvist, Spacecraft wakes in the solar wind, In Proceedings of the 10th Spacecraft Charging Technology Conference (SCTC-10), (2007) URL: http://www.space.irfu.se/aie/publ/Eriksson2007b.pdf ˚ en, F. 5. Gustafsson, G., R. Bostr¨om, B. Holback, G. Holmgren, A. Lundgren, K. Stasiewicz, L. Ahl´ S. Mozer, D. Pankow, P. Harvey, P. Berg, R. Ulrich, A. Pedersen, R. Schmidt, A. Butler, A. W. C. Fransen, D. Klinge, M. Thomsen, C.-G. F¨althammar, P.-A. Lindqvist, S. Christenson, J. Holtet, B. Lybekk, T. A. Sten, P. Tanskanen, K. Lappalainen, and J. Wygant, The Electric Field and Wave Experiment for the Cluster Mission, Space Sci. Rev. 79, 137–156 (1997) 6. Lindqvist, P.-A. and Y. Khotyaintsev, Cluster Active Archive: Interface control document for EFW, ESA document CAA-EFW-ICD-0001 (2007), URL : http://caa.estec.esa.int/caa/ instr doc.xml 7. Pedersen, A., B. Lybekk, M. Andr´e, A. Eriksson, A. Masson, F. S. Mozer, P.-A. Lindqvist, P. M. E. D´ecr´eau, I. Dandouras, J.-A. Sauvaud, A. Fazakerley, M. Taylor, G. Paschmann, K. R. Svenes, K. Torkar, and E. Whipple, Electron density estimations derived from spacecraft potential measurements on Cluster in tenuous plasma regions, J. Geophys. Res. 113, A07S33 (2008). doi: 10.1029/2007JA012636 8. Puhl-Quinn, P. A., H. Matsui, V. K. Jordanova, Y. Khotyaintsev, and P.-A. Lindqvist, An effort to derive a convection electric field model in the inner-magnetosphere: Merging Cluster EDI and EFW data, J. Atmosph. Solar-Terr. Phys., 70, 564–573 (2007). doi: 10.1016/ j.jastp.2007.08.069 9. Torkar, K., A.I. Eriksson, P.-A. Lindqvist, and W. Steiger, Long-term study of active spacecraft potential control, IEEE Trans Plasma Science, 70, 2294–2300 (2008)

Chapter 7

FGM Data Products in the CAA J.M. Gloag, E.A. Lucek, L.-N. Alconcel, A. Balogh, P. Brown, C.M. Carr, C.N. Dunford, T. Oddy, and J. Soucek

Abstract Set out in this paper is a description of the magnetic field and supporting data products that the FGM team have generated and submitted to the Cluster Active Archive. A significant amount of effort has been put into the calibration and validation of the magnetic field data products and details of the calibration techniques used to produce these data are given here. The quality control tools used to validate the extensive magnetometer dataset and the quality indicators used to label the data itself (namely the caveat and accuracy) files are described. A summary of overall magnetometer calibration status, accuracy level and conclusions from the analysis work undertaken so far is also included.

7.1 FGM Instrument Description The FGM instrument consists of two three-axis fluxgate magnetometer (FGM) sensors located on a radial boom and an onboard data processing unit. The outboard sensor (OB) is located on the end of a 5.2 m boom and the inboard sensor is located 1.5 m in from the end of the boom. Each of the four Cluster spacecraft carries an identical FGM instrument and a full description of the hardware can be found in Balogh et al. [1, 2]. In flight either sensor can be designated as the primary sensor for acquiring the main data stream of the magnetic field vectors and since the J.M. Gloag, E.A. Lucek (), L.-N. Alconcel, P. Brown, C.M. Carr, C.N. Dunford, and T. Oddy Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London SW7 2BW, UK e-mail: [email protected] A. Balogh Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London SW7 2BW, UK and International Space Science Institute (ISSI), Hallerstrasse 6, 3012 Bern, Switzerland J. Soucek Department of Space Physics, Institute of Atmospheric Physics, Academy of Sciences of the Czech Republic, Prague, Czech Republic

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 7, c Springer Science+Business Media B.V. 2010

109

110

J.M. Gloag et al. Table 7.1 FGM operating ranges Range number Range (nT) 2 64 to C63:97 3 256 to C255:87 4 1;024 to C1;023:5 5 4;096 to C4;094 7 65;536 to C65;504

Resolution (nT) 7:8 103 3:1 103 0.125 0.5 8

beginning of nominal operations in 2001 the OB sensor has been set as primary on all four spacecraft. The instrument is designed to be highly failure-tolerant through a full redundancy of all its functions. The wide range of magnetic field values to be sampled across the different regions encountered in the Cluster orbits necessitates the use of instrument ranging in order to prevent saturation of the fluxgates during perigee pass while at the same time providing high resolution in the solar wind where the ambient field is typically a few nT. The sensors measure the three components of the field in seven ranges, although only four of these (numbered 2–5) are used on Cluster, with full scales and corresponding digital resolutions as shown in Table 7.1. Another range (range 7) was used only for ground testing. The instrument data processing unit automatically switches the ranges by continually monitoring each component of the measured magnetic field vector. There are two telemetry modes in common use with a rate derived from the instrument raw sample rate of 201.75 vectors/s. In normal mode the data are digitally filtered and decimated down to 22.416 vectors/s and in burst mode the data are filtered and decimated to 67.249 vectors/s. In both cases the digital filter is an FIR (Finite Impulse Response) filter derived from a Gaussian shaped impulse response. The fluxgate sensors are most sensitive in the range (0–10) Hz. The signal chain from the sensor acquisition to the telemetered data stream has two filter stages – an anti-aliasing analogue filter on the output of the raw sensor voltage and the DPU implemented digital filter(s) described above (See also Figures 5 and 6 in Balogh [2]). The following section of this paper introduces the challenges of calibrating the magnetometer data. After an introduction and overview of the in-flight calibration procedure there are sections which give an introduction and useful practical insights into the implementation of the different calibration analyses required to produce the highest quality FGM data. Section 7.3 introduces FGM data products which have been submitted to the CAA, sets out the data production procedure and finally describes the quality control procedure for the FGM data products.

7.2 The FGM Calibration Procedure for CAA Data Products The calibration problem for a fluxgate magnetometer can be most clearly expressed in terms of the calibration equation. This is shown in Eq. 7.1 to relate the magnetic

7

FGM Data Products in the CAA

111

field in a calibrated orthogonal coordinate system in nT to the magnetic field in the fluxgate sensor coordinate system in engineering units. 0

1 0 10 1 0 1 BS1 Bx G1 cos 1 G1 sin 1 cos 1 G1 sin 1 sin 1 O1 @ BS2 A D @ G2 cos 2 G2 sin 2 cos 2 G2 sin 2 sin 2 A @ By A C @ O2 A BS3 G3 cos 3 G3 sin 3 cos 3 G3 sin 3 sin 3 Bz O3 (7.1) There are three sets of physical calibration parameters in the calibration equation. There are the sensor aligned offsets O1 , O2 , O3 which alter the overall level of the magnetic field, the gains G1 , G2 , G3 which scale the field and the angles ™1 , ™2 , ™3 , ®1 , ®2 , ®3 which transform from the almost orthogonal sensor coordinate system to an exactly orthogonal coordinate system. The subscripts s1, s2, s3 refer to the sensor coordinate system and the subscripts x, y, z refer to an orthogonal coordinate system. The equation has been written so that the s1 direction is in the spin axis sensor direction which is consistent with FGM on Cluster and the x direction is in precisely the spin axis direction. The aim is to determine values for these 12 calibration parameters. However there is no single way to solve this calibration equation and in reality several different analysis methods are employed to gain as much information about the 12 calibration parameters as possible. The type of analysis that can be done depends on details of a particular spacecraft mission such as whether it is a spinning spacecraft or three-axis stabilised and what type of space plasma regions are encountered. In addition some of the ground calibration values are used as input which can provide a starting point for some of the analysis. The approach taken for the calibration of Cluster FGM for the CAA is to start each analysis period with a well defined starting point and work with the physical calibration parameters as much as possible when performing the calibration analyses in order to make the results at each stage as transparent as possible. What are meant here by physical calibration parameters are the sensor offsets gains and misalignment angles. Each of the four instrument ranges used during Cluster FGM flight operations have a different set of calibration parameters, which means that for any particular analysis period there are 48 calibration parameters (i.e. 4 ranges 12 parameters for each range) to be determined. As a general guide to the physical calibration parameters it can be noted that the sensor offsets tend to vary over the shortest timescales. It is possible to see very small changes in offsets over times scales as short as minutes or hours but the more significant changes tend to be seen over weeks or months. As the offsets only change the overall level of the magnetic field, they tend to be more important in the lower instrument ranges where the magnetic field is lower. The gains and misalignment angles can be considered together as generally they change by much smaller amounts over longer time scales of months. One notable exception to this is when the spacecraft go through eclipses and the sudden changes in temperature can cause gains and misalignment angles to change significantly over the period of minutes and hours. Both gains and angles tend to be more important at the higher end of ranges and in the higher ranges as they are both multiplying factors to the field.

112

J.M. Gloag et al.

7.2.1 Overview of the Calibration Procedure Figure 7.1 shows an overview of the CAA calibration procedure. The central calibration method is the Fourier analysis technique, based on Kepko et al. [4]. This technique is applicable on Cluster as the spacecraft are spin stabilised and it is used to calculate the bulk of the calibration parameters. The range 2 spin axis offset is required as an input to the Fourier analysis which means this needs to be determined first. During the solar wind season the range 2 spin axis offset can be determined using the solar wind analysis and during the tail season a linear interpolation is made for the correction to the spin axis to the next solar wind season. It is possible to use the EDI (Electron Drift Instrument, Paschmann et al. [7]) data to monitor the magnetic field magnitude and make sure that no large deviations from the interpolated spin axis offset occur during the tail season. After the Fourier analysis has been performed there are still a number of calibration parameters that have not been estimated. Due to a number of these parameters being incorrect at this stage it may be possible to observe jumps in the magnetic field at instrument range changes. It is possible to use the measurement of these jumps at range changes to refine a number of calibration parameters. Under very specific conditions (see Section 7.2.5) it is possible to apply an inter-spacecraft calibration analysis to the data in order to determine a further set of parameters but it is very rarely the case that the correct conditions exist. The standard calibration files produced to calibrate the data for submission to the CAA are valid for the period of an orbit from perigee to perigee. These calibration files have had input from the solar wind analysis, Fourier analysis and range change analysis and are produced for every orbit of data submitted to the CAA. A small number of files have also had the results of inter-spacecraft

Calibration in

Solar wind analysis

Spin axis offsets

Fourier analysis (Kepko et al., 1996)

EDI

Bulk of cal. set

Refinement files to compensate for short term variations in calibration

Tweaks to some cal.

Intercalibration on selected data periods

Tweaks to some cal.

Jumps at range changes

Fig. 7.1 A diagram illustrating an overview of the calibration procedure used to generate the calibration parameters that are applied to the FGM data products submitted to the CAA

7

FGM Data Products in the CAA

113

calibration analysis incorporated into the parameters. It is possible for some calibration parameters to change on shorter timescales than an orbit which results in small residual observable signals in the calibrated data. The final calibration analysis is designed to produce a refinement to the orbit calibration and take account of the short timescale changes in a particular set of calibration parameters. Due to the intensity of processing and validation required to produce these calibration refinement parameters, only a subset of prioritised data periods are being considered for this analysis, and to date no FGM data within the CAA have calibrated using this method. Therefore we do not describe it in detail here. In the following sections an outline of each of the other calibration techniques is given. These descriptions do not aim to rigorously define the basis of each technique but instead the aim is to give an insight into the challenges of implementing these analyses in practice.

7.2.2 The Solar Wind Analysis The determination of offsets for non-spinning axes is generally considered to be one of the more challenging calibration tasks. However if several hours of solar wind data are available with the physical properties described below then it is possible to obtain corrections to these type of offsets to very high accuracy. The general idea is to exploit a particular natural property of the solar wind as done by Hedgecock [3], for example. The solar wind contains short period fluctuations which are predominantly rotational in nature. These fluctuations are therefore observed as changes in field direction. This means that measurements made in any particular non-spinning axis being considered should have on average no correlation with the measured magnetic field magnitude. However if an offset exists on a non-spinning axis then measurements of the magnetic field on this axis will have a correlation with the magnetic field magnitude. The offset on this axis is then estimated by determining a field value that when removed from the magnetic field on the axis under consideration results in a zero correlation between this corrected field and the magnetic field magnitude. In reality compressive fluctuations are also observed in the solar wind to a certain extent and the key to applying this calibration method successfully and accurately is in the selection of data periods to be used in the calibration analysis. The implementation of the solar wind calibration analysis used in the calibration procedure for producing Cluster FGM data for the CAA was originally developed by FGM CoIs at UCLA (H.K. Schwarzl, K. Khurana, M. Kivelson, 2005, private communication) who have also collaborated with us on its implementation at Imperial College. The data selection scheme is applied to data for an analysis period which have been split into bins of a fixed width. Each bin is also overlapped by a fixed time period. The length of analysis period, bin widths and overlap periods are discussed in the next paragraph. The data in the non-spinning axis being considered for calibration first has an artificial positive offset applied to it and the correlation coefficient between this field and the magnetic field magnitude is calculated for each bin. The data in

114

J.M. Gloag et al.

the same axis then has the same artificial offset applied in the negative direction and the correlation coefficient calculated for each bin. In order for this method to work most effectively it is important that the artificial offset is greater than the offset correction being calculated. This means that there can be some trial and error involved in deciding on what value to give this artificial offset. The data selection parameters can then be calculated by taking the absolute value of the difference between the two correlation coefficients for each bin. The greater the value of the data selection parameter, the more Alfv´enic the fluctuations are in that particular data bin and the more suitable the data are for use in the solar wind calibration analysis. The data selection parameters are then put into descending order and the bins that fall in the top 75th, 80th, 85th, 90th and 95th percentile boundaries are taken forward as the five sets of bins to be used in the calibration analysis. The procedure is applied to data from each of the four spacecraft independently. The period of time used for a particular analysis is usually 2 weeks but can be 1 month if less solar wind data are available. The analysis is normally done for three sets of bin widths and overlaps and then all the calibration results considered together to decide what correction to the spin axis offset is most suitable. The following combinations of bin width and overlap are used: (3,600, 200 s); (900, 50 s); (200, 20 s). The solar wind analysis is an iterative process and the input data to the analysis is calibrated using calibration from an orbit which is immediately before the analysis period being considered. This means that the analysis data is in an orthogonalised coordinate system with one axis precisely in the spin axis direction. The spin axis offset corrections are therefore also in this orthogonalised coordinate system. The spin axis offset corrections for the first iteration are generated by taking the median value of the corrections determined in each of the five percentile groups of bins containing the top 75th, 80th, 85th, 90th and 95th percentile bins from the data selection analysis. The final offset correction for this iteration is then calculated by taking the mean of these five median values. The next iteration then repeats the process using data which has been modified using the offset correction from the present iteration. Usually a stable result is established in three to five iterations. As stated already, this final offset correction is in an orthogonalised coordinate system. However this value should then be applied as a correction to the input sensor offset so must be transformed into the sensor coordinate system in order to do this. It is this corrected sensor offset which is required as an input to the Fourier analysis calibration. The results generated using different data bin widths generally give slightly different results and occasionally, if there is particularly little data for instance, the longest bin widths may not give good results. In order to check that the best spin axis offset corrections have been chosen the data are visually validated at this stage. The four spacecraft data are plotted together with the chosen spin axis offset corrections applied. It is expected that in the solar wind the four spacecraft data will lie on top of each other. There is always a level of natural variation in each spacecraft data but on average and particularly when the data are quiet it should not be possible to observe a significant systematic difference in one or more spacecraft spin axis data. If this is observed the use of a different spin axis offset correction should be considered for the affected spacecraft.

7

FGM Data Products in the CAA

115

Once the standard solar wind calibration has been performed successfully it is possible to further refine the spin axis offset corrections by considering the four spacecraft data together. A set of equations can be written which relate differences in the spin axis field between every combination of spacecraft to the refinements in spin axis offsets for each spacecraft. Singular value decomposition can be used to solve this set of equations for the refinements to the spin axis offsets for each spacecraft. In practice it is found that this technique can only be used rarely because exceptionally quiet solar wind data are required. It is important to visually identify the systematic small differences between the four spacecraft field in smooth field regions and then confirm that the results of this refinement analysis are consistent with these differences. This ensures that the analysis is determining refinements to the spin axis offsets and is not simply producing a measure of fluctuations in the field between spacecraft.

7.2.3 The Fourier Analysis Technique The Fourier calibration analysis [4] is central to the calibration procedure and based on the fact that a number of calibration parameters, if incorrect can produce spin signals in the despun magnetic field data. If the calibration Eq. 7.1 is linearised, has the despin matrix applied to it and then the Fourier transform applied to it as described in Kepko et al. [4], it is found that a set of the calibration parameters are directly related to first and second harmonic spin signals in the despun data. Measurement of these signals in the data can therefore be used to calculate estimates to a set of calibration parameters. The first harmonic signal in the spin axis magnetic field can be used to estimate the spin axis elevation and azimuthal angles. The first harmonic signal in the spin plane magnetic field can be used to estimate the two spin plane offsets and two spin plane elevation angles. Finally the second harmonic signal in the spin plane magnetic field component can be used to estimate the ratio of the spin plane gains and the relative spin plane azimuthal angles. The total, final estimates of the calibration parameters may not be small but the results come from a linearised set of equations. Therefore the final set of calibration parameters are made by using an iterative approach where at each iteration the new estimates of the calibration parameters are applied to the data to be used in the next iteration. For each iteration the calibration parameters are estimated using least squares analysis applied to the equations that come from the manipulations made to the calibration equation described above. Data selection is an important step to ensure the best results are obtained. The data are split into approximately two minute bins and each instrument range is calibrated separately. Normal mode FGM data at 22 Hz time resolution is used. It is essential that the data bins are chopped to size so that their length is an exact integer number of the spacecraft spin periods. The best data bins for use in the analysis are those which have generally low levels of fluctuations, field which is reasonably constant over the bin width but that also have good signal to noise on the spin signals

116

J.M. Gloag et al.

that are to be removed. However it is also important that there is a good spread of magnetic field magnitude and direction over the full set of bins so a compromise on these conditions is required. Three tests are made on the data bins in order to choose the best ones for analysis. The first calculates the median power spectral density between 0.1 and 1 Hz and the data that exhibit the highest level of fluctuations are removed. The second test calculates the standard deviations for each of the remaining data bins and these statistics used to remove the data bins with the most change in field over the bin width. The third test uses the calculation of the signal to noise ratio of the first harmonic signal to find the intervals with the clearest spin signals. Each of these tests are carried out on the spin axis and spin plane data bins independently. Fourier analysis is applied to data from a full orbit defined from perigee to perigee. Data quality for calibration varies from orbit to orbit in terms of coverage and data characteristics. For instance if a large amount of data in a particular range is sheath data which contains high levels of fluctuations over a wide band of frequencies it can be difficult to measure the spin signal associated with certain calibration parameters. Some calibration parameters change over shorter time scales than others. Generally speaking, the angles and gains tend to change over a longer timescale than the offsets. Under normal circumstances offsets can change by small amounts over the period of an orbit but angles tend to be reasonably constant for many orbits. However during events such as eclipses when the temperature drops and then rises significantly all the calibration parameters can change on a timescale significantly shorter than an orbit. In these circumstances the analysis results is the determination of average calibration parameters over the period of the orbit and it is sometimes clear from the least squares fitting routine that there are multiple solutions which correspond to different parts of the orbit. The starting point for this Fourier analysis is to use a calibration file which has the range 2 spin axis offset predetermined using the solar wind analysis for instance and the range 3–5 spin axis offsets input from a previous orbit within 1 month. The spin plane offsets are set to zeros and the gains are set to the ground determined values. Finally the angles are set to 0ı or 90ı in such a way as to align the three sensor axes in an orthogonal set. The initial results are evaluated for calibration quality. The first check is to create a table of all the spin axis angles that have been determined for each range. These angles should only change relatively slowly through time which means that if large variations are seen on any particular orbits it may be the case that the data quality was not sufficient to obtain good results. This can often be the case in range 2 as there tend to be larger levels of fluctuations in this data and the angle errors don’t produce as big a signal in lower field magnitudes. In practise the best orbits for calibration within a month are identified and the spin axis angles used from these in the orbits where the data are not as good for calibration. Secondly a calibration report is automatically produced for each orbit and spacecraft containing information about the three data selection tests and plots of the least square results from the first iteration. It is also recorded here if there are not enough data intervals to perform the analysis on a particular range. Finally spectrograms are generated for the spin axis field and spin plane magnitude field

7

FGM Data Products in the CAA

117

using the calibration parameters calculated at this stage. All this information together can be used to decide what calibration parameters may need to be substituted into one orbit calibration from another orbit where the data were more suitable for performing the analysis. In our implementation of the Fourier analysis technique, a first analysis is performed on a months worth of orbits. The set of spin axis angles for all orbits and ranges are used to determine which orbits are the best for calibration purposes and spin axis angles substituted from these best orbits to those where the analysis was less successful. If any orbits have too low data coverage in a particular range then the full set of calibration for that range is substituted from an adjacent good orbit. The spectrograms are then rerun with the updated calibration and then these are compared to the original spectrograms to make sure an improvement has been made. There may be more than one possibility for updating certain calibration parameters. All possible improvements are tried and then the calibration that produces the best spectrogram is chosen to be the calibration for that orbit. If a particular problem still remains with the calibration for a certain orbit then the automatically produced calibration report described previously can be used to try and establish the problem with the orbit. One point to note is that there is a spacecraft generated field on spacecraft 1 which manifests as a small first and second harmonic spin signal in the spin axis field. This cannot be removed with this calibration analysis but it has been found that the major part of the signal is in the imaginary component of the first harmonic spin signal so the real part is used to estimate the spin axis angles successfully. It should also be noted that the spin axis angles can be more precisely determined in the higher ranges because the angles produce a bigger signal in higher magnetic fields and also because the data tends to be quieter in the higher ranges. This means that the angles from the higher ranges can be used as guidance for the lower ranges if the data in the lower ranges are not good quality for calibration. It is possible for the calibration to vary within an orbit. The calibration from this analysis is effectively an average calibration over the period of an orbit and the calibration refinement method is required to remove the residual spin tone caused by these short timescale variations of calibration parameters. Figure 7.2 presents example spectrograms for a complete orbit of spacecraft 2 data from February 2006. Figure 7.2a is the spectra prior to Fourier analysis and Fig. 7.2b is the spectra produced following the analysis. The red lines and white lines overlaid on the plots mark out demarcations of instrument range switches with the long middle period corresponding to range 2 data. Looking at the period immediately following the 02/08 time marker in Fig. 7.2a where the instrument range switches from 2 to 3 it can be seen that there is significant increase in power at 0.25 and 0.5 Hz indicating that the initial range 3 calibration is incorrect. Following the Fourier analysis this power burst is identified as unnatural and corrected by the re-calculated calibration parameters.

118

J.M. Gloag et al.

b

Dynamic spectrum, |Byz|(Spin plane) sensor coords − SC2 orbit 863 V01

Dynamic spectrum, |Byz|(Spin plane) sensor coords − SC2 orbit 863 V03 1

0.9

0

0.8

−1

0.7

−2

0.6

−3

0.5

−4

0.4

−5

0.3

−6

log(PSD) nT2 Hz−1 Frequency / Hz

Frequency / Hz

1

0.9

0

0.8

−1

0.7

−2

0.6

−3

0.5

−4

0.4

−5

0.3

log(PSD) nT2 Hz−1

a

−6

0.2

0.2

−7

0.1

−7 0.1

02/06

02/07

02/08

Time

02/06

02/07

02/08

Time

Fig. 7.2 Spectrograms of spin plane magnitude magnetic field data from the 6th, 7th and 8th of February 2006 which has been calibrated using parameters derived using the standard Fourier analysis procedure. The horizontal axis shows time in month and day, the vertical axis shows frequency in Hz and the colour scale is a measure of the power spectral density: (a) corresponds to the spectrogram prior to the analysis and (b) corresponds to the spectrogram achieved following this calibration step. The red and white overlaid lines represent the different instrument ranges used through time. The lowest level is range 2, the mid level range 3 and highest level range 4

7.2.4 The Range Change Analysis Following the range 2 spin axis offset determination and the application of the Fourier analysis method there are still a number of calibration parameters which may not be correct. If the natural fluctuations in the data are small enough it may be possible to see small jumps in the magnetic field across instrument range changes when the calibration at this stage is used. These jumps are an indication that some calibration parameters are incorrect as the data should be continuous across the range changes. A method was developed by FGM CoIs at UCLA (H.K. Schwarzl, K. Khurana, M. Kivelson, 2005, private communication) to use these jumps at range changes in order to correct a subset of calibration parameters. Through collaboration with these FGM CoIs at UCLA this range change calibration analysis has been implemented at Imperial College. The analysis is performed in the spin plane and the spin axis magnetic field separately. In the spin plane a jump in the field at a range change can be ascribed to a gain error and an error in azimuthal angle. In the spin axis a jump in the field at a range change can be accounted for by a gain error and an offset error. A set of equations can be derived which relate these calibration parameters to the jumps in the field observed at the range changes. In order to cause the smallest change to the overall field the gain corrections are applied to the lower range in both the spin axis and spin plane and the other parameters are applied to the upper range, the offset in the case of the spin axis and the azimuthal angle in the case of the spin plane. The implementation of this range change analysis involves creating analysis tools which are used to plot the data close to range changes, select which range changes are most suitable for the analysis and measure the jump in each component of the

7

FGM Data Products in the CAA

119

magnetic field. The period of data used for analysis is generally 2 weeks or 1 month, which usually has enough suitable range changes to give good results although occasionally slightly longer periods of time need to be considered. The calibration of changes between ranges 2 and 3 is usually the most difficult because the data often have a significant amount of natural signal compared with the size of the jumps being measured. Ideally the data either side of the range change should be smooth and the jump in the field to be measured should be clear. The data tend to be smoother close to range changes between ranges 3 and 4 which means that more jumps in the field can be used and generally it is possible to do the analysis with 2 weeks of data. The range 2 and 3 calibration is done first, the data reprocessed with the new calibration and then the range 3–4 calibration done. When all the ranges have been calibrated the data are reprocessed again and both the range 2–3 and 3–4 changes are compared with the original data to make sure that the tweaks to the calibration have been successful. This is an important step because when correcting the range 3 and 4 calibration small jumps may be reintroduced at the range 2 and 3 changes. One important point to note is that when doing the spin axis calibration it is important to make sure that the spin axis has a good range of magnetic field values. This is because the calculation of the calibration corrections involves a linear fit so the larger the range of field values, the more accurate the gradient and offset of this fit. An example of time series data from spacecraft 2 before and after the range change analysis is shown in Fig. 7.3. The upper plots show a range change from 4 to 5 for each vector component Bx, By and Bz. One can see there is a discontinuity where the range change occurs due to a calibration error. The lower plots display the same range change plotted following re-calibration of the time series using the parameter modifications determined during the analysis.

7.2.5 Inter-Spacecraft Calibration It is possible to use the four spacecraft data together to perform inter-spacecraft calibration and a method for doing this was developed by Khurana et al. [5]. The method demonstrates that if the assumption can be made for a period of data that the Curl of the magnetic field vector should be zero and it is taken that the divergence of the magnetic field is zero then it is possible to estimate a correction to a set of calibration parameters. This is done by minimising the Curl and divergence of the magnetic field in regions of magnetic field where the current density and therefore Curl of the magnetic field is expected to be zero. Using this technique it is possible to calculate a set of misorientation angles, a set of spin axis gain corrections and a set of spin axis offset corrections for three spacecraft relative to a pre-chosen mother spacecraft. This method has been implemented by FGM CoIs at UCLA (H.K. Schwarzl, K. Khurana, M. Kivelson, 2005, private communication) and a collaboration with these CoIs has lead to an implementation of this technique at Imperial College. Data selection is of prime importance when implementing this calibration technique and the initial task is to choose general data periods when the current density is

120

J.M. Gloag et al. Range 45 jumps before jump calibration −325

388

−26

−327

384

−27

−28

Bz /nT

By /nT

Bx /nT

−25

380

−329

18:50:00 18:50:30 UTC time

−331

18:49:40 18:49:50 UTC time

376

18:49:40 18:49:50 UTC time

Range 45 jumps after final jump calibration −325

388

−26

−327

384

−27

Bz /nT

Bx /nT

By /nT

−25

380

−329

−28

376

−331

18:50:00 18:50:30 UTC time

18:50:00 18:50:30 UTC time

18:50:00 18:50:30 UTC time

Fig. 7.3 A typical figure generated during the range change analysis. The data are from 30th of January 2009. The upper three plots correspond to a range 4–5 jump for each field component utilising the calibration file produced by the Fourier analysis step. The lower panel shows the same jumps after the corrected file has been further modified by the range change analysis. We can see on all three components that the jump offset has been reduced to almost zero and the data are now continuous across the range change

expected to be zero, for example, in the magneto-tail lobe regions. The lobe regions are located by finding the point where the magnetic field in the x-axis of the J2k coordinate system (Geocentric inertial reference frame using the Julian Epoch of Jan. 1, 2000) smoothly moves through zero and by considering the data either side of this point in the magneto-tail. Data is then selected where the Curl of the magnetic field is smooth, the magnetic field in the x-axis of the J2k coordinate system is smooth, the inter-spacecraft distances are less than 4,000 km and the Q-factor [8] for the spacecraft tetrahedral is greater than 2. In practice the most successful results between spacecraft have been obtained in the tail season of 2003 when the inter-spacecraft distances were of the order 100 km. It was also found that more stable results were found when only solving for the spin axis gain corrections and spin axis offset corrections. Generally between 2 weeks and 1 month of data are considered when performing the data selection for the inter-spacecraft calibration analysis. When the minimisation analysis has been performed checks are made to determine if good results have been obtained. The Curl and divergence of the magnetic field are calculated, which should be small, considered and the changes to the calibration parameters made after the last three or four iterations are checked to make sure a

7

FGM Data Products in the CAA

121

stable solution has been achieved. A visual check is made to make sure the four spacecraft spin axis field values are close together in the regions selected for the analysis. The inter-spacecraft calibration is rewarding in that it yields corrections to calibration parameters which could otherwise not be estimated.

7.3 The FGM Data Products 7.3.1 Primary Data Products The primary FGM data products consist of magnetic field data products at three different time resolutions. Each of the three data products are processed independently and are calibrated using the same calibration file. Each calibration file is valid for the period of one orbit defined from perigee to perigee. The highest resolution magnetic field data product (Cx CP FGM FULL) contains data which has no time averaging applied. If the instrument is in normal mode the data is at approximately 22 Hz resolution and if it is in burst mode the data is at approximately 67 Hz resolution. The second magnetic field data product (Cx CP FGM 5VPS) is averaged to a time resolution of 5 vectors/s and the third magnetic field data product (Cx CP FGM SPIN) is averaged over one spacecraft spin where the spin phase for averaging is the same as that used for the Prime Parameter data (26:367ı). Each of the three primary data products have the same format below the CEF-2 (Cluster Exchange Format, version 2) header which consists of the following columns of data: ISO time; half time interval/s; magnetic field vector (3 components)/nT; magnetic field magnitude/nT; spacecraft position (3 components)/km; instrument range; telemetry mode. The magnetic field components and the spacecraft position are in the GSE coordinate system. An example 5 lines of 5 vectors/s data directly below the CEF-2 header for the cluster 4 spacecraft is shown below. DATA_UNTIL = EOF 2003-02-01T07:00:00.100Z,0.1,9.070,-20.395,2.979,22.519,104287.6,63758.5,1468.6,2,22 2003-02-01T07:00:00.300Z,0.1,8.972,-20.451,2.665,22.491,104287.7,63758.4,1468.4,2,22 2003-02-01T07:00:00.500Z,0.1,8.819,-20.501,2.418,22.448,104287.8,63758.4,1468.2,2,22 2003-02-01T07:00:00.700Z,0.1,8.635,-20.526,2.290,22.386,104287.9,63758.4,1468.0,2,22 2003-02-01T07:00:00.900Z,0.1,8.426,-20.608,2.326,22.385,104288.0,63758.3,1467.8,2,22

$ $ $ $ $

7.3.2 Support Data Products There are several support data products produced for the FGM data which give additional information about the magnetic field data products or about the calibration used to generate the data. There are two data products which give information about data gaps in the FGM data. The first product (Cx CQ FGM GAPF) lists the data gaps that are introduced at the processing stage which may be due to times when no data were taken for instance. The second product (Cx CQ FGM VALF) lists time periods where invalid data have been removed at the data validation stage.

122

J.M. Gloag et al.

Caveat files (Cx CQ FGM CAVF) are produced for periods of data when the calibration is known to be changing on short timescales near eclipses and for other events when visible effects have been observed in the data. Calibration accuracy files (Cx CQ FGM CALA) contain detailed information on the accuracy of the magnetic field data products and can be used to select periods of data which would be particularly good for a certain type of analysis from an accuracy point of view.

7.3.3 Calibration Files The calibration files (Cx CC FGM CALF) contain the nine elements of the 3 3 calibration matrix and the three offsets for each instrument range which are used to transform the FGM data from uncalibrated data in the sensor coordinate system to calibrated data in an orthogonal coordinate system. Each of the calibration files are valid for the period of one orbit and are produced for each spacecraft for every orbit where magnetic field data products have been submitted to the CAA. There are several different calibration analyses that are used to produce these standard orbit calibration files including the use of inter-spacecraft comparisons and which methods are employed at a particular point in time depends on properties of the data that are available at that time.

7.3.4 The Data Production Procedure There are several steps involved with producing the FGM data products. The first and most time intensive step is to determine calibration parameters for every orbit for which data is submitted to the CAA. The procedure for producing these calibration parameters is described in the next section. The next step is to validate the data which is also a time intensive step and is discussed in detail below. The data are then processed from raw data into calibrated data in the GSE coordinate system. In order to do this processing the calibration files are applied to the data and other auxiliary data are used to transform the data into its final form. Finally the data products are submitted to the CAA. Further information on the production and quality control procedures along with further information about the data products can be found in the FGM ICD (Interface Control Document), the newest version of which can be found in the CAA, Lucek et al. [6].

7.3.5 Quality Control for FGM Data Products Submitted to the CAA The outcome of the calibration procedure is one calibration file for each spacecraft for each orbit. However as already outlined some calibration parameters are

7

FGM Data Products in the CAA

123

necessarily derived from periods of data much longer than an orbit, typically of 2 weeks or 1 month. Each calibration method is unable to track changes in calibration parameters that occur on a timescale shorter than the data interval used to estimate the calibration coefficients. If the changes in calibration parameters are large, then the effects will be visible in the final magnetic data product. Therefore FGM calibration analysis does not always result in perfect calibration. The accuracy of the calibration depends on a number of factors including: Data coverage – extensive data gaps can have a strong negative impact on the

calibration. Data characteristics – most corrections to the calibration parameters are small.

Therefore power within the natural signal can make it difficult to identify the correction required to the calibration matrix. Spread of B and jBj within the data – The majority of calibration techniques require measurements over a range of magnetic field magnitudes and directions. A narrow spread in either of these can adversely impact the quality of the calibration. The time over which the calibration is calculated depends on each of these three factors. Poor data coverage might, for example, suggest that using data over a longer period would be helpful. However, all the calibration methods assume that the underlying calibration parameters do not change over the period of data being analysed. If there is a change in calibration parameters during the interval of analysis, then since only a single estimate of the parameters is derived, some part of that period will be less well calibrated. The quality control system for the FGM data products has three parts to it. The first is the validation procedure which is applied to all periods of data which are submitted to the CAA. The validation procedure is based on visual inspection of the 5 vectors/s data product. Semi-automated software searches for possible spikes in the data and these spikes are then visually inspected to determine whether they are invalid data (and therefore removed) or a natural feature. High resolution data may be inspected if the 5 vectors/s data are inconclusive. The start and end times of the periods of invalid data to be removed are recorded using the validation tools and then the data processed to remove these intervals. At this point the validation gap files (VALF) are produced which contain the start and end times for data periods removed at the validation stage. The second aspect of quality control is in the form of caveat files (CAVF). CAVF files are generated for intervals of data when the calibration procedure is compromised, often for one of the reasons listed in the previous section. The signatures of poor calibration at this level are generally either a visible signal at the spacecraft spin frequency, or a spread of magnetic field values between the four spacecraft while they are in a region when we would expect all spacecraft to measure the same average magnetic field. The latter is most commonly seen in the solar wind.

124

J.M. Gloag et al.

Within a CAVF file for a particular spacecraft for a particular interval, the signature of poor calibration, and (where possible) the cause are given. Common reasons for caveats are: 1. Spacecraft eclipses, where the temperature change experienced by the instrument during short or particularly long eclipses can often lead to a small but significant change in calibration parameters during that orbit. This leads to a larger than normal residual spin tone in the data that can be identified at the validation stage. 2. Spacecraft manoeuvres can also cause changes in the calibration parameters, leading to a similar signature as in 1. 3. An unexplained spread in sensor spin axis magnetic field data exceeding 0.25 nT is seen between the four spacecraft. This signature can generally only be seen in the solar wind where the magnetic field magnitude is low, and the magnetic field data are very quiet. This can arise if the spin axis offset changes during the 2 weeks or 1 month over which the spin axis offset calibration method is applied. However, poor data coverage might also influence the accuracy with which the spin axis offset can be estimated. 4. Unexplained instances where higher than normal spin tone is observed. These occur occasionally. Finally the third aspect of the quality control are the Calibration Accuracy files (CALA). CALA files give information on the accuracy of the calibration used to process each data interval. Three different categories of calibration accuracy parameters are recorded for users of the data: The first category is a set of parameters that are derived from the solar wind calibration analysis. Recorded here are the median values of the differences in spin axis magnetic field between all combinations of the four Cluster spacecraft. Also recorded are the standard deviations associated with the differences, calculated for each of the spacecraft combinations. The values are calculated over the period of each solar wind analysis, which is either 2 weeks or a month. The idea behind using these parameters is that after the solar wind analysis has been performed independently for each spacecraft it is expected that the spin axis field measurements from the four spacecraft would lie on top of each other when the spacecraft are in the solar wind. The measured differences in spin axis magnetic field between each spacecraft combination therefore indicates how well this calibration has been performed in a particular region or may indicate if the calibration is changing within a certain period. An example of when you might be interested in these parameters is if a study is being done which uses inter-spacecraft comparisons. The median parameters would give an indication of the accuracy of these inter-spacecraft comparisons in low field regions. The second category of calibration parameters are connected with the calibration analysis performed at range changes. The principal used in this method is that the magnetic field should not change as a result of the instrument changing range. Only a subset of range changes can be used to measure this effect, because the power in the science signal needs to be low enough for the small jump in B associated with the error in calibration parameters to be observed. The parameters

7

FGM Data Products in the CAA

125

recorded in the files are the residual jump in the magnetic field magnitudes for each spacecraft at both range 2-to-3 and 3-to-4 changes after the calibration has been performed. Only the range changes that are selected in the calibration analysis are used here, as data close to other range changes may contain too much natural signal for it to be possible to measure the accuracy parameters. In addition the residual angle deflection in magnetic field after the calibration analysis has been performed is also recorded for range 2-to-3 and 3-to-4 changes. All of these parameters are an average value taken over the period of a month. These parameters are important if analysis is being performed on data in higher field regions where the fluxgate magnetometer is in a higher instrument range. The parameters give an indication of how successful the range change analysis has been in a particular region and therefore an idea of the accuracy of the magnetic field data in the higher ranges. A user of the data can identify times when range changes occur from the primary data products described in Section 7.3.1. As shown in the sample of data at the end of Section 7.3.1 the second to last column of data is the instrument range so when these numbers change from one sample to the next, a range change has occurred. The third category of calibration accuracy parameters are extracted from the Fourier analysis calibration. They comprise of hourly averaged values of the residual first harmonic signal (spin frequency) to noise ratio in the spin axis and spin plane and the residual second harmonic signal (twice spin frequency) to noise ratio in the spin plane after the calibration has been performed. The residual power spectral density after calibration are also recorded for the corresponding first and second harmonic signal to noise ratios. These parameters demonstrate how well the Fourier analysis calibration has been done for a particular orbit or whether the calibration is changing within an orbit. More generally, these calibration accuracy parameters give an indication of how well a particular orbit is calibrated. Factors which influence how well it is possible to calibrate a particular orbit include the data coverage in each instrument range and also the type and level of physical signal in the data which is often connected with the magnetospheric region being encountered at the time. As the Fourier analysis method is applied to every orbit, the hourly averaged calibration accuracy values described above demonstrate how suitable the data available in a particular orbit is for producing a good set of calibration parameters. In addition to these parameters a single score value is also recorded in the calibration accuracy files for each category. The score values are on a scale of 1–4 (1 being the best) and are based on the long-term distributions of each parameter. A score is assigned for a particular period of time by considering in which quartile of the long-term distribution the parameter falls for that period. For each category of parameters an average score value is determined for a period and rounded to the nearest score value. In this way a single score value for each category is determined for each period of time. This simple scoring system makes it possible to run a rough filter on a proposed set of data periods in order to extract which periods are most suitable for a particular type of analysis.

126

J.M. Gloag et al.

0 −90

180

−5 −10

10

By

Bx

0

0

0

10

−10

0

Bz

φ°

360

−10

10

|B|

θ°

90

8 6 01 00

01 10

01 20

01 30

01 40

2006 Feb 07 (038) hour min (UT)

Fig. 7.4 Time series plot showing a typical piece of FGM data in the solar wind submitted to the CAA. We can recognise this data set as well calibrated due to the fact the traces lie directly on top of each other in quiet regions together with the absence of spin tone. The four traces correspond to each of the spacecraft using the conventional Cluster colours (C1 Black, C2 Red, C3 Green and C4 Magenta)

An example of a validated FGM time series plot is shown in Fig. 7.4. The panels correspond to , ', Bx, By, Bz, and jBj with all four spacecraft traces displayed in the traditional Cluster colours. The data is taken from a solar wind interval from the same orbit (863) as that of the spectrograms used in Fig. 7.2 although only 47 min of data are plotted in order to show the fine structure within the time series. We can see that in regions where the background field is smooth the four traces lie on top of each other indicating that the data is well calibrated whereas during sudden field changes the spacecraft traces are not overlaid reflecting the fact that different spacecraft do not encounter the discontinuity at the same time.

7.4 Conclusion This paper gives an introduction to the FGM data products submitted to the CAA along with a description of how the data were produced and an outline of the quality control procedures used to ensure the integrity of the data. This is the first time that such a high quality of calibration has been applied across large multi-spacecraft magnetometer datasets and it has so far resulted in an exceptional magnetic field data set in the CAA. The routinely achieved accuracy level for well calibrated science data in instrument ranges 2, 3 and 4 is between 0.1–0.2 nT in regions where jBj < 200 nT and better than 0.4 nT in regions where jBj > 200 nT. Spin tone may be seen in the data

7

FGM Data Products in the CAA

127

on occasion. This usually indicates a small residual error in the calibration, typically due to calibration drifts occurring on timescales much shorter than that of a single orbit. The highest case of spin tone occurs during perigee pass during passage through maximum jBj, where it may reach up to 0.4 nT. Where it occurs elsewhere during an orbit it will generally be less than 0.1 nT. If an extreme amount of spin tone (i.e. greater than 1 nT) is observed within a period of interest (a given range), a caveat will be produced along with the data that forewarns the user as to the nature of the excessive spin tone. The presence of spin tone in an FGM time series inevitably means its spectrogram will show residual power at spin (0:25 Hz) and twice spin (0:5 Hz) frequencies. Generally speaking if a particular narrow band signal is seen at either of these frequencies, it is likely that it arises from small calibration errors and is not natural in origin. This derives from the fact that it is not possible to distinguish power between natural signal and calibration error at these frequencies. 2 4 p The instrument noise level above 1 Hz is of the order 1 10 nT =Hz.10pT= Hz/. This may be compared with a digital resolution of 7.8 pT in the highest resolution range 2. Highest resolution data in burst mode are primarily designed for use in time series analysis, for example discontinuity analyses at the bow shock or magnetopause. CAA users wishing to study waves in the frequency range above 10 Hz should note that FGM is a DC instrument, and that filters within the instrument will significantly attenuate wave power above this frequency. It is recommended that data from the STAFF instrument be used to analyse waves in this frequency range. The FGM team does not recommend the use of 67 Hz for spectrograms. If a user of the data at a later date is interested in attempting an improvement to a short period of data of particular interest then the only way to achieve this is to develop a good understanding of the in-flight calibration techniques which are used to produce the calibration files. The detailed sections set out in this paper on the different calibration analyses give a good starting point to gaining this calibration knowledge required. Although the details of specific algorithms will need to be looked up in other sources, the sections in this paper contain valuable practical knowledge on implementing the calibration procedures. Acknowledgments We would like to acknowledge the invaluable assistance of the FGM CoIs and FGM CoI teams who have contributed to the calibration effort. We would also like to thank M. Kivelson, K. Khurana and H.K. Schwarzl for their generosity in helping with the implementation at Imperial College of the calibration methods developed at UCLA.

References 1. Balogh A., M. W. Dunlop, S. W. H. Cowley, D. J. Southwood, J. G. Thomlinson and the Cluster magnetometer team: The Cluster magnetic field investigation, Space Sci. Rev., 79, 65, (1997) 2. Balogh A., C. M. Carr, M. H. Acuna, M. W. Dunlop, T. J. Beek, P. Brown, K.-H. Fornacon, E. Georgescu, K.-H. Glassmeier, J. Harris, G. Musmann, T. Oddy, and K. Schwingenschuh: The Cluster Magnetic Field Investigation: Overview of in-flight performance and initial results, Ann. Geophysicae., 19, 1207–1217 (2001)

128

J.M. Gloag et al.

3. Hedgecock, P. C.: A correlation technique for magnetometer zero level determination, Space Sci. Inst., 1, 83–90 (1975) 4. Kepko, E. L., K. K. Khurana, M. G. Kivelson, R. C. Elphic, and C. T. Russell: Accurate determination of magnetic field gradients from four point vector measurements: I. Use of natural constraints on vector data obtained from a single spinning spacecraft, IEEE Transactions on Magnetics, 32, 377–385 (1996) 5. Khurana, K. K., E. L. Kepko, M. G. Kivelson, and R. C. Elphic: Accurate determination of magnetic field gradients from four point vector measurements: II. Use of natural constraints on vector data obtained from four spinning spacecraft, IEEE Transactions on Magnetics, 32, 5193–5205 (1996) 6. Lucek, E. A., J. M. Gloag, A. Balogh, C. Carr, Cluster Active Archive: Interface control document for FGM, http://caa.estec.esa.int/documents/ICD/CAA FGM ICD V40.pdf (2008) 7. Paschmann, G., J.M. Quinn, R.B. Torbert, H. Vaith, C.E. McIlwain, G. Haerendel, O.H. Bauer, T. Bauer, W. Baumjohann, W. Filliius, M. F¨orster, S. Frey, S.S. Kerr, C.A. Kletzing, P. Puhl-Quinn, and E.C. Whipple, The electron drift instrument on Cluster: overview of first results, Ann. Geophysicae, 19, 1273, (2001) 8. Robert, P., A. Roux, C. C. Harvey, M. W. Dunlop, P. W. Daly, and K.-H. Glassmeier: Tetrahedron Geometric Factors. In: Paschmann G. and P. W. Daly (Eds.) Analysis Methods for Multi-Spacecraft Data, pp. 323–348, ISSI, Bern (1998)

Chapter 8

PEACE Data in the Cluster Active Archive A.N. Fazakerley, A.D. Lahiff, R.J. Wilson, I. Rozum, C. Anekallu, M. West, and H. Bacai

Abstract We describe the data products of the Cluster Plasma Electron and Current Experiment (PEACE) instruments which have been produced for the Cluster Active Archive. This article is intended to introduce the PEACE instruments, the measured data and the data products provided through the CAA. We aim to help the researcher to choose the data products most suited to their needs and to avoid inadvertent misuse of the data.

8.1 Introduction The Plasma Electron and Current Experiment (PEACE) measurements of the thermal electron plasma population enable characterisation of the local plasma – its density, bulk flow, electron temperature – and electron pitch angle data reveal whether the local region is magnetically connected to regions of electron acceleration or lies between magnetic mirrors. PEACE electron data reveals whether electron pitch angle or energy scattering processes are active, for example as waveparticle interactions occur. Good quality moments of the electron distribution may also be used to study the contribution of electrons to currents in key regions (e.g., magnetotail current sheet, auroral regions, reconnection sites). The four spacecraft capability allows determination of the three-dimensional motion of plasma boundaries (which may not always have a corresponding magnetic signature) and probing A.N. Fazakerley (), A.D. Lahiff, I. Rozum, C. Anekallu, and H. Bacai Mullard Space Science Laboratory, University College London, UK e-mail: [email protected] R.J. Wilson Mullard Space Science Laboratory, University College London, UK and Now at Los Alamos National Laboratory, Los Alamos, USA M. West Mullard Space Science Laboratory, University College London, UK and Now at Royal Observatory, Belgium

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 8, c Springer Science+Business Media B.V. 2010

129

130

A.N. Fazakerley et al.

of three dimensional structures such as Flux Transfer Events or the magnetotail plasmasheet, where electrons on neighbouring field lines may have had quite different recent histories. The dataset is key to studies of plasma boundary regions (magnetopause, bowshock, cusp etc.) and how mass, momentum and energy are transferred across them. Each Cluster spacecraft carries an identical PEACE instrument, consisting of two sensors, HEEA and LEEA, and a data processing unit, the DPU. Below, we briefly describe a PEACE instrument and how it works, after which we introduce the Cluster Active Archive PEACE data products. Readers may also be interested in the PEACE Science User Guide available at the CAA website, which complements this paper.

8.2 The Cluster PEACE Instruments 8.2.1 Instrument Description The Low Energy Electron Analyser (LEEA) is designed to specialise in coverage of the very lowest electron energies (<10 eV) but is also capable of covering the full energy range up to 26 keV. LEEA has a geometric factor appropriate for the higher electron fluxes usually found at lower energies (e.g., in the solar wind and magnetosheath). The High Energy Electron Analyser (HEEA) has a larger geometric factor, better suited to the outer magnetosphere, which extends the dynamic range of the instrument. Each sensor usually covers about 70% of the instrument energy range, and the sensors are used together to cover the full range every spacecraft spin. An Inter-Experiment Link (IEL) connects the HEEA sensor to the Digital Wave Processor (DWP) experiment providing detailed timing information about each arriving electron to the Particle Correlator. An IEL between the Fluxgate Magnetometer (FGM) and PEACE is used to share raw magnetic field data, for use in onboard pitch-angle selection. The sensors conform to the Cluster II specification (Fazakerley et al., manuscript in preparation, 2009) which is similar to the Cluster I specification [1], except that the sensor anodes were modified to delete the fine angular resolution capability and the MCP resistivity was altered, in both cases to improve performance under high electron fluxes. 8.2.1.1 The Electrostatic Analysers Each sensor consists of a “Top Hat” Electrostatic Analyser with an annular microchannel plate chevron pair (MCP) and segmented anode to provide position sensitive detection of arriving electrons, together with a supporting Sensor Electronics Unit. Figure 8.1 illustrates the operating principle of the PEACE electrostatic analyser. Each PEACE analyser is mounted with its field of view fan lying perpendicular

8

PEACE Data in the Cluster Active Archive

131

Fig. 8.1 Illustration of the principle of the PEACE Top Hat electrostatic analyser. The voltage applied to the analyser hemispheres diverts electrons (shown in blue) of a specific narrow band of energy and arriving within the acceptance angle ‰ through the analyser to the detector, while electrons of other energies (e.g., in red) strike one of the analyser hemispheres and are not detected. The semi-annular microchannel plate amplifies the signal of an electron reaching it, and the resulting charge cloud is detected by one of 12 segments of the anode beneath, giving information about the electron arrival direction ™. The Top Hat design is able to provide a focussed spot in the detector plane for a parallel electron beam arriving from anywhere in the aperture plane. At times when the aperture looks sunward, the majority of photons (black wiggly arrow) pass through the aperture and do not find their way to the detector

rather than tangential to the spacecraft surface, as shown in Fig. 8.2, in order to minimise the entry into the aperture of photoelectrons and secondary electrons from the local spacecraft surface. Consequently, each sensor has a 180ı field of view. The two sensors are mounted on opposite sides of the spacecraft thus their combined field of view is 360ı (though they may not always be sampling the same energy at the same time). In combination, the two sensors view the complete 4 solid angular range during each half rotation of the spacecraft. Figure 8.2 shows the direction from which electrons arrive to be counted on anodes 0 to 11.

8.2.1.2 The Data Accumulation Time Interval and the PEACE Spin The basic measurement made by PEACE is the number of counted electrons per accumulation time. The satellite spin period, Tspin , is subdivided into 1,024 equal parts, of duration Tacc , the accumulation time. For a nominal 4 s spin, Tacc 3:9 ms. The start of a PEACE spin occurs on reception of a rephased sun pulse. PEACE azimuth angles in the spinning frame are defined with 0ı as the LEEA look direction at that time. The rephase angle can be changed on command. It is used to minimise measurements of solar UV generated photoelectrons when the sensor apertures look sunwards. The rephase angle used during early operations was modified in August 2003 in order to better correspond to the CSDS spin definition.

132

A.N. Fazakerley et al. Zb Yb

30° Spin

LEEA

HEEA

10 11 8

9 Xb

7 6 5 4 3 2 1

0

Fig. 8.2 The physical deployment of the sensors on the spacecraft. The spacecraft body coordinate system is shown .Xb , Yb , Zb ). The sensor numbering with respect to the spin axis is indicated, e.g., zone 0 looks towards –Xb , and sees electrons with velocities along the spin axis direction, CXb . In orbit the spacecraft spin axes are maintained roughly anti-parallel to the GSE Z axis

8.2.1.3 Energy and Solid Angle Coverage, and Resolution The energy range of the instrument is nominally from 0.6 to 26,460 eV. The Sensors can each make measurements at 88 distinct non-zero levels. The first 16 energy levels are equally spaced linearly in the range from 0.6 to 9.5 eV. Levels 16 to 87 inclusive are equally spaced logarithmically over the rest of the range. In normal operation, an energy sweep, during which the energy is continuously varied from a high to a low level, over a time interval Tsweep , is repeated 16, 32 or 64 times per spin, depending on the selected sweep mode (LAR, MAR or HAR). During a sweep, the measured energy is reduced by one “step” from energy level n to level n–1 in sweep mode LAR or else by two “steps” from energy level n to level n–2 in sweep modes MAR and HAR, during each measurement period Tacc . A complete sweep also has a “flyback” component, during which the sweep HV is returned to the high value needed at the start of the next measurement sweep. Sweep characteristics such as the energy range coverage, flyback and sweep duration, are shown in Table 8.1. The energy sweeps, including the flybacks, are illustrated in Fig. 8.3. During the satellite spin period the satellite rotates through 360ı . The angle of rotation between two measurements at the same energy is called the azimuth angle resolution, and is given by 360ı Tsweep =Tspin . The azimuthal angle

8

PEACE Data in the Cluster Active Archive

133

Table 8.1 Properties of the Energy Sweep Modes: LAR, MAR and HAR. Note that MAR is the most frequently used energy sweep mode Energy Polar Azimuth Energy Polar Azimuth Flyback Energy resolution resolution resolution samples samples samples samples coverage LAR 1 step /bin 15ı 22:5ı 60 bin 12 bin 16 bin 4 bins 60 step ı MAR 2 step /bin 15 11:25ı 30 bin 12 bin 32 bin 2 bins 60 step HAR 2 step /bin 15ı 5:625ı 15 bin 12 bin 64 bin 1 bin 30 step

Fig. 8.3 The pattern of the energy sweeps in each of the three sweep modes (flybacks shaded)

resolutions vary with sweep mode as shown in Table 8.1; thus the sweep mode names LAR/MAR/HAR refer to Low/Medium/High Angular Resolution. The azimuthal acceptance angle (i.e., ‰ in Fig. 8.1) is 5ı for HEEA and 3ı for LEEA. The look direction rotates through only 0:35ı of azimuth angle during Tacc . A related useful concept is the “basic sector” which has duration 64 Tacc . Each spin contains 16 basic sectors. The basic sector contains 1 sweep in LAR, 2 sweeps in MAR or 4 sweeps in HAR sweep mode. The 180ı detector field of view is divided up into 12 equal parts of angular width 15ı as illustrated in Figs. 8.1 and 8.2. Measurements are made simultaneously in each of these 12 “polar zones”. In any sweep mode, a single PEACE sensor measures 11,520 values per spin (e.g., in MAR 30 energy 32 azimuth 12 polar) which we call 3DX.

134

A.N. Fazakerley et al.

8.2.2 How PEACE Science Data Is Generated and Transmitted The DPU is able to produce a full resolution three dimensional (3-D) data product (3DX) from one or both sensors. However, these are too large to be transmitted at a rate of one every spin. Therefore, to achieve good temporal coverage, it is necessary to create reduced resolution data products onboard (i.e., in the PEACE DPU) which can be transmitted within the available normal mode telemetry (NM) or burst mode telemetry (BM). Uncompressed 3DX data is sometimes transmitted, albeit at a slow rate, e.g., to support in-flight calibration or other instrument tests. The most compressed set of information is the “onboard moments sums dataset” (OMS), from which electron moments may be calculated after transmission to the ground. The next most compressed dataset is the “onboard selected pitch angle data” (PAD). Here, the reduction in data volume is achieved by sending the bare minimum of solid angle information needed to reconstruct the pitch angle distribution for a single gyrophase. A third compressed dataset (LER) focuses on the lowest energy part of the low energy sensor’s coverage. These data products have been sized to be available every spin for both NM and BM (though for LER on C2 the rate is sometimes slightly lower; see Section 8.4.2.1). The remaining transmitted science datasets are various forms of 3-D data. The most commonly available is 3DR (3DX compressed 8) which is used onboard as the basis for calculation of the onboard moment sums. During BM intervals, 3DXP (3DX compressed 2) is often used, though this is further compressed on spacecraft C1 and C3 by slightly reducing energy coverage (discussed in more detail below). The rate at which these products are transmitted using standard NM and BM is shown in Table 8.2. For most of the mission, spacecraft telemetry capacity which was not being used by out-of-commission instruments (CIS on C2 and EDI sensor on C4) has been made available to PEACE. The enhanced NM has allowed routine transmission of 3DR every spin on C2, and every 3 spins (or so) on C4. The enhanced BM enables the transmission of full energy range 3DXP data from both sensors every spin, on both C2 and C4.

Table 8.2 Rate of transmission of the main scientifically relevant data products. Table shows the number of spins needed to transmit a data product. NM1 and BM1 refer to enhanced data rates available on C2 since 25 November 2001 and on C4 since 21 March 2002. From July 2003 to June 2004, the C2 BM rate fell to that of C4 BM on some orbits (TM sharing with RAPID). Note that on C2, a slightly enhanced NM was initially available from 16 January 2001 Data NM1 NM1 BM1 BM1 OMS, PAD 1 1 1 1 3DR 40 C2: 1 C4: 3 1 (one sensor) Not sent 3DXP 1 (other sensor) 1(both sensors) 3DX (H&L) 315 C2: 9 C4: 31 5 3

8

PEACE Data in the Cluster Active Archive

135

8.2.3 PEACE Instrument Status At the time of writing (summer 2008) all four PEACE instruments are working well during their eighth year of science operations. The only hardware failure of any kind is the loss of information from the Cluster 3 PEACE HEEA anode 2 since August 22 2005, i.e., data is not available from one of the 12 anodes in one of the eight PEACE sensors. The consequences are a loss of accuracy in moments generated with data from that sensor and incomplete pitch angle distributions from that sensor when the magnetic field is within 30ı of the spin axis direction. Consequently, in-flight calibration of the remaining anodes is also more difficult.

8.3 The CAA PEACE Science Data Products This section describes the CAA science data products that we expect a typical scientist to be interested in working with. Many of the data files have a suffix H for HEEA or L for LEEA. The 3-D and 2-D data products have some elements in the energy matrix which do not relate to measured energy. These relate to the HV sweep flyback and are used in the files to allow proper representation of the timeline of events during each spin. The 3-D and 2-D data products contain the measured quantity (i.e., count rate, differential number flux, differential energy flux or phase space density) in a suitable array, and also contain a reference “background noise” level quantity, based on a rate of 1/8th count per accumulation bin time interval derived experimentally in flight. In the case of pitch angle data, it is essential to provide “background” data for every record, as the anode which corresponds to a given pitch angle may change frequently in a non-steady magnetic field, and different anodes have different calibrations. There is an additional background signature associated with the spin phase at which the aperture faces the Sun, but this is not included at the time of writing.

8.3.1 Three Dimensional Distribution Data A variety of 3-D data products are produced by PEACE, but only one, 3DR, is routinely produced. These differ only in the degree of data decimation, which is tailored to the telemetry rate available at the time of data collection in orbit (see Table 8.2). The 3-D data is provided in different combinations of energy, azimuth and polar bins for different sweep modes (except for 3DR). We use one file per data product to contain the commonly used MAR and HAR sweep mode data, and another file structure for the less frequently available LAR sweep mode data. The approach is a compromise between having many different types of file (e.g., one per sweep mode per 3-D data product) and a one-size-fits-all file structure which would require very large files, which are very sparsely populated.

136

A.N. Fazakerley et al.

8.3.1.1 Standard Reduced Resolution 3-D Data (3DR) As noted above, 3DR is compressed by a factor 8 relative to 3DX. Unlike the other 3-D data files, the 3DR files have a standard size independent of sweep mode (see Table 8.3). In all sweep modes, adjacent pairs of 3DX polar bins are added together. The energy/azimuth compression depends on sweep mode: consecutive groups of four energy bins are summed for LAR; consecutive groups of four azimuth bins are summed for HAR; pairs of energy bins and pairs of azimuth bins are summed for MAR. The “flyback” information is correspondingly merged. The rate of 3DR data transmission varies with spacecraft and telemetry mode (see Table 8.2). Occasionally, 3DR is transmitted for only one sensor, and thus there may not always be both 3DRL and 3DRH files. While 3DR is the default 3-D product, less compressed 3-D data is often sent during BM intervals. These 3DX or 3DXP data have been used in ground data processing to make corresponding 3DR data files, in order to give continuity of 3DR file coverage at the request of the CAA. When the source 3DXP file has reduced energy coverage (see below) the corresponding 3DR will have only 13 as opposed to all 15 energy bins.

Table 8.3 This table shows the organization of data in 3-D data files. Fill factors are never 100%, as space allocated to flybacks holds fill values. The 3DX spin data record is sized to hold both MAR and HAR mode data, using 32 MAR energy and flyback bins and 64 HAR azimuth bins, for each of the 12 polar zones; hence 3DX files are roughly half filled with data. The same principle applies to 3DXP and 3DXPA, although in these cases only six polar zones (each 30ı wide) are applicable, and for 3DXPA fewer azimuths are relevant as explained in the text. The LAR files are separate as (a) LAR is rarely used and (b) a file able to hold data from all sweep modes would have a fill factor of only 23.4% CAA data Sweep Flyback Energy Azimuth Polar Data Fill factor product mode bins bins bins bins bins/spin (%) Reduced energy/polar/azimuth resolution (/8) 3DR H/M/LAR 1 15 16 6 1,440 94 Full resolution 3DX MAR 2 30 32 12 11,520 47 3DX HAR 1 15 64 12 11,520 47 3DXLAR LAR 4 60 16 12 11,520 94 Reduced polar resolution (/2) 3DXP MAR 2 30 32 6 5,760 47 3DXP HAR 1 15 64 6 5,760 47 3DXPLAR LAR 4 60 16 6 5,760 94 Reduced azimuth coverage, and reduced polar resolution (/2) 3DXPA MAR 2 30 4 6 720 47 3DXPA HAR 1 15 8 6 720 47 3DXPALAR LAR 4 60 2 6 720 94

8

PEACE Data in the Cluster Active Archive

137

8.3.1.2 Flexible Reduced Polar Resolution 3-D Data (3DXP, 3DXPLAR) The 3DXP data are compressed from 3DX by adding pairs of polar bins together (see Table 8.3). While such data is well sized to be sent every spin from both sensors in BM on C2 and C4, it is too large to be sent every spin from one sensor only with 3DR from the other sensor, on C1 and C3. In this case, to maintain a spin rate data stream, some energy coverage is sacrificed and a 2/15th subset of the measured energy range is discarded from either the top or bottom of the range. Fill values replace the data in the CAA files. 8.3.1.3 Full Resolution 3-D Distributions (3DX, 3DXLAR) The CAA 3DX data files are used for full resolution 3-D data, and also (rarely) for transmitted “summed energy” data which has been compressed onboard by adding pairs of energy bins together. The use of separate files for 3DX and 3DXP data is more efficient in terms of file size than using 3DX for both. 8.3.1.4 Partial Azimuth Content 3-D Data (3DXPA, 3DXPALAR) The 3DXPA data are a subset of 3DX, produced by reducing angular coverage rather than resolution. The files contain only two basic sectors, 180ı apart and each spanning 22:5ı in azimuth, selected to contain the magnetic field direction as determined onboard. If these data were provided in the 3DX (or 3DXLAR) files, 7/8ths of the file would be empty, so it is more efficient to provide a dedicated alternative file. Such data is available relatively rarely. It is regarded as an enhanced form of PAD.

8.3.2 Pitch Angle Distribution Data The PEACE DPU produces “onboard selected pitch angle data” (PAD) designed to use the minimum possible telemetry (see Section 8.4.1.1). As a result, these data are inconveniently formatted from the point of view of a scientific user, and may not always capture the full pitch angle range. For scientific use, the PEACE team provides pitch angle data to the CAA in a family of PITCH data files, which are more userfriendly and which incorporate corrected (“rebinned”) pitch angle data. The data are given in scientific units and are in the spacecraft frame, rather than the plasma rest frame. Some characteristics of these PITCH data products are summarized in Table 8.4, and they are discussed in more detail below. 8.3.2.1 Standard Spin Rate Pitch Angle Data (PITCH SPIN) The recommended CAA pitch angle product for routine use is PITCH SPIN. The data are derived from PAD data (see Section 8.4.1.1) which has been rebinned

138

A.N. Fazakerley et al.

Table 8.4 This table illustrates how data is organized in the PITCH family of data files. PITCH SPIN has no sub-spin timing, and so no azimuth/flyback information. It uses data from both sensors and caters for all three sweep modes. The PITCH SPIN data record is sized to hold up to 44 energy bins, and 12 pitch angle bins. The fill factor depends on the sweep modes and energy coverage of both sensors. PITCH FULL also caters for all three sweep modes. The PITCH FULL data record is sized to hold up to 32 energy and flyback bins, and 12 pitch angle bins, for each of the two sensors, all of this for each of two azimuths per spin. PITCH 3DR, PITCH 3DX and PITCH 3DXLAR files are for single sensor data only, and have similar structure to 3DR, 3DX and 3DXLAR files except that 12 pitch angles replaces the polar zones. Pitch angle data from 3DXP and 3DXPA is provided in the PITCH 3DX files Flyback Energy Azimuth P.A. Data Max. fill CAA data product Sweep mode bins bins bins bins bins/spin factor (%) Pitch angle data from both sensors, merged in one record, for each spin (2-D) PITCH SPIN MAR–MAR 44 12 528 100 PITCH SPIN HAR–HAR 30 12 360 68 PITCH SPIN LAR–LAR 44 12 528 100 Pitch angle snapshots, both sensors, two two-sensor records, for each spin (2-D) PITCH FULL MAR–MAR 2 30 2 12 720 94 PITCH FULL HAR–HAR 2 15 2 12 360 47 PITCH FULL LAR–LAR 2 30 2 12 720 94 Pitch angle information, one sensor, determined for every energy-azimuth sweep (3-D) PITCH 3DR Any 1 15 16 12 3,072 94 PITCH 3DX MAR 2 30 32 12 11,520 47 PITCH 3DX HAR 1 15 64 12 11,520 47 PITCH 3DX LAR LAR 4 60 16 12 11,520 94 PITCH 3DXPA MAR 2 30 4 12 1,440 47 PITCH 3DXPA HAR 1 15 8 12 1,440 47 PITCH 3DXPA LAR LAR 4 60 2 12 1,440 94

using ground calibrated magnetic field data, using FGM team software and daily calibration files (not the CAA FGM data, at the time of writing). PITCH SPIN is usually available every spin, except during weekly MCP tests (see Section 8.4.2.2). PITCH SPIN contains merged data from both the HEEA and LEEA sensors. It is a two dimensional product, with twelve 15ı pitch angle bins (covering 0ı to 180ı pitch angle) and 44 energy bins (sufficient to cover the full instrument energy range). The energy bin boundaries can be defined on a spin by spin basis. This enables efficient packaging of all data collected in all three sweep modes. At energies and pitch angles measured by both HEEA and LEEA, the average of the measured values from the two sensors is used. The data is provided in spin records, with no sub-spin timing information. Full pitch angle coverage is only possible if the magnetic field are slowly varying compared to the spin period. This condition is not always met, and the user is encouraged to check for such cases by plotting the Status PADSelectionQuality parameters together with the PITCH SPIN data.

8

PEACE Data in the Cluster Active Archive

139

8.3.2.2 Full Time Resolution Pitch Angle Data (PITCH FULL) The PITCH FULL product is designed to provide the rebinned PAD data at sub-spin time resolution from both the HEEA and LEEA sensors. This can be useful in a rapidly changing plasma environment. In effect, the electron data is collected as two-dimensional pitch angle snapshots, of duration an energy sweep (or two sweeps in HAR mode), collected twice within each spin, as illustrated in Fig. 8.4. The PITCH FULL data for each sensor is organized by pitch angle and energy, and azimuth angle to provide sub-spin timing information. The data can be conceptually divided up into up to four energy regions (top, bottom, HEEA overlap and LEEA overlap). For any given snapshot, part of the pitch angle coverage will be provided by LEEA while the remaining part is provided by HEEA (see Fig. 8.4), so that pitch angle coverage will only be complete in the sensor energy overlap region. A PITCH FULL snapshot has, for each sensor, twelve 15ı pitch angle bins (covering 0ı to 180ı pitch angle) and 32 energy bins, though usually the energy-pitch angle matrix for a given sensor will only be partially filled with data, for the reasons given above. The data in a single snapshot record from one sensor has an azimuth value 180ı apart from that of the other sensor. There are two snapshot records per spin, at azimuths which are 180ı apart. The snapshot azimuth angles may vary from spin to spin as the magnetic field direction varies. The leading energy bins are flyback related and contain fill values, which are useful in establishing accurate timing within the spin. As with PITCH SPIN, the energy bin boundaries can be redefined in the metadata from spin to spin, enabling efficient packaging of all data collected in all three sweep modes. The use of 32 energy bins is sufficient to cover the full energy range used by a single sensor.

Fig. 8.4 Illustration of onboard selected pitch angle data collection. The figure shows which polar zones (shaded) are selected for transmission. These are selected only for the azimuths containing the magnetic field direction

140

A.N. Fazakerley et al.

8.3.2.3 Pitch Angle Data from 3-D Data (PITCH 3DR, PITCH 3DX, PITCH 3DXLAR, PITCH 3DXA, 3DXA LAR) The PITCH 3D products are expected to sometimes be of interest in rapidly varying plasma environments, and are provided as a complement to PITCH FULL. The advantage of PITCH 3D data is that they contain pitch angle data collected throughout a spin, in contrast to PITCH SPIN and PITCH FULL which provide only a 2-D cut through the distribution. Thus PITCH 3D products may show pitch angle information at specific times of interest, including times not sampled in PITCH FULL. Such data may sometimes be useful, even if only a restricted pitch angle range is available. Also, under conditions of rapidly varying magnetic field or plasma environment, it may be that only the PITCH 3D products contain the 0ı and 180ı pitch angle, as in such cases the onboard pitch angle selection may have been unsuccessful (so that PAD and hence PITCH SPIN and PITCH FULL, do not contain these pitch angles). The disadvantage of PITCH 3D are that a separate file is needed for each sensor and that if we use only one file to contain any possible type of 3D data it will always be large and usually only sparsely populated. In order to avoid providing many very large sparsely filled files, it was decided to provide a separate files for PITCH 3DR, PITCH 3DX and PITCH 3DXPA, and for the rarely used LAR mode, additional files for PITCH 3DXLAR and PITCH 3DXPALAR. PITCH 3DR is the smallest and is required most often as 3DR is the most commonly available 3D product. PITCH 3DX is required usually only for BM telemetry intervals. It is associated with 3DX and 3DXP data. PITCH 3DXPA is only needed in rare cases when 3DXPA is available. These PITCH 3D products will often have poorer pitch angle and energy resolution than PITCH SPIN and PITCH FULL. The energy resolution will be highest when the parent data is 3DXLAR or 3DXPALAR. The energy and pitch angle resolution are most reduced for PITCH 3DR.

8.3.3 Electron Moments Data The CAA provides the CSDS Prime Parameter data (see Section 8.3.3.2) which have some inherent limitations and the MOMENTS data (see Section 8.3.3.1) which offer the highest quality moments, though not always at spin rate. 8.3.3.1 Moments from 3-D Data (MOMENTS) The CAA “Moments” data are electron density, velocity vector, pressure tensor and heat flux vector. The vectors and tensors are provided in GSE coordinates. Velocity and temperature are also provided as components perpendicular and parallel to the magnetic field (where the magnetic field data is taken from CSDS FGM PP data, at present). In particular, T? and Tjj are determined by projecting the pressure tensor

8

PEACE Data in the Cluster Active Archive

141

along the magnetic field and dividing by the density and by Boltzmann’s constant. T? is the average of the two perpendicular components. Similarly, vjj D .v:b/b and v? D v .v:b/b. where v is the velocity vector and b is the magnetic field unit vector. Supporting information on data quality is provided (see Section 8.5). These data are produced on the ground (as opposed to onboard the spacecraft), using 3-D data and thus are available at time resolutions which are limited by the rate at which 3-D data can be transmitted (see Section 8.2.2). The data source used to generate the moments is always the best available resolution 3-D data. Choices on the best way to combine data from the two sensors can be made in the moments production software; in the first release the moments are made using LEEA data in the energy overlap region. The spin rate C2 electron moments can be a useful substitute for ion moments that are lacking due to the non-availability of C2 CIS data. The advantage of these ground-calculated moments is that the best available calibration data is used, a correction can be made for the spacecraft potential, and spacecraft electrons can be properly removed. The current release has been generated automatically using default values to control the approach to potential correction and photoelectron removal, and has not been systematically humanvalidated, so some errors may exist. Specifically, the spacecraft potential is set to be 1 V greater than the measured probe-spacecraft potential provided by the electric field experiment (EFW) and the lower energy cutoff for the moments integration is set to reject the energy bin containing the EFW potential and the one above that (which in practice may contain counts due to spacecraft electrons). A further release is likely, pending improved calibrations. Moments production software will be provided by the PEACE team, via the CAA, to allow scientists to calculate their own moments, if they wish, for example using a subset of the measured energies and polar zones. 8.3.3.2 Prime Parameters The Cluster Science Data System (CSDS) Prime Parameter (PP) data are provided by the CAA. PEACE PP data are produced at MSSL and sent to the UK CSDS data centre, from where the CAA collects them. The data are derived from the onboard moment sums (see Section 8.4.1.2), which are converted to electron moments data (density, velocity, perpendicular and parallel temperature, parallel heat flux). The PP data are provided in the GSE coordinate system. Geometric factor updates based on in-flight calibration are used to correct for errors due to the less up-to-date onboard calibrations. However, difficulties with onboard inter-anode calibration cannot be corrected during ground data processing, which may lead to errors in the spin axis component of velocity. It is not possible within the PEACE DPU, or late on the ground) to make corrections for the acceleration of plasma electrons due to the spacecraft-plasma potential. Fortunately, this does not introduce significant error in all plasma environments, only those such as the solar wind where the energy gained by electrons due to the spacecraft potential is not negligible compared to the mean measured energy of the

142

A.N. Fazakerley et al.

distribution. The data are not corrected for possible interference from WHISPER, EDI or EFW. It is recommended that the CAA Moments data (and not the PP data) are preferred for science analysis, despite the reduced time resolution in some cases, and that use of the PP data is only made after careful consideration of their limitations.

8.4 Other Data Products 8.4.1 Pre-science Data Here we very briefly discuss the data products produced onboard which are the basis for scientifically useful products discussed above. More detail is available in the Interface Control Document [2].

8.4.1.1 Onboard Selected Pitch Angle Data (PAD) The routinely transmitted, “onboard selected pitch angle” product is provided to the CAA as PADHAR, PADLAR and PADMAR files. PAD data is sent routinely in both NM and BM intervals, with few exceptions. The magnetic field direction estimate used to select these data onboard is usually based on a single vector extracted at the start of each spin from the data stream provided by the Inter-Experiment Link from FGM. The vector is applied after being corrected onboard using uplinked values of the estimated magnetometer offset calibrations. On rare occasions this onboard data link gave corrupted data; for Cluster 2 this occurred in 2002 between 05–30 April and between 02–24 May. The data are organized as indicated in Table 8.5. The selected azimuths look parallel and anti-parallel to the (onboard determined) magnetic field. The selected polar zones are from these azimuths, and are chosen as shown in Fig. 8.4. The PAD data product usually contains both HEEA and LEEA data. From 2007, in cases when HEEA is turned off, LEEA data may be returned for the key azimuths from all polar zones, each half spin.

Table 8.5 This table illustrates the way that the data is organized in the PAD data files. In LAR mode, only 50% of the energy bins and flyback bins are transmitted each spin; these are the even numbered bins on one spin and the odd numbered bins on the next. For the special case of LEEA only PAD, used sometimes from 2007, all 24 polar bins in a spin contain data Polar bins DPU data Sweep Flyback Energy Azimuth Polar bins second product mode bins bins bins first half spin half spin PAD MAR 2 30 2 n 13-n PAD HAR 1 15 4 n 13-n PAD LAR 2 30 2 n 13-n

8

PEACE Data in the Cluster Active Archive

143

8.4.1.2 Onboard Moments Sums (OMS) Onboard moment sums data are sent routinely in both Normal Mode and Burst Mode spacecraft telemetry operations. Data that will be used to calculate the parameters mentioned in Section 8.3.3.2 are calculated onboard from 3DR data, for the Top and Bottom energy regions once per spin, and the Overlap regions for each sensor, twice a spin. Data from energies below 10 eV are excluded, on the assumption that they contain spacecraft electrons. In practice, spacecraft electrons may be measured above 10 eV. Then, the partial density in the Bottom region and sometimes in the Overlap regions will be overestimated. Users may check for this by viewing energy spectra.

8.4.2 Diagnostic Data These data may not interest the typical scientist, but are provided to the CAA for completeness. 8.4.2.1 LER The LER data product contains data from only one sensor, usually LEEA. It is a reduced 3-D distribution designed to enable estimation of spacecraft-plasma potential using PEACE spectra (provided the potential lies within the LER energy range). LER contains only the lowest 16 LAR bins, or 8 MAR/HAR bins. Polar data is summed in sets of four bins, so that three LER polar bins cover 180ı . Only a subset of evenly spaced azimuths are transmitted each spin. Neighbouring azimuths are sent on successive spins, in a cycle that has eight steps before all azimuths have been transmitted in MAR and HAR mode, or four steps in LAR mode. LER data is usually sent routinely in both NM and BM, though on C2, from 25 November 2001, LER is sent less often, ensuring spin rate 3DR. The LER label refers to fact that this data product contains only the “low energy, reduced” part of the distribution. 8.4.2.2 NOI The NOI data product is used during routine weekly MCP operational level tests, to capture anode-by-anode responses to changing MCP voltages. It is transmitted in place of PAD data. NOI always contains both HEEA and LEEA data (LEEA appear first). All energy bins are summed together. Azimuth bins are summed as for 3DR to provide 16 equal azimuth sectors. The “NOI” label is an abbreviation of “noise”, since this product was first used during ground testing, to monitor MCP count rates vs. anode.

144

A.N. Fazakerley et al.

8.5 Caveats and Status Information Our approach to assisting the user in avoiding pitfalls when working with PEACE data centres on provision of detailed spin by spin status information. Examples include warnings of saturation of the HEEA sensor in strong fluxes (leading to under-counting) and warnings of occasions when the PAD selection has erred (e.g., in rapidly changing magnetic fields) which should aid interpretation of PITCH

products. To aid interpretation of moments data, warnings about incomplete energy coverage of the measured distribution, or suspected coverage of more than one plasma population are provided. Full details are given in the Interface Control Document [2], available at the CAA website.

8.6 Summary Our article has briefly introduced the PEACE instruments, the measured data and the data products provided through the CAA. We hope that together with the Status information in the CAA files, and documentation at the CAA website, this paper will help the researcher to understand what the PEACE data offers, and how to work effectively with PEACE data. In addition to moments software, the PEACE team is providing a simple tool (plotcf) via CAA which facilitates quick plotting of PEACE and other CAA data. Acknowledgments The authors thank ESA and STFC for funding, and also gratefully acknowledge the excellent work of the engineers and scientists who conceived, built and are operating the PEACE instruments.

References 1. Johnstone A.D., C. Alsop, S. Burge, P.J. Carter, A.J. Coates, A.J. Coker, A.N. Fazakerley, M. Grande, R.A. Gowen, C. Gurgiolo, B.K. Hancock, B. Narheim, A. Preece, P.H. Sheather, J.D. Winningham, and R.D. Woodliffe, Peace: A Plasma Electron and Current Experiment, Space Sci. Rev. 79, pp. 351–398, 1997. 2. Fazakerley A.N, A.D. Lahiff, R.J. Wilson, M.G.G.T. Taylor, CAA Interface Control Document for PEA, CAA-PEA-ICD-0001, v5, 2008 (http://caa.estec.esa.int/caa/documentation.xml).

Chapter 9

RAPID Products at the Cluster Active Archive Patrick W. Daly and Elena A. Kronberg

Abstract The RAPID instrument on board the four Cluster spacecraft delivers 3-D distributions of energetic (&30 keV) ion and electron flux for up to eight energy channels. To make this data set available to the general scientific community, both during the operations phase and beyond the end of the mission, the RAPID Team is delivering processed data to the Cluster Active Archive (CAA). These data consist of both fluxes generated with the best current calibrations available, as well as raw count rates for knowledgeable users. Diagnostic products, caveats, documentation, and summary plots are also part of the deliverables. The intention is that once the Cluster Mission is completed and the instrument teams disbanded, this will be the sole repository of all available RAPID data and knowledge about it. This article gives an overview of the RAPID data products at CAA, and is intended to summarize the more complete description given in the Interface Control Document.

9.1 Introduction The RAPID experiment (Research with Adaptive Particle Imaging Detectors) on board Cluster measures 3-D energetic electron and ion fluxes in the energy range above 30 keV. Since the raw data delivered by the instrument are of no use without detailed knowledge of their construction, functioning, calibrations, and limitations, it is essential that the RAPID Team process this initial data set into usable, physical parameters, and make them available to the scientific community. This has been done for a restricted subset of the data by means of the Cluster Science Data System (CSDS at http://sci2.estec.esa.nl/cluster/csds/csds.html). Also, the RAPID community has had access through the Principal Investigator to the entire data set in a special format developed only for this purpose. With the Cluster Active Archive (CAA), a mechanism has been created to make the entire data set from all

P.W. Daly () and E.A. Kronberg Max Planck Institute for Solar System Research, 37191 Katlenburg-Lindau, Germany e-mail: [email protected]; [email protected]

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 9, c Springer Science+Business Media B.V. 2010

145

146

P.W. Daly and E.A. Kronberg

11 instruments available to all users in a standardized format, known as Cluster Exchange Format (CEF) [1]. Furthermore, a uniform set of metadata [4] has been defined, based on that of CSDS but further extended. The metadata allow data sets to be “self-describing”, essential for analysis and plotting tools that process them. We are in the process of delivering all RAPID data to CAA, both in the form of physical parameters (particle flux) and raw data (count rates) as well as diagnostic products such as housekeeping data and caveats. As CAA is an active archive, it is envisaged that redeliveries will take place, due to improved calibrations and additional or altered data products. This indeed is happening with RAPID. In the rest of this article, we describe briefly the RAPID instrument, then the data as obtained from the instrument (the base products), and finally the products as they are to be found at CAA. The complete description of the RAPID CAA data products is given in the Interface Control Document [3] (hereafter referred to as ICD). A brief description (of the initial release) has been given by Daly and M¨uhlbachler [2]. What is presented here is intended as a more detailed overview of the current situation.

9.2 The RAPID Instrument A complete technical description of the RAPID instrument has been given by Wilken et al. [5]; here we present a brief summary to aid understanding of the nature and structure of the data products. RAPID contains two independent detections systems, IES (Imaging Electron Spectrometer) for electrons, and IIMS (Imaging Ion Mass Spectrometer) for ions. Each of these consists of three detector heads, each covering an opening of 60ı acceptance angle (Fig. 9.1). Altogether, the instrument is sensitive to incoming particles within one half-plane containing the spin axis (Fig. 9.2).

Time-of-flight measurement Start Stop

Incoming Ion 15∞

Incoming Electrons

3 Microstrip solid-state detectors Energy and direction

15∞

15∞

Energy measurement in solid-state detectors

15∞

Collimators

Fig. 9.1 Schematic of a single detector heads for IES (left) and IIMS (right)

9

RAPID Products at the Cluster Active Archive IIMS 0

IES

4

3 2

8 3

1

9

5 6

4

0

10 5

7 8

6

9

Spin

4

7 2

3

10

5

6 1

8 Spin

7

in Sp

0

1 2

11

147

15

11 12

13

14

RAPID at 60.167 deg from Sun at start of Spin Sun Sensor Offset = 26.367 deg

Fig. 9.2 The IIMS and IES polar segments relative to the spin axis (left and center) and the RAPID sectorization relative to the sun (right)

9.2.1 Electrons One of the three IES detector heads is shown in the left side of Fig. 9.1. Each functions as a pin-hole camera, with three microstrip detectors to locate the incoming particles to within 20ı . Not shown is a foil between the pin-hole aperture and detectors that removes any ions up to 350 keV. A single head covers an acceptance angle of 60ı in all; a total of three heads span the entire range of 180ı with 9 “pixels” (middle of Fig. 9.2) in one half-plane. During one spacecraft spin (4 s), this plane rotate through 360ı, during which time, 16 measurements are made; one speaks of 16 spin sectors which are synchronized to the sun (right side of Fig. 9.2). A 3-D distribution thus consists of 9 polar 16 azimuth directions. The 3rd dimension, energy, is determined by the charge deposited in the detector by the absorbed electron, which is proportional to that energy (minus a constant). The accumulated charge is swept out and converted to a digital signal at regular intervals. Because of background currents, zero energy corresponds to some finite signal, called the pedestal; eight energy channels are defined in terms of the 256 bin numbers, relative to the expected pedestal location (Fig. 9.3). The read-out interval is called the integration time; it is adjusted to ensure that there is no more than one count per interval, otherwise pile-up occurs and it is the summed energy that is read out instead. The integration time switches automatically between values of 2, 5, 15, and 50 s. The read-out time itself is about 50 s, which is why it is desirable to keep the integration time as long as possible, to maximize the duty cycle. Complications There can be jumps in the flux values at such switching times as the

erroneous pile-up effects are suddenly removed. Corrections for this are not possible; the user must be careful with such jumps.

148

P.W. Daly and E.A. Kronberg

Pedestal

Electron Line

Electron energy relative to pedestal

PHA Ch# 0

N

N+45

N+90

N+135

0

100

200

300

255

Energy, keV

Fig. 9.3 The IES pulse-height analysis. The analog signal, reflecting the electron’s energy on top of a background pedestal, is digitized from 0 to 255, and then sorted into broad energy bins that must be set relative to the pedestal

Since the pedestal position moves with time (aging), temperature, and even

count rate, its position is constantly monitored and corrected for in the data. This is described in Sect. 9.3. However, the corrections are not always perfect and the user should be aware of this.

9.2.1.1 The Electron Base Products The electron data sets as delivered directly by IES exhibit some importance differences between the nominal telemetry mode (NM, low bit rate) and burst mode (BM, high bit rate). Burst mode: There is one full 3-D electron distribution per spin, with 9 polar directions, 16 azimuthal sectors, and eight energy channels. The polar directions 0–2 and 6–8 (those closest to the spin axis, Fig. 9.2) really have only eight sectors per spin; the CAA products artificially double this to simplify handling the data as a rectangular array of size f8,16,9g. Nominal mode: Counts are summed over the whole spin, so angular information is limited to nine polar directions; thus this is often called 2-D data, as reflected in the CAA product name. There are also only six energy channels, the top two corresponding to the upper 4 in burst mode. The array size is f6,9g. Omnidirectional: There are no omnidirectional electron fluxes in the base products, but one is created artificially for CAA from the 3-D/2-D data. It always has the six energy channels of nominal mode: array size f6g.

9

RAPID Products at the Cluster Active Archive

149

EPAD: To compensate for the missing 3-D data in NM, there is an additional product called EPAD (electron pitch angle distribution): it contains fluxes from three polar directions for each of 16 sectors, but only for two energy channels. One direction is perpendicular to the magnetic field in that sector, another 90ı to the first, the third is either parallel or anti-parallel to the spin axis. To simplify life for the user, the CAA product places these values into their proper location within an array of f2,16,9g with the other six directions having fill values. The definitions of the two energy channels has changed several times over the first years of operation; hence three separate versions of this product exist with different energies. All three versions cease after May 2004 (the CAA data files exist but are empty), when they were replaced by the next product. Electrons lite: Since May 2004, a new 3-D electron product is available in nominal mode. It is the same as the burst mode distribution but for energy channels 1 and 3 only. For CAA, this product is also provided in burst mode, by simply extracting it from the existing burst mode distribution. The array size is f2,16,9g. All the above products have a time resolution of one spin 4 s.

9.2.2 Ions A single ion head is illustrated on the right side of Fig. 9.1. The entering particle first traverses collimating plates, penetrates a thin foil that emits electrons to generate a start signal, and then strikes the solid state detector, where electrons emitted at the surface produce a stop signal. The particle energy is deposited in the detector and recorded. The opening angle for a single ion detector is also 60ı , but unlike for the electrons, the subdivision into four 15ı slots is done by determining the location of the start signal. This of course has a certain efficiency associated with it. The start and stop signals are used to measure the time-of-flight (TOF) over the 32 mm separation. This time plus measured energy locate the event within the energy-TOF space of Fig. 9.4 which then determines the species and energy channel. The number of events in each of the species-channel bins in Fig. 9.4 is read out and reset once per sector (1/16 spin). Only the hydrogen, helium, and carbon-nitrogen-oxygen (CNO) data are delivered by RAPID as part of the regular products; the silicon-iron channels are also available but only in the direct events product (see below). Complications The start and stop signals are generated by multichannel plate detec-

tors, whose performance change considerably over time. Thus the overall efficiency of the TOF mechanism is monitored continuously and the

CNO7

100 H2

SiFe1

SiFe2

SiFe6

He2

H3

CNO 1

CNO2

H4

CNO4 CNO3

He5

He1

H5

SiFe5 SiFe4 SiFe3

He6

CNO5

Measured energy, keV

H6

Over Range = 1500 keV

He3

1000

CNO6 SiFe7

He7

He4

H7

He7

CNO7

H6

CNO8

H7

He6

P.W. Daly and E.A. Kronberg

CNO8

150

H1

Under Range = 30 keV

H1

He1

CNO1

10 1.25

2.50

5.00

10.00 20.00 Time of Flight, ns

40.00

80.00

Fig. 9.4 The RAPID ion classification system. Every ion “event” occupies a point in the 256 256 energy-TOF space, which is divided into areas for different ion species and up to eight energy channels

calibration files regularly updated. The user must realize that interspacecraft cross-calibrations are only approximate and must be used with great care. The lower part of Fig. 9.4 shows the “underrange” region, where no energy signal is received (<30 keV); ion classification is done solely on TOF information. These events contribute to energy channel 1. The species determination is not so precise: low energy hydrogen enters and dominates the helium channel 1 (which is therefore suppressed) and the CNO channel 1 exhibit signs of contamination as well. The central ion head on all four spacecraft deteriorated within the first months of the mission; most likely sunlight saturated the time-of-flight system. The result is that ion distributions are not available within 30ı of the ecliptic plane. This is often referred to as the “donut” effect (hole in the middle). There is a bug in the onboard programming that causes the 3-D ions to be accumulated at different times in different sectors. The CAA product provides a key to help recombine these distributions. This is described in an appendix to the ICD.

9

RAPID Products at the Cluster Active Archive

151

9.2.2.1 The Ion Base Products The ion data products delivered directly by IIMS can be summarized as follows. 3-D data: Ion fluxes for three species (H, He, CNO) in 12 polar and 16 azimuthal directions over eight energy channels. For CAA, this is split up into three separate products, one per species. Array size is f8,16,12g. Each distribution is accumulated over 8 spins, except on spacecraft 2 where they are accumulated over a single spin. In burst mode, there is one such distribution every 8 spins; in nominal mode, there is 1 every 32 spins. This is a result of the limited telemetry rate for RAPID. There are thus data gaps between the delivered accumulations. Omnidirectional: There is an omnidirectional base product for H, with eight energy channels. Since these events do not depend upon the efficiency of the directional signal mentioned above, they are not identical to the sum of the 3-D data. Array size is f8g, frequency is once per spin. There are also omnidirectional He and CNO spectra at a rate of 1 per 4 spins: their array sizes are f8g. IPAD: As for the electrons, there is a sparse 3-D hydrogen distribution available every spin, with three polar directions per sectors, which vary with the magnetic field. Like EPAD, it too is expanded to fill the complete 3-D array with fill values for those directions not measured. The array size is f2,16,12g. As of May 2004, this product exists only in burst mode. Direct events: The complete energy and time-of-flight values of Fig. 9.4 plus polar and azimuthal direction numbers are transmitted for a limited number of events per spin: 20 in NM and 106 in BM. A priority scheme ensures that the rarer heavy ion events are read out before the dominating hydrogens. For CAA, the species and energy channels determined from the scheme in Fig. 9.4 are also provided. Details are to be found in the ICD.

9.3 Standardized Energy Channels A number of different values for the RAPID energy channels have been published over the years, especially for the electrons. The reason is that each of the electron and ion sensors in the different heads and on different spacecraft really do have deviating values. For the electrons, this is mainly due to setting some lower thresholds high enough to avoid contamination from a wide pedestal, while for the ions, slightly differing foil thicknesses lead to variations in energy losses at the start signal. However, a uniform set of energy channel definitions is essential for any physical analysis of the data. Hence, the variations in the spectra among the sensors are corrected to bring them effectively to such a standardized set. The “official” values now used for the CAA products are listed in Table 9.1. The algorithm used for this correction is to fit the data between adjacent channels to a power law which is then integrated over the energy shift region, with the resulting flux subtracted from the one and added to the other channel, as shown in Fig. 9.5.

152

P.W. Daly and E.A. Kronberg Table 9.1 The standardized RAPID energy channels Electrons Channel 1 2 3 4 5 6 7 8 Upper

Hydrogen a

27.7–64.4 75.3 92.2 159.7 374. 962. 1885. —e 4007.

Helium b

29.9 137.8 172.3 235.1 351. 737. 1689. —e 3799.

CNO

BM

c

NM d

37:3d 50:5 68:1 94:5 127:5 244:1 — — 406:5

37:3 50:5 68:1 94:5 127:5 175:9 244:1 336:5 406:5

83.8 274.4 414. 498. 638. 948. 1414. 2539. 4046.

Lower thresholds, all units keV a Energy gap between hydrogen 1 & 2. b Helium channel 1 contains hydrogen, suppressed. c CNO channel 1 contaminated at times, not suppressed. d Some electron channels start at 46.1 keV. e 8th channel not accessible for hydrogen and helium.

Fitted power law j = AE

−γ

Ln j

j1

j2

E1

e1

E2 E2’

e2

E3

Ln E

Fig. 9.5 Shifting the energy threshold between two channels: the differential flux in two channels is fitted to a power law, that fit is integrated over the energy shift E20 E2 (shaded area), and the result moved from one channel to the other

For the electrons, there is a further adjustment to be made: because the pedestal that defines the energy origin can move in a way that depends on count rate, its location is monitored by some additional channels. The spectral shift introduced by this effect can be deduced and then also corrected for as part of the energy channel standardization.

9

RAPID Products at the Cluster Active Archive

153

Note that the spectral correction is applied only to flux data, not to the count rates since they are to remain as original raw data.

9.4 CAA Products So far we have described the RAPID products as they come from the instrument. Here we explain what goes into the Cluster Active Archive specifically. The primary idea is to archive as much as possible so that the RAPID data may be used both today and in the far future when the RAPID Team is no longer available for advice. Obviously this includes processed data as flux, with all calibrations, cleaning, corrections, and glitch removal included; but it also means archiving raw data as count rates, housekeeping data, instrument settings and other data sets that normally can only be exploited by very knowledgeable users. If these expert data sets are not archived they will later be lost forever. They are essential to understand any puzzling and possibly erroneous effects in the processed data. Here we will not detail the names of the CAA files or the variable names in them. That can be found directly from CAA or from the ICD.

9.4.1 Main Products The most important products are those derived from the base products described above. The following points apply to them: Processed data are in units of differential particle flux, 1/(cm2 sr s keV). Raw data are count rates (1/s); these contain no corrections, no calibrations, no

suppressions whatsoever; only the accumulation time is employed to convert from counts to rates. This accumulation time is well-defined but not trivial to determine, hence it is included as an item in every count-rate record (the original number of counts can thus be reconstructed if wanted). For both fluxes and rates, the standard deviation is also included in each record, in the same format and array size as for the data. This is obtained from both the Poisson statistical error of the counts plus a factor for the data compression uncertainty; the former normally dominates. There is a CEF file for every product for every day and spacecraft; for those times when the product does not exist (e.g. the Electrons Lite distributions before May 2004) the files are empty. The record time tag refers to the middle of the record interval; the time-halfinterval for each record is also given (CAA metadata standard). Note that the record interval and the accumulation time are two separate things: data are accumulated over a certain interval but not necessary for 100% of that time.

154

P.W. Daly and E.A. Kronberg

The direct events are also a main product, but are different in that they are a listing of a limited number of event parameters. No rates or fluxes can be derived from them: they only indicate the presence of particles. For measured zero flux, the standard deviation is also set to 0, although it should really be that for a single count, to express the uncertainty level. However, such a procedure would lead to errors when summing over longer times. Recall that errors are combined by summing the variance, i.e. the sum of the squares of the standard deviations.

9.4.2 Support Products Some additional data sets are provided to assist users with the interpretations; their derivation requires detailed knowledge of the RAPID location on the spacecraft, the spin axis, and sun sensor. Thus all this is worked out for the user, as separate products. Flow directions: For each of RAPID’s 9 16 (electrons) and 12 16 (ions) polarazimuth directions, the GSE coordinates of the flow vector at their centers are provided. Since these values change very slowly, they are only given once per hour. SC-GSE rotation matrix: Provided in the same files as the flow directions is the actual 3 3 rotation matrix to do a general transformation to GSE from the RAPID spacecraft frame ( D angle to spin axis, D sector 22:5ı ). Pitch angles: For each of the electron and ion polar-azimuth flow directions, the angle to the magnetic field (pitch angle) plus a magnetic azimuth is provided; this is done for each spin. Warning: this calculation uses the magnetic data delivered directly to RAPID on-board and not the final calibrated FGM data from CAA. Proper pitch angle calculations must be done by combining the flow directions with that magnetic data set.

9.4.3 Diagnostic Products These are products that are intended mainly for experts and are explained further in the ICD. Housekeeping parameters: Various flags and parameters (137 in all) used to indicate the instrument status or mode or performance. Instrument settings: A subset of the housekeeping data, 11 in all, for the most important aspects of the instrument status; this includes telemetry mode, multichannel plate voltages, electron integration time, ion accumulation mode.

9

RAPID Products at the Cluster Active Archive

155

Pedestals: Information on the electron pedestals: counts (not rates) in the pedestal channels, calculated pedestal positions, used for spectral corrections, for burst and nominal mode separately.

9.4.4 Graphics and Software The RAPID Team also makes available an IDL visualizing software package and summary plots. The software package can be used to read RAPID data files (the CAA CEF format or the RAPID-specific SCI format) to produce various types of plots: line plots, spectrograms, angle-angle plots, and pitch angle distributions. The user can combine different spacecraft or different data sets from one spacecraft. Summary plots, as illustrated in Fig. 9.6, are of 6-h duration, and contain omnidirectional spectrograms for electrons, hydrogen, helium, and CNO, with various instrument settings as well.

9.5 Status of RAPID Data at CAA Since the CAA is an active archive, it is expected that the contents shall be replaced, refined, and/or updated in the course of time. There are several reasons for doing this.

9.5.1 Recalibration The calibration factors used for the first number of years were preliminary ones based on pre-launch measurements and tests during commissioning. Only over the longer operational phase could actual data be used to establish the need for recalibration. The original data delivered to CAA for 2001–2004 were made using these preliminary calibrations. New calibrations are now available and were used for the initial delivery of the 2005 data. Reprocessing of the 2001–2004 data (and reissuing of 2005) will take place after the delivery of the 2006 data, starting mid-2008. The calibration version can be found in the data files. There is a metadata parameter called Dataset Version, which for RAPID is a 4-digit code, the most significant digit of which is 1 for the older and 2 for the newer calibrations. See the ICD for further details. Ultimately, all the older versions will be replaced with the newer ones.

156

P.W. Daly and E.A. Kronberg Cluster/RAPID SC3 (Samba) 2008-02-20 18:00:00 – 23:59:59 Time resolution = 1 min Energy, keV

103

Electrons 104 103 102 101

102

100 Energy, keV

Protons

104 103

103

102 101

102

100

Energy, keV

Diff. flux [1/(cm2 sr s kev)]

Helium 102

103

101

102

100 103 102

103

101 102

100

ON

AutoSW

OFF

HIST

50 us 15 us 5 us 2 us A B TM 10 8 6 4 2 0

HV, kV

IIMS Modes

Int Time

Energy, keV

CNO

XGSE YGSE ZGSE R

S/B

Serial

Paralel IFFT

HV off

HV on

HV ok ENA

NM

BM

BM3

STA 18 1.86 3.73 3.37 5.36

NM3

STO 19 3.84 4.68 3.18 6.84

20 5.58 5.34 2.77 8.20

DEF 21 7.12 5.79 2.26 9.45

22 8.50 6.10 1.70 10.59

23 9.73 6.31 1.10 11.64

24 10.83 RE 6.43 RE 0.50 RE 12.61 RE

Hours of 2008 Feb 20 Generated 2008-Jul-16 21:41:39; V2.1; Data Vers 2123 (E) & 2220 (I)

Fig. 9.6 RAPID 6-h summary plot

9

RAPID Products at the Cluster Active Archive

157

9.5.2 Reissues A re-release of the data products is also necessary when new or revised products are ready. The first release is missing the diagnostic products and some minor main products, which are coming with the next release. Even so, some of the main products require format changes and must therefore be replaced. In particular, the original nominal mode electron products were artificially given eight energy channels, to be consistent with burst mode. This has led to misunderstandings and even errors, so these products are being replaced by new ones with the proper six channels.

All the product descriptions in this article are for the newer release.

9.6 Summary The RAPID products at CAA aim to be as complete as possible to ensure reliable and convenient usage by the general scientific community. For this reason, the Team provides not only processed data as fluxes, but also raw data as count rates, and standard deviations for error estimates for both fluxes and rates. Support data for coordinate transformations as well as diagnostic products for “experts” are also available. Delivery of new data is proceeding on schedule; reprocessing of older data with improved calibrations is also under way. The CAA repository and its CEF format will soon become the primary means of obtaining and using RAPID data, especially as more data is ingested and the calibrations improved. Acknowledgements The authors wish to thank the CAA and its Team for all the efforts they have put in to making the Active Archive a functioning and useful system.

References 1. Allen A., Schwartz S.J., Harvey C., Perry C., Huc C., Robert P.: Cluster Exchange Format— Data File Syntax. ESA document DS–QMW–TN–0010, Queen Mary University of London, London, UK (2004). Latest version available at http://caa.estec.esa.int/caa/documentation.xml 2. Daly P.W., M¨uhlbachler S.: The Cluster Active Archive—The RAPID contribution. In: P. Escoubet, H. Laakso, M. Taylor, A. Masson (eds.) Proceedings Cluster and Double Star Symposium—5th Anniversary of Cluster in Space, Noordwijk, The Netherlands, 19–23 September 2005, ESA SP-598. ESA Publ. Div., Noordwijk (2006). http://caa.estec.esa.int/ documents/publications/P6.11 Daly.pdf

158

P.W. Daly and E.A. Kronberg

3. Daly P.W., M¨uhlbachler S., Kronberg E.: Cluster Active Archive: Interface Control Document for RAPID. ESA document CAA–RAP–ICD–0001, Max-Planck-Institut f¨ur Sonnensystemforschung, Katlenburg-Lindau, Germany (2008). Latest version available at http:// caa.estec.esa.int/caa/instr doc.xml 4. Harvey C.C., Allen A.J., D´eriot F., Huc C., Nonon-Latapie M., Perry C.H., Schwartz S.J., Eriksson T., McCaffrey S.: Cluster Metadata Dictionary. ESA document CAA–CDPP–TN– 0002, European Space Agency, Noordwijk, Netherlands (2008). Latest version available at http://caa.estec.esa.int/caa/documentation.xml 5. Wilken B., Axford W.I., Daly P., et al.: RAPID: The Imaging Energetic Particle Spectrometer on Cluster. Space Science Reviews 79, 399–473 (1997)

Chapter 10

STAFF Instrument Products Distributed Through the Cluster Active Archive N. Cornilleau-Wehrlin, L. Mirioni, P. Robert, V. Bouzid, M. Maksimovic, Y. de Conchy, C.C. Harvey, and O. Santol´ık

Abstract Cluster Spatio-Temporal Analysis of Field Fluctuations (STAFF) high resolution data that are available at Cluster Active Archive (CAA) comprise two main parts, corresponding to data issued from the two onboard data analysers. The STAFF waveform analyser (STAFF-SC) provides data in the frequency range 0.1–10 Hz or 0.1–180 Hz, depending on the spacecraft telemetry rate. The Spectrum Analyser (STAFF-SA) calculates the complete spectral matrix elements for five wave components, the three magnetic components from the STAFF experiment and the two electric components from the Electric Field and Wave (EFW) experiment, in the frequency range 8–4,000 Hz. From STAFF-SC the CAA data comprise waveform data in telemetry units, complex spectra in physical units, dynamic spectra plots, and possibly in the future waveform data in physical units. The CAA products from STAFF-SA are the spectral matrix data in physical units. In the future, STAFF-SA value added products containing wave polarisation and propagation characteristics will be delivered.

10.1 Introduction The purpose of the present paper is to give a short overview of the STAFF instrument itself and its data available through the Cluster Active Archive (CAA). N. Cornilleau-Wehrlin (), L. Mirioni, P. Robert, and V. Bouzid Laboratoire de Physique des Plasmas, Ecole Polytechnique, UPMC, CNRS, route de Saclay, 91128 Palaiseau, France e-mail: [email protected] N. Cornilleau-Wehrlin Station de Radioastronomie de Nanc¸ay, 18330 Nancay, France M. Maksimovic and Y. de Conchy LESIA, Observatoire de Paris-Meudon, 5 place Jules Jansen, 92195 Meudon Cedex, France C.C. Harvey CESR, 9 avenue du Colonel Roche, BP 4346, 31028 Toulouse Cedex 4, France O. Santol´ık Institute of Atmospheric Physics, ASCR, Bocni II/1401, CZ-14131 PRAHA 4, Czech Republic H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 10, c Springer Science+Business Media B.V. 2010

159

160

N. Cornilleau-Wehrlin et al.

The scientific rationale underpinning the CAA is to: Maximise the scientific return from the mission by making all Cluster data avail-

able to the worldwide scientific community Ensure that the unique data set returned by the Cluster mission is preserved in a

stable, long-term archive for scientific analysis beyond the end of the mission Provide this archive as a major contribution by ESA and the Cluster science

community to the International Living With a Star programme In the case of the Spatio-Temporal Analysis of Field Fluctuations (STAFF) experiment, the main responsibilities are to deliver to CAA high resolution data in an agreed format at the best possible quality level, as far as possible in physical units. Additional value added products will be delivered too.

10.2 Instrument Hardware The Cluster STAFF experiment (see Cornilleau-Wehrlin et al. [2] and CornilleauWehrlin et al. [3]) comprises a boom-mounted three-axis search coil magnetometer to measure magnetic fluctuations in the frequency range 0.1 Hz–4 kHz, a preamplifier and an electronics box that houses the two complementary data analysis packages: an on-board waveform unit and a digital Spectrum Analyser. The later uses the five electromagnetic wave components available, the three magnetic coming from STAFF and the two electric coming from the Electric Field and Wave experiment (EFW [4]). STAFF is one of the five experiments of the Wave Experiment Consortium (WEC [7]).

10.2.1 The Magnetic Waveform Unit The magnetic waveform unit (STAFF-SC) is made of three sections in order to fulfil different filtering and waveform digitalisation, output interface and onboard calibration (for the whole STAFF experiment). The three magnetic components of the magnetic field (Bx , By and Bz ), at the output of the search coil preamplifier are filtered simultaneously in either one of the two bandwidths 0.1–10 and 0–180 Hz depending on the spacecraft telemetry mode, then they are digitised simultaneously by a 16 bits sampling and hold device at 25 and 450 Hz, respectively. STAFF and EFW samplings are synchronized by the Digital Wave Processor (DWP). It is also to be noticed that EFW and STAFF filters are identical.

10.2.2 The Spectrum Analyser The Spectrum Analyser (STAFF-SA) is designed to perform the complete auto and cross correlation matrix of five sensor channels (Bx , By , Bz , Ey , Ez ) over a

10

STAFF Instrument Products Distributed Through the Cluster Active Archive

161

frequency range of more than nine octaves at a high rate. The two electric components come from EFW. The “frontend” of the analyser is analogue. The analysis band of 8 Hz–4 kHz is divided into three logarithmically distributed frequency subbands (A: 8–64 Hz, B: 64–512 Hz and C: 512–4,096 Hz), each one being divided into nine frequency channels. The analysis band is therefore divided into 27 frequency bands, logarithmically spaced. For each of the three sub-bands and for each of the five sensor channels there is a separate band pass filtering. For each sub-band there are three automatic gaincontrolled amplifiers. It is worth noting that in High Bit Rate telemetry mode there are no data for the lowest frequency band A as there is an overlap with the STAFF and EFW waveform data (up to 180 Hz), which permits to save telemetry for the benefice of other WEC experiments.

10.3 Data Processing Figure 10.1 shows a general picture of how STAFF data are processed on, and how CAA files are created. The DWP team produced the TED software that extracts and time-tags the STAFF Level 0 data from the WEC raw data packets, producing different data packets for STAFF-SC and STAFF-SA data that are then decommutated (Level 1). The more recent version of TED has an option allowing improving the precision of the time tag, from 2 ms (standard precision) to about 20 s. For this application DWP uses tables given by ESOC and DSN [Yearby et al., Chapter 4]. For inter-spacecraft analysis this correction is specially needed for waveform data in the High Bit Rate

CWF CS

DWF WEC Level

Level 0

Level 1

STAFF L0 File

STAFF L1 File

Level 2

Normal Modes Burst Modes

TED

Decom

...

Level 3

STAFF−SC L2 File

RC L

STAFF−SC L3 Files

STAFF−SA L2 File

PRAS− SADCO

STAFF−SA L3 Files

Calib

House Keeping

SATT File

Plots

Calibration Files

PSD SM AGC

Polarisation Propagation Parameters

Fig. 10.1 STAFF data ground processing chain. Products available at CAA are, on one hand, Decommutated Waveform (DWF), Complex Spectra (CS) and graphical plots of spectrograms (Plots) for STAFF-SC data, and on another hand, Power Spectral Density (PSD), Spectral Matrix (SM) and Automatic Gain Control (AGC) for STAFF-SA data

162

N. Cornilleau-Wehrlin et al.

(HBR) mode (up to 180 Hz). At 180 Hz, the phase resolution is 260ı for 2 ms precision, going to 3ı with the improved time tag. This option is used for STAFF waveform Level 1 data delivered in CAA. For telemetry limitation reasons, waveform data are compressed onboard in the DWP unit. During the decompression of STAFF-SC data on ground, it is possible to estimate the loss due to the on-board compression from 16 to 12 bits (maximum value of the error). This maximum error is given with the Level 1 waveform data for each component and each sample. Other data are also recorded, in particular spin phase and spin axis direction for polarisation and propagation direction computation. Level 1 file also contains a status word which records operation mode information.

10.3.1 STAFF-SC STAFF-SC data are calibrated with dedicated software [8]. From Level 1 files, this software performs time checking, calibration, filtering, coordinate system transformations and other processing operations including dynamic spectra and wave polarisation computations to produce calibrated data files (Level 2) and value added products (Level 3). Value added products such as dynamic spectra produced routinely by the STAFF team (Level 3 graphical plots) are also stored in the CAA.

10.3.2 STAFF-SA The calibration procedure for the STAFF-SA is fully described in Harvey et al. [5]. The calibrated data (Level 2) are then processed by the PRASSADCO software [9] in order to get the polarisation and propagation parameters, in addition to the five components of the power spectral density.

10.4 Available STAFF Data Products in the CAA Before delivery to CAA, these data are translated into the Cluster Exchange Format (CEF, [1]), and then stored in the CAA database [6], together with the PNG graphical plots (e.g. Fig. 10.2). The Cluster Exchange Format is ASCII which allows the end-user to actually see, with a simple text editor, what the file does contain. The file header may seem hard to decrypt because it records every piece of information one may expect, but the way data are recorded is as clear and simple as possible and may be read quite easily with any usual scientific software.

10

STAFF Instrument Products Distributed Through the Cluster Active Archive

Fig. 10.2 Example of STAFF-SC NBR dynamic spectra

163

164

N. Cornilleau-Wehrlin et al.

Table 10.1 Summary of STAFF data files in the CAA. CWF are not yet available, they will be delivered on a best effort basis Level Product Experiment Mode Number of files Format 1 Decommutated WaveForm (DWF) STAFF-SC NBR 1 file/1 sat./24 h CEF 1 Decommutated WaveForm (DWF) STAFF-SC HBR 1 file/1 sat./24 h CEF 2 Calibrated WaveForm (CWF) STAFF-SC NBR 1 file/1 sat./24 h CEF 2 Calibrated WaveForm (CWF) STAFF-SC HBR 1 file/1 sat./24 h CEF 2 Complex Spectra (CS) STAFF-SC NBR 1 file/1 sat./24 h CEF 2 Complex Spectra (CS) STAFF-SC HBR 1 file/1 sat./24 h CEF 2 Automated Gain Control (AGC) STAFF-SA All 1 file/1 sat./24 h CEF 2 Power Spectral Density (PSD) STAFF-SA All 1 file/1 sat./24 h CEF 2 Spectral Matrix (SM) STAFF-SA All 1 file/1 sat./24 h CEF 3 Dynamic spectra plots STAFF-SC NBR 1 file/1 sat./3 h PNG 3 Dynamic spectra plots STAFF-SC HBR 1 file/1 sat./3 h PNG 3 Polarisation parameters STAFF-SA All TBD CEF 3 Plots STAFF-SA All TBD TBD

Level 0 (L0) data are raw data. None is stored in the CAA. Level 1 (L1) data are only decommutated and time-tagged. This is the case of STAFF-SC Decommutated WaveForm (DWF) which is provided un-calibrated. Level 2 (L2) data are calibrated (physical units). One will find STAFF-SC Complex Spectra (CS), STAFF-SA Spectral Matrix (SM) and Power Spectral Density (PSD) data in the CAA. Level 3 (L3) data are value-added products. The only type of L3 products delivered to CAA yet, are graphical plots of dynamic spectra for browsing purpose only, but STAFF-SA polarisation and propagation parameters data will also be stored in CAA. The reader will find a summary of CAA-STAFF products main characteristics in Table 10.1.

10.4.1 STAFF-SC Decommutated Waveform These L1 CEF files contain the Decommutated WaveForm (DWF) of the magnetic field (un-calibrated) sampled at 25 Hz for the Normal Bit Rate (NBR) mode and 450 Hz for the High Bit Rate (HBR) mode. Other variables or constant data (constant for one given file) are also stored here: the spin axis direction, the mean spin rate, the status (containing HK information), the phase angle and the maximum compression error. Bx , By and Bz are given in telemetry counts units. To keep as much information as possible, it has also been decided to store Level 1 data in the STAFF instrument reference frame (so-called SSW6RF). This minimizes information loss (this way loss can only result from compression error which is estimated in the instrument reference frame) as data records are as close to raw data as possible except that they are nice-looking and time tagged with the highest precision available.

10

STAFF Instrument Products Distributed Through the Cluster Active Archive

165

10.4.2 STAFF-SC Complex Spectra These L2 files give the values of Complex Spectra (CS) of the magnetic field in GSE coordinate system, in the 0.1–12.5 Hz and 0.1–225 Hz frequency range in NBR mode and in the HBR mode respectively, with a time resolution of 10 s and a frequency resolution of 0.1 Hz.

10.4.3 STAFF-SC Dynamic Spectra These L3 files are graphical plots showing dynamic spectra and integrated power for the component parallel to the spacecraft spin axis and for the four satellites. At the bottom the amplitude of the DC field perpendicular to the spin axis, computed from the search-coil spin signal, is also plotted for the four spacecraft. Plots are delivered in time series of 3 h for HBR or NBR (e.g. Fig. 10.2). They are provided in Portable Network Graphics (PNG) format for preview.

10.4.4 STAFF-SA Automatic Gain Control These L2 CEF files contain the three measured values of the Automatic Gain Control (AGC) for Bx , By C Bz and Ey C Ez , for each of the three frequency bands (A, B and C). The AGC data are not subject to direct scientific use. However they represent a measure of the total wave power for the corresponding sensor or group of sensors (B or E in the spacecraft spin plan) and frequency band.

10.4.5 STAFF-SA Power Spectral Density and Spectral Matrix These L2 CEF files store the records of the 5 5 complex Spectral Matrix (SM) which is the cross-product of the magnetic and electric fields values computed onboard the spacecraft: 0 B B B B B @

Bx2 By :Bx

Bz :Bx

Ex :Bx

Ey :Bx

Bx :By

By2 Bz :By

Ex :By

Ey :By

Bx :Bz

By :Bz

Bz2 Ex :Bz

Ey :Bz

Bx :Ex

By :Ex

Bz :Ex

Ex2 Ey :Ez

Bx :Ey

By :Ey

Bz :Ey

Ex :Ey

Ey2

1 C C C C C A

As this matrix is complex, the real and imaginary parts have been separated in the SM CEF file.

166

N. Cornilleau-Wehrlin et al.

It has also been decided to cut it in three pieces to be unit consistent, the first being the magnetic elements B B, the second the electric ones E E and the third the cross-product B E. As the matrix is also non-hermitian, that is Si :S j D Sj :S i (where S D B or E, and i, j D x, y, z), only half of the non-diagonal elements are stored. The Power Spectral Density (PSD) file is a record of the SM diagonal values. PSD are averaged over shorter time scales than the SM elements. That is why they are stored in a separated CEF file. All of these values are given for each of the 27 frequency channels (3 bands 9 sub-bands logarithmically spaced). The time resolution is 1 s for PSD and 4 s for SM in NBR mode, and 125 ms or 250 ms for PSD and 1 s for SM in HBR mode. There is no A-band measurement in the HBR mode.

10.4.6 Caveats Presently there are few caveat files for STAFF CAA products. Most of the data gaps, errors, telemetry overflow relevant to WEC data, either as a whole or for individual experiments, thus in particular STAFF, are listed in the WEC status log provided by the DWP team (see Chapter 4). STAFF caveats include information on absence of sun pulse due to eclipses and the presence of erroneous STAFF-SA PSD negative values. In the future, possible new caveats will be produced in conjunction with the new products to be delivered. One possible source of caveat is linked to the possibility to remove the spin signal nicely or not in the vicinity of the perigee of the Cluster orbit, where this signal is the strongest. This is not an issue with the data that are presently delivered to CAA. This might become a need once the waveform data in physical units are delivered, as for this some data processing is needed, as a Fourier transform to correct from the transfer function followed by an inverse one to get wave form in physical units. This explains partly why such a product is not delivered yet. The CSDS product that is the integrated power density in the frequency range 1–10 Hz is accompanied by caveats sometimes when in the vicinity of the perigee. This is translated in the delivered data by an increase of the signal level at 0.25 Hz and maybe at 0.5 Hz, close to perigee, those frequencies corresponding to the spin frequency and its first harmonic. Another caveat would be linked to the EFW experiment for the availability of electric component data for the Spectrum Analyser. This should be found in EFW experiment caveats: it concerns periods after one of the EFW boom suffers malfunction on one spacecraft or another (filter or preamplifier failure). In that case the electric field data are meaningless as STAFF-SA combines onboard the two components to de-spin the data. The priority has been given for STAFF-SA to keep good magnetic field data, onboard de-spun, allowing to get the polarisation parameters of the electromagnetic waves.

10

STAFF Instrument Products Distributed Through the Cluster Active Archive

167

10.5 Prospects 10.5.1 CAA STAFF Data Status At present time STAFF Level 2 data that are available at CAA cover at least years 2001–2005. All data (Levels 1, 2 and 3) covering the years 2001–2009 should be available by the end of 2010. STAFF data that are available at CAA are also delivered by STAFF to the French data centre Centre de Donn´ees de la Physique des Plasmas (CDPP).

10.5.2 STAFF Data Prospects The polarisation and propagation parameters from STAFF-SA data will be provided (delivery start is scheduled on 2009). STAFF-SC waveform data in physical units (CWF) will be provided on best effort basis. Acknowledgments STAFF experiment realisation and software have been founded by CNES and ESA.

References 1. Allen, A., Schwarz, S.J., Harvey C.C., Perry, C., Huc C., Robert P.: Cluster Exchange Format: Data FileSyntax. Technical Report DS-QMW-TN-0010, issue 2, rev. 0.4, ESA (2008) 2. Cornilleau-Wehrlin, N., Chauveau, P., Louis, S., Meyer, A., Nappa, J.M., Perraut, S., Rezeau, L., Robert, P., Roux, A., de Villedary, C., de Conchy, Y., Friel, L., Harvey, C.C., Hubert, D., Lacombe, C., Manning, R., Wouters, F., Lefeuvre F., Parrot, M., Pinc¸on, J.L., Poirier, B., Kofman, W., Louarn, P.: The Cluster Spatio-Temporal Analysis of Field Fluctuations (STAFF) Experiment, Space Science Reviews, 79:107–136 (1997) 3. Cornilleau-Wehrlin, N., Chanteur, G., Perraut, S., Rezeau, L., Robert, P., Roux, A., Villedary, C. de, Canu, P., Maksimovic, M., Conchy, Y. de, Hubert, D., Lacombe, C., Lefeuvre, F., Parrot, M., Pinc¸on, J.L., D´ecr´eau, P.M.E., Harvey C.C., Louarn, Ph., Santol´ık, O., Alleyne, H.St.C., Roth, M., STAFF team: First results obtained by the Cluster STAFF experiment, Annales Geophysicae, 21:437–456 (2003) 4. Gustafsson, G., Bostr¨om, R., Holback, B., Holmgren, G., Lundgren, A., Stasiewicz, K., ˚ en L., Mozer, F.S., Pankow, D., Harvey, P., Berg, P., Ulrich, R., Pedersen, A., Schmidt, R., Ahl´ Butler, A., Fransen, A.W.C., Klinge, D., Thomsen, M., F¨althammar, C.-G., Lindqvist, P.-A., Christenson, S., Holtet, J., Lybekk, B., Sten, T.A., Tanskanen, P., Lappalainen, K., Wygant J.: The Electric Field and Wave Experiment for the Cluster Mission, Space Science Reviews, 79:137–156 (1997) 5. Harvey, C.C., Belkacemi, M., Manning, R., Wouters, F., de Conchy, Y.: STAFF Spectrum Analyzer, Conversion of the Science Data to Physical Units. Technical Report OBSPM-TN-0001, issue 7, rev. 5, Observatoire de Paris-Meudon, http://www.cesr.fr/harvey/TN01 V75.pdf (2004)

168

N. Cornilleau-Wehrlin et al.

6. Mirioni, L., Cornilleau-Wehrlin, N., Maksimovic, M., Burlaud, C., Robert, P.: Cluster Active Archive: Interface Control Document for STAFF, Technical Report CAA-STA-ICD-0001, issue 3.0, ESA (2008) 7. Pedersen, A., Cornilleau-Wehrlin, N., de la Porte, B., Roux, A., Bouabdellah, A., D´ecr´eau, P.M.E., Lefeuvre, F., Sen´e, F.X., Gurnett, D., R. Huff, R., Gustafsson, G.,, Holmgren, G., Woolliscroft, L.J.C., J.A. Thompson, J.A., and Davies, P.H.N., The Wave Experiment Consortium (WEC), Space Sci. Rev., 79, 93–106 (1997) 8. Robert, P.: Les Proc´edures Roproc pour le Traitement des Donn´ees de Cluster/STAFF-SC, FGM et ORBIT, Technical Report, CETP, http://cluster.cetp.ipsl.fr/ressources/Roproc V2p0.pdf (2003) 9. Santol´ık, O.: PRopagation Analysis of STAFF-SA Data with COherency tests (PRASSADCO), http://oberon.troja.mff.cuni.cz/santolik/PRASSADCO/staff sa/ (2003)

Chapter 11

Cluster Wideband Data Products in the Cluster Active Archive J.S. Pickett, J.M. Seeberger, I.W. Christopher, O. Santol´ık, and K.M. Sigsbee

Abstract Cluster Wideband Data (WBD) plasma wave receivers are mounted on all four of the Cluster spacecraft providing approximately 4–7% orbit coverage since mission operations began in February 2001. Data obtained by the WBD instruments are downlinked in real time to various Deep Space Network (DSN) and Panska Ves (PV) ground stations. The strengths of the WBD instruments lie in their high time and frequency resolution data, which allow analysis of the wave fine structure in order to more accurately investigate the linear and nonlinear nature of the waves, and in their use in carrying out Very Long Baseline Interferometry (VLBI) investigations to locate and characterize remote wave sources and their characteristics. The WBD team has already archived at the Cluster Active Archive (CAA) the documentation, software and data plots required for carrying on an independent scientific investigation. A document on interpretation issues has also been archived to help ensure that signals which are not naturally in the plasma are not misinterpreted. The WBD team also plans to archive digital, calibrated data files for the entire mission in cdf format at NASA’s CDAWeb and submit these files for conversion to the Cluster standard cef format and archiving at the CAA. In order to ensure that the calibration of the WBD data has been done properly, WBD has taken part in Cluster cross calibration activities with the other wave instruments with regard to wave amplitudes in the time and frequency domains and with the PEACE instruments with regard to low density measurements.

11.1 Introduction The Cluster Wideband Data (WBD) plasma wave receiver is a part of the Cluster Wave Experiment Consortium (WEC) which consists of four other wave instruments (EFW, STAFF-SC, STAFF-SA and WHISPER) and a central Digital Wave J.S. Pickett (), J.M. Seeberger, I.W. Christopher, O. Santol´ık, and K.M. Sigsbee Department of Physics and Astronomy, The University of Iowa, Iowa City, Iowa, 52240, USA e-mail: [email protected] O. Santol´ık Institute of Atmospheric Physics and Charles University, Faculty of Mathematics and Physics, Prague, 18000, Czech Republic H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 11, c Springer Science+Business Media B.V. 2010

169

170

J.S. Pickett et al.

Processor (DWP) (see Yearby et al. [Chapter 4]). WBD instruments are mounted on all four Cluster spacecraft providing high time resolution (5–36 s depending on filter mode) waveforms in the range of 100 Hz to 577 kHz. In order to obtain such high time resolution, WBD measurements along only one axis are telemetered to the ground at 220 kbits/s, with the signal being picked up by either NASA’s Deep Space Network (DSN) or the Czech Republic’s station at Panska Ves (PV). The one axis measurements consist of either the AC electric field from one of the two 88 m spin plane electric field antennas or the AC magnetic field from one of the two searchcoils available for WBD measurements (one is in the spin plane, the other along the spin axis). The WBD data are obtained in one of three filter bandwidth modes: (1) 9.5 kHz, (2) 19 kHz, or (3) 77 kHz. The minimum frequency of each these three frequency bands can be shifted up (converted) from the default 0 kHz base frequency by 125.454, 250.908 or 501.816 kHz in order to obtain higher frequency measurements However, due to internal WBD and other external filters and to antenna constraints (e.g., the magnetic searchcoils have a maximum frequency extent of 4 kHz), the useable portion of each frequency band, i.e., that part which can be calibrated, is restricted to the following ranges: Filter bandwidth (kHz) 9.5 19 77

Electric 100 Hz–9.5 kHz 100 Hz–19 kHz 700 Hz–77 kHz

Magnetic 70 Hz–4 kHz 80 Hz–4 kHz 700 Hz–4 kHz

For example, measurements made with the electric antenna, filter bandwidth of 9.5 kHz and a conversion to 250.908 kHz would provide valid calibrated data in the frequency range of 251.008 to 260.408 kHz. Although frequency components are often observed in the WBD data either below the lower frequency cutoff or above the upper frequency cutoff as shown in the above table, the amplitudes of these components are invalid because their frequencies fall outside the calibrated analysis bandwidth. Thus, use of the WBD data outside the valid frequency range shown above is discouraged. In addition, the resolution of the data can be set to 8, 4, or 1 bit(s). Normal operation is set to 8 bits resolution which means that only the 9.5 kHz filter mode has a 100% duty cycle. Lowering the resolution to 4 bits allows the 19 kHz filter mode to send back continuous data as opposed to 50% duty cycle with 8 bits, and to 1 bit allowing for the 77 kHz filter to obtain continuous data, as opposed to 12.5% duty cycle with 8 bits and 25% with 4 bit mode. For more information on the instrument and its various modes, please refer to Gurnett et al. [1] and the Cluster Active Archive (CAA) WBD Interface Control Document. The WBD data chain is shown in the block diagram of Fig. 11.1. Once the data are received at one of the DSN or PV ground stations, they are Reed Solomon decoded in order to ensure that every transfer frame sent by the spacecraft is received exactly as it was transmitted, and are time tagged with the ground receive time with an accuracy of 10–50 s. DSN and PV then transfer via ftp these raw WBD data files to The University of Iowa for processing. Iowa performs Level 1 processing

11

Cluster Wideband Data Products in the Cluster Active Archive

Fig. 11.1 Block Diagram of the Cluster Data Chain. ( WBD data have been recorded only once onboard (BM2 mode on 7 March 2001) since start of mission operations phase)

171

Input Wave

WBD INSTRUMENT Make waveform measurement using appropriate mode for region, with one parameter selected from each of the following sets: Antenna Conversion Freq. 0 kHz (Baseband) Ez 125.454 kHz Ey Bx 250.908 kHz By 501.816 kHz Bandwidth 9.5 kHz 19 kHz 77 kHz

Resolution 8 bits 4 bits 1 bit Transmit to Ground (220 kbits/s) *

DSN or Panska Ves Processing

FTP Transfer

The University of Iowa Level 1 (L1) Processing Creation of L1 files Validation of L1 files Write L1 files to DVD Mail DVDs

French, Swedish and UK Cluster Data Centres, Stanford University, SwRI Extract L1 Data Local Institute Analyze waveform Or convert to frequency domain

on these files in order to: (1) check for expected synchronization bits to verify that the data in any one data frame are good, (2) convert the onboard counter parameter in each transfer frame to UT time of measurement using the ESOC-provided time

172

J.S. Pickett et al.

calibration files and known delays in the onboard data system, (3) convert the ground receive time supplied by DSN or PV to UT time of measurement with microsecond accuracy taking into account the propagation delay in addition to the delay in the onboard data system, (4) create Level 1 data files with at most 10 min of uncalibrated data together with the aforementioned time tags (steps 2 and 3) attached to each minor frame of data (1,090 data points covering a period of approximately 5 to 39.7 ms, depending on filter mode). The Level 1 data files are then validated by using a set of plots and time analysis routines to check for the goodness of the time tags and to verify that the science data are not corrupted. Once validated, the Level 1 data files are made available on the Cluster WBD home page and written to DVDs. These DVDs are distributed once per month to the WEC data centers in France, Sweden and the United Kingdom and to Stanford University and Southwest Research Institute in the U.S. They contain the validated WBD waveform data, explanation documentation about the instrument, its modes and its calibration, and the calculated time tags described above, as well as software programs that easily and quickly can be used by anyone to read, calibrate and write out the WBD data. At this point the data are now ready for use by the scientific community. A scientist can either create plots or download WBD digital data files in text mode through the Cluster WBD website (http://www-pw.physics.uiowa.edu/cluster/) under the “Digital Data Download Tool” link or download the Level 1 files and accompanying software from one of the remote sites to their home site and use the supplied simplified software programs to read, calibrate and write out the WBD science data for analysis using their own analysis and plotting software. For a current listing of publications, including peer-reviewed journal and proceedings articles and graduate and undergraduate degree theses, in which WBD data were highlighted, please refer to the Cluster WBD website (http://www-pw.physics. uiowa.edu/cluster/publications.html). One important factor to keep in mind with regard to the WBD instrument is that it is dependent on real time reception on the ground by either DSN or PV. Except for a few periods during the Cluster mission, the spacecraft are too widely separated in space for any one antenna on the ground to pick up the signals from at least two spacecraft simultaneously (called MSPA, or Multi-spacecraft Per Aperture). Therefore, we are dependent on having at least one ground antenna available per Cluster spacecraft. Obtaining simultaneous WBD coverage from four spacecraft is thus somewhat limited and only possible at either the DSN Goldstone or Canberra complexes where at least four antennas are available and by prior negotiations, not in conflict with the needs of other missions. In the planning stage of WBD operations, science targets are defined based primarily on the needs of WBD and WEC. DSN then schedules reception of WBD data during some of these science target periods, providing, by agreement, about 2 h per spacecraft per 57-h orbit, or about 24 h per week. DSN provides as much simultaneous coverage from the four Cluster spacecraft as possible given the numerous constraints placed on WBD operation by ESOC and resources available at DSN within the context of its many other missions. PV has at most two available antennas, and usually captures 20 h of WBD data per week. Operations are greatly constrained at PV by the higher priority operations of the

11

Cluster Wideband Data Products in the Cluster Active Archive

173

Cluster spacecraft (uploading commands and downloading all other recorded data) when the ground stations they use are in view of the Cluster spacecraft. These constraints are also applicable to DSN reception of WBD data but are not as severe as at PV because DSN visibility of the Cluster spacecraft does not overlap as greatly with the ground stations used by ESA for Cluster. WBD data can be recorded onboard in a special burst mode (BM2) with either one-third the bandwidth or at one-third the duty cycle of the chosen WBD filter. This mode has only been used once for 1 h on March 7, 2001 because the Whisper instrument gets no data and all the other instruments have reduced telemetry rates. In the remainder of this paper, we first provide an example of WBD data and briefly discuss the types of studies for which WBD is ideally suited. Then we describe the documentation, software and Iowa-generated data plots archived at the CAA, followed by a brief description of some of the interpretation issues related to analyzing WBD data. We follow this with a discussion of our plans for eventually archiving the WBD data in cef format at the CAA as is done for all other Cluster instruments. We close with a brief description of the cross-calibration activities with respect to the WBD data.

11.2 WBD Data and Science Investigations WBD data are best suited for detailed analysis of the fine structure of linear and nonlinear waves observed throughout the Cluster orbit in support of: WEC scientific objectives by providing information about the fine structure of

the waves observed by the other WEC instruments WBD-specific objectives such as Very Long Baseline Interferometry (VLBI) and

cross-spacecraft waveform correlation These science objectives are discussed in more detail in the WBD instrument paper of Gurnett et al. [1] and in the Cluster WBD first results paper of Gurnett et al. [2]. An example of a two spacecraft WBD observation of fine structured triggered emissions (hooks) is shown in Fig. 11.2. In the spectrogram of Fig. 11.2a we see a series of seven prominent riser/hook combinations on both spacecraft, with those seen on spacecraft 2 being slightly before and more intense than those on spacecraft 4, suggesting that the source is closer to spacecraft 2. When we look at a short 10 ms section of the filtered waveforms in Fig. 11.2b, FIR filtered to retain only the 2–3 kHz components, taken near the top of the correlated hook pointed out in the spectrogram, we can easily see these same correlations with respect to the wave packets and the shape of their envelopes, which is quite remarkable. At this time the spacecraft are separated by about 150 km and located at 4:4 RE , 12ı Magnetic Latitude and 16.8 Magnetic Local Time. This example points out the beauty of the WBD instrument, i.e., the capability to resolve the fine structure of waves in both the time and frequency domains.

174

J.S. Pickett et al.

a

Cluster WBD 9.5 kHz

V2m-2Hz-1 10-11

4

3

SC2 Salsa

Ez Frequency [kHz]

Hook

2

10-12

10-13

1

0

10-14

4

10-11

SC4 Tango

Ez Frequency [kHz]

Hook 3

2

10-12

10-13

1 10-14

0 03:45:05

03:45:10

03:45:15

03:45:20

03:45:25

b Filtered Waveforms from Center of Hook

Fig. 11.2 Example of high frequency and time resolution WBD data showing several hooks, which are a triggered emission. (a) Spectrograms from two Cluster spacecraft showing the cross spacecraft correlation of hooks. (b) Filtered waveforms of a part of one hook showing the cross spacecraft correlation. Adapted from Pickett et al. [3]

11

Cluster Wideband Data Products in the Cluster Active Archive

175

11.3 WBD Products Currently Archived at the CAA WBD has already archived documentation, software and plots at the CAA for the first 4 years of the Cluster mission. Below we briefly describe each of these products.

11.3.1 Documentation The following general documents, with their format specified in brackets following the document name, have already been made available at the CAA. They are accessible through the “Documentation” link on the CAA home page by selecting “ICD and Publications” under the Cluster Instrumentation section. WBD ICD [pdf ], the WBD Interface Control Document, which supplies detailed

information about the WBD instrument and its various modes, references to other instrument documents, and contact information regarding the WBD investigation Listing of Coordinated Cluster WBD/PEACE Chorus Operations [html], which supplies the times when WBD and PEACE carried out coordinated observations of chorus emissions for the purpose of investigating the generation of chorus and wave-particle interactions Listing of Coordinated Cluster WBD/WHISPER/PEACE Nonthermal Continuum Operations [html], which supplies the time when WBD, Whisper and PEACE carried out coordinated operations for studying the generation and propagation of Nonthermal Continuum Listing of Other Special Operations (HAARP, IMAGE RPI Transmissions, Mutual Impedance tests) [html], which contains times when WBD carried out coordinated operations with HAARP (a ground facility located in Alaska for temporarily modifying a limited portion of the ionosphere), with the IMAGE spacecraft when the RPI instrument was transmitting at high frequency and the WBD instrument was observing the transmissions a few RE away, and with the Cluster WHISPER instrument, which was actively working in the mutual impedance mode Cluster WBD Interpretation Issues [pdf ], a document that details most of the known interferences and other non-natural features that might possibly be observed and misinterpreted in the WBD data (see Section 11.4 below) WBD Raw Telemetry to Calibrated Values Conversion [pdf ], a document that describes the calibration function for the WBD instrument

In addition, several documents that are included on every WBD Level 1 data DVD are available as follows: Instrument Description [txt] Description of the format of the Cluster WBD LEVEL1 data files contained on

the Level 1 DVDs [txt] Description of the contents of a Level 1 data DVD volume [txt] List of known problems and anomalies in the data [txt]

176

J.S. Pickett et al.

Fig. 11.3 Sample of a small section of a WBD “Coverage” listing (from file wbd mop. jan apr2002)

Description of the origin of time stamps, their accuracies, and their preferred

applications [txt] Finally, some very useful WBD documents for identifying WBD intervals for scientific study are the two types of listings of when WBD data have been taken, the WBD mode in which they were taken, etc. “Coverage” files in text format, which can be used to check if WBD data are

available at the time of a user’s interest and in the preferred mode, along with a document describing the format. File names contain the start and ending month followed by the year, e.g.: wbd mop.jan apr2002, where “mop” stands for mission operations phase. Note that listings for 2000 and 2001 contain all months. A sample from one of these files is shown in Fig. 11.3. “Science Summary” files in html format, which show times of WBD data coverage, instrument mode, science target (orbital region such as bow shock, magnetopause, etc. or wave emission) and information about active experiments (EDI, ASPOC) during the coverage period (i.e., if active and level of interference, if any). These files can be used to obtain a subset of WBD coverage times that fit specified investigation criteria. File names contain the two-digit year, followed by the inclusive months (as numbers), e.g.: scidatarec 02 1-3.html.

11.3.2 Software The following software and associated documentation are available through the “Team Software” section of the “Software” link, as well as from the

11

Cluster Wideband Data Products in the Cluster Active Archive

177

“Documentation” link on the CAA home page. All WBD software and associated documentation will be downloaded as a zip file. EXAMPLES Directory: EX README.TXT: Explanation of how to use the software CLOG.C: Example C code to output a list of all data modes contained in a Level1

file (if no mode changes, only one line is output) PARSETIME.C: C code used to parse the data (needs to be linked to

RD WBD WF.C) RD WBD WF.C: Example C code to read the Level1 data files and output the

calibrated data in binary format WBDL1LIST.C: Example C code to extract and list most fields contained in the

Cluster WBD Level 1 file records WBDL1.H: Contains macros, tables, etc. for use in processing Cluster WBD

Level1 data MAKEFILE: Converts to Unix-friendly format, compiles the programs, runs the

test programs, and outputs data 03082749.9C1: Test WBD Level1 data file WBDL1LIST03082749 9C1.TXT: Output listing of all field values in file CLOG03082749 9C1.TXT: Output listing of all data modes found R WBD WF0308271001.TXT: Output Data File (waveform data only)

CALIBRATE.C: C code to calibrate the WBD Level 1 data. Note that the calibration is applied in the time domain and is thus frequency independent. The electric field calibration assumes a physical antenna length of 88 m, whereas the effective length may in fact be less than that in some regions and at some frequencies. Units for electric field are mV/m. The magnetic field calibration is in nT. FIX SC2.C: C code to identify known bad data points in the time series and average across adjacent points for spacecraft 2 data. This problem originates within the Onboard Data Handling system and was discovered from data obtained while spacecraft 2 was on the launch pad and thus not possible to be fixed.

11.3.3 WBD Plots Two types of WBD plots serving different purposes have been generated at The University of Iowa and archived at the CAA. Overview Spectrograms: Used to identify events for in depth study and are generally not meant to be

published. Generally range from about 0.5–6 h (see example in Fig. 11.4). Show WBD spectral data (via FFT on ground) from all spacecraft being sampled

for the data overview interval (see full plot description in WBD ICD). Obtained by following the “CAA Quicklook Plots” link on the CAA home page,

selecting the Plot Type, CM CG WBD SPECPLOT OVERVIEW PNG, and entering a plot start time (the first plot with a start time later than the chosen start time will be displayed).

178

J.S. Pickett et al. Cluster WBD 9.5 kHz DSN

R

V2m-2Hz-1 -10

SC1 Rumba

6 4 2

SC2 Salsa

10 8 6 4 2

SC3 Samba

10 8 6 4 2 10 8

E MLAT MLT L

SC4 Tango

Ey Freq. [kHz] Ez Freq. [kHz] Ez Freq. [kHz] Ez Freq. [kHz]

NOT FOR PUBLICATION : Plot intended for browse purposes only since cyclic 42-second data intervals are dilated across 10-second data gaps throughout.

FFT Size 1024 (Overlap 0) 10 8

6 4 2 07:30

08:00

08:30

5.04 -38.42 21.55 8.20

4.68 -25.57 21.34 5.80

4.44 -10.74 21.14 4.60

09:00 4.35 5.44 20.95 4.40

09:30 4.42 21.78 20.76 5.10

10:00 4.64 37.09 20.58 7.30

10 -11 10 -12

10 -13 10 -14

10 -15 10 -16 10-10 10 -11 10 -12

10 -13 10 -14 10 -15

10 -16 10-10 10 -11 10 -12 10 -13 10 -14 10 -15 10 -16 10-10 10 -11 10 -12

10 -13 10 -14 10 -15 10 -16 10

10:30 4.99 50.60 20.40 12.40

UIowa 060227

UT_OBT: 2002-04-18T07:13:00 to 2002-04-18T10:36:00

Fig. 11.4 Sample WBD Overview spectrogram

Available as png files with the date, start time of the data and version provided in

the file name, e.g.: CM CG WBD SPECPLOT OVERVIEW PNG 20020418 1036 V00.PNG Postscript versions generated by Iowa and archived at the CAA will eventually be available from the CAA home page. CAA currently contains overview plots for 1 Feb 2001 thru 31 Jan 2005 (will be expanded as resources and time permit; see Section 11.5). High Time Resolution Spectrograms: Used to resolve the fine structure of waves. Cover a 30-s time span with full resolution (no averaging) (see example in

Fig. 11.5). Provide WBD gain and mode information, spectral data (via FFT on ground)

from only one spacecraft, start date and time, and ephemeris data (see full plot description in WBD ICD). Obtained by following the “CAA Quicklook Plots” link on the CAA home page, selecting the Plot Type, C CG WBD GIFPLOT, where is the number of the Cluster spacecraft of interest, ranging from 1 to 4, and entering a plot start time (the first plot with a start time later than the chosen start time will be displayed).

11

Cluster Wideband Data Products in the Cluster Active Archive

Gain (dB)

30

40

50

179 00

DATA MODE 80 60 40 20 0

80 60 40 20 0

DSN Filter DC ANTENNA Ez Bx By Ey RESOLUTION 8-bit 4-bit 1-bit

dB

–70

10

10 Frequency (kHz)

–80 –90 –100

5

–110

5

–120

2002 108 09:10: 30.000 (18 April) Cluster - C4

TRANSLATION 0 kHz 125 kHz 250 kHz 500 kHz

RE

4.4 MLAT 11.2 MLT 20.9 L 4.5

0

0 30

40

50

00

Time (sec)

Fig. 11.5 Sample WBD High Resolution Spectrogram showing the gain and instrument mode at the top, date and start time at right, and ephemeris data in the lower right

Available as gif files, with the date and start minute followed by the date and

end minute and version number contained in the file name (if the start and end minutes are identical, the plot covers 00 to 30 s of that minute, whereas plot names with different start and end minutes have plots starting at 30 s of the start minute and ending at 00 s of the end minute) e.g.: C4 CG WBD GIFPLOT 20020418 0910 20020418 0911 V00.gif CAA currently contains overview plots for 1 Feb 2001 thru 31 Jan 2005 (will be expanded as resources and time permit; see Section 11.5).

11.4 Interpretation Issues with WBD Data In certain regions of the orbit and during certain planned operations, interference from other instruments and/or spacecraft systems affects the WBD measurements. At these times one must use caution so as not to misinterpret the WBD data. In addition certain design features of the WBD hardware and software and of the electric antenna that is used by WBD make the WBD instrument (1) vulnerable to nonlinear effects in the form of harmonic generation under certain environmental conditions

180

J.S. Pickett et al.

and (2) susceptible to distorting waveforms when measuring specific types of waves, most notably solitary waves. In these cases, extreme caution must be used when analyzing the data. We have created an “Interpretation Issues” document that describes most of these interferences and instrument limitations and shows how they are manifested in the data through actual WBD data examples. This document is available for download through the CAA Documentation link (see Section 11.3.1 above). The list of issues currently covered is:

Electron Drift Instrument (EDI) interference Gain change effects within the WBD receiver Spacecraft battery interference Spacecraft power bus interference WHISPER Sounder interference DWP 900 Hz clock interference Periodic interference patterns of unknown origin, which are probably not naturally-occurring plasma emissions (see example in Fig. 11.6) Waveform distortion by the WBD receiver (see Section 11.6) Harmonic generation due to clipping (amplitude of received signal greater than that allowed by the 8 bit system, but receiver not in saturation) and due to amplitude saturation resulting in nonlinear effects Cluster WBD 9.5 kHz

9.5 9.0 11.0 10.5 10.0 9.5 9.0 11.0

10-16

10-16

10-17 10-15

10.0 9.5 9.0 11.0 10.5 10.0

09:35:00 5.47 -53.61 22.62 15.50

09:35:05 5.46 -53.58 22.62 15.48

09:35:10 5.46 -53.55 22.62 15.45

09:35:15 5.46 –53.53 22.62 15.43

09:35:20 5.46 -53.50 22.62 15.40

SC3 Samba

10-17 10-15

10.5

9.5 9.0 09:34:55 RE 5.47 MLAT-53.63 MLT 22.62 L 15.52

SC1 Rumba

10.0

10-16

SC4 Tango

Ez Freq. [kHz] Ey Freq. [kHz] Ez Freq. [kHz] Ez Freq. [kHz]

10.5

SC2 Salsa

V2m-2Hz-1 10-15

FFT Size 1024 (Overlap 0) 11.0

10-16

10-17 10-15

10-17 09:35:25 5.46 -53.47 22.62 15.38 UIowa 080912

UT_OBT : 2002-04-06T09:34:55 to 2002-04-06T09:35:25

Fig. 11.6 Sample Spectrogram of interference of unknown origin but probably not natural

11

Cluster Wideband Data Products in the Cluster Active Archive

181

11.5 Plans for Archiving WBD Digital Data Files Currently, it is not possible to download the WBD digital calibrated waveform data in cef format from the CAA. A link (“CLUSTER WBD Digital Data Download”) is provided under the WBD section of “Team Software” under the “Software” link from the CAA home page. This link will take the user to the Cluster WBD home page located at The University of Iowa. Data in text format will be made available for download. These download files currently provide a header containing the spacecraft name, start time and date of the data requested, description of the parameters contained in each column in the data which follow and a description of how to convert the data to the frequency domain via a standard FFT routine. The data columns provide the time since the defined start time, the calibrated data point, a flag indicating whether the data point is clipped (original raw value of 0 or 255) or not, the instrument mode (filter bandwidth, conversion frequency, gain, antenna selected and resolution), and the sampling frequency which is required for FFT analysis (different filter modes have different sampling frequencies; see Gurnett et al. [1]). NASA has just recently made available to The University of Iowa funds for WBD data archiving at NASA Goddard Space Flight Center’s CDAWeb (http://cdaweb. gsfc.nasa.gov). We envision that the archived WBD data files will each be 10 min in length and contain all of the parameters currently available from the Digital Data Download tool at Iowa discussed above, as well as auxiliary parameters such as the angle between the magnetic field and the antenna WBD is using to make the measurement, and the two angles defining the orientation of this antenna in a GSE coordinate system. Pursuant to standard cdf formatting, these files will also contain information about the data set, pertinent references, and information on how to reverse calibrate the data in order to analyze the raw data numbers (0 to 255) returned by the instrument. Once the cdf files have been generated they will be made available to the CAA for conversion to cef format and archiving at the CAA. We expect that the first files should be available to the CAA as early as May 2009.

11.6 Cross Calibration Activities WBD has participated in the following cross calibration activities:

11.6.1 Waveform Waveform cross calibration carried out with EFW, which identified some timing, sign (of E-field) and receiver response issues (see WBD report at the February 2006 Cross Calibration Workshop available through the CAA website). The timing and sign issues have since been corrected. With regard to WBD receiver response issues, this primarily concerns WBD response to pulses observed in the data. This is a

182

J.S. Pickett et al.

serious issue if one wishes to use the WBD data to study solitary waves or impulsive bursts of data. Some guidelines in this regard are provided in the document “WBD Response to Bipolar and Tripolar Pulses: Bench Tests vs. in Flight Observations”, by J. M. Swanner, J. S. Pickett, J. R. Phillips, and D. L. Kirchner, which is available in pdf format for download from the CAA website by following the “Documentation” link, then the “ICDs and Publications” link, to the WBD section.

11.6.2 Power Spectral Density Power Spectral Density cross calibration was carried out with WHISPER and STAFF-SA (see WHISPER report at the February 2006 Cross Calibration Workshop available through the CAA website). While the agreement with WHISPER was good, it was not good with regard to STAFF-SA in the overlapping frequency ranges. An investigation of this disagreement is ongoing.

11.6.3 Density Density cross calibration was carried out with PEACE and EFW of densities in low density regions, more specifically the magnetotail (e.g., see WBD reports at the February and May 2006 Cross Calibration Workshops available through the CAA). The method that WBD uses to obtain densities in these regions is to identify the L-X mode cutoff at L D 0 around a few kHz in time and frequency spectrograms. Provided the plasma frequency, fpe , is hgreater than the cyclotronifrequency, fce , this 1=2 frequency cutoff, FL , is equal to 1=2 fce C fce 2 C 4fpe 2 or approximately fpe – 1=2 fce . Thus, fpe FL C 1=2 fce , and ne .fpe =9/2 , where ne is the plasma density in cm3 and fpe is in kHz.

11.7 Summary All of the WBD documentation, software, overview and high time resolution plots, and digital data files necessary for any scientist anywhere in the world with internet access to the Cluster Active Archive to carry out independent science investigations are described above. We encourage such investigations and remain available to answer questions and/or participate in these investigations. The WBD observations, together with the observations from all of the other Cluster instruments, provide a unique observatory for answering many of the still outstanding questions with respect to the various boundary layers, the inner magnetosphere, magnetic reconnection, and substorms, as well as the auroral acceleration region from Cluster crossings of this region beginning in 2009.

11

Cluster Wideband Data Products in the Cluster Active Archive

183

Acknowledgements The authors would like to acknowledge support for this work from NASA Goddard Space Flight Center under Grant NNX07AI24G.

References 1. Gurnett DA, Huff, RL, Kirchner DL (1997) The Wide-band plasma wave investigation. Space Sci. Rev 79:195–208 2. Gurnett DA, Huff RL, Pickett JS, Persoon AM, Mutel RL, Christopher IW, Kletzing CA, Inan US, Martin WL, Bougeret J-L, Alleyne H St C, Yearby KH (2001) First results from the Cluster wideband plasma wave investigation. Annal Geophys 19:1259–1272 3. Pickett JS, Santol´ık O, Kahler SW, Masson A, Adrian ML, Gurnett DA, Bell TF, Laakso H., Parrot M, D´ecr´eau P, Fazakerley A, Cornilleau-Wehrlin N, Balogh A, Andr´e M (2004) Multipoint Cluster observations of VLF risers, fallers and hooks at and near the plasmapause. In: Sauvaud JA and N˘emeˇcek Z, (Eds), Multiscale processes in the Earth’s magnetosphere: From Interball to Cluster, Kluwer Academic Publishers, Netherlands, 307–328

Chapter 12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive J.G. Trotignon, P.M.E. D´ecr´eau, J.L. Rauch, X. Valli`eres, A. Rochel, S. Kougbl´enou, G. Lointier, G. Facsk´o, P. Canu, F. Darrouzet, and A. Masson

Abstract The Waves of HIgh frequency and Sounder for Probing of Electron density by Relaxation (WHISPER) instrument is part of the Wave Experiment Consortium (WEC) of the CLUSTER mission. With the help of the long double sphere antennae of the Electric Field and Wave (EFW) instrument and the Digital Wave Processor (DWP), it delivers active (sounding) and natural (transmitter off) electric field spectra, respectively from 4 to 82 kHz, and from 2 to 80 kHz. These frequency ranges have been chosen to include the electron plasma frequency, which is closely related to the total electron density, in most of the regions encountered by the CLUSTER spacecraft. Presented here is an overview of the WHISPER data products available in the CLUSTER Active Archive (CAA). The instrument and its performance are first recalled. The way the WHISPER products are obtained is then described, with particular attention being paid to the density determination. Both sounding and natural measurements are commonly used in this process, which depends on the ambient plasma regime. This is illustrated using drawings similar to the Bryant plots commonly used in the CLUSTER master science plan. These give a clear overview of typical density values and the parts of the orbits where they are obtained. More information on the applied software or on the quality/reliability of the density determination can also be highlighted. J.G. Trotignon (), P.M.E. D´ecr´eau, J.L. Rauch, X. Valli`eres, A. Rochel, S. Kougbl´enou, G. Lointier, and G. Facsk´o Laboratoire de Physique et Chimie de l’Environnement et de l’Espace, Universit´e d’Orl´eans, Orl´eans, France e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected] P. Canu Laboratoire de Physique des Plasmas, V´elizy, France e-mail: [email protected] F. Darrouzet Institut d’A´eronomie Spatiale de Belgique, Bruxelles, Belgique e-mail: [email protected] A. Masson RSSD, ESA, Noordwijk, The Netherlands e-mail: [email protected] H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 12, c Springer Science+Business Media B.V. 2010

185

186

J.G. Trotignon et al.

12.1 Introduction The CLUSTER Active Archive (CAA) has been created to house data delivered from all of the CLUSTER instruments and to make these data available to the wide scientific community [18] In this context, the WHISPER (Waves of HIgh frequency and Sounder for Probing of Electron density by Relaxation) team provides two principal types of data: the total electron density at standard time resolutions of either 2.15 or 52 s, and electric field spectra in the frequency range 2–82 with time resolution 1.7 or 3.4 s. The total wave energy in the range 2–80 kHz is also provided with higher time resolutions. The density can be deduced from the characteristics of natural waves monitored whenever WHISPER is in natural mode (transmitter off) and/or from resonances triggered in the sounding mode. After a brief description of the instrument design and operation (Section 12.2) the data products ingested into CAA (Section 12.3) and some of the cross calibration activities (Section 12.5) are presented. The way the electron density is determined is then recalled (Section 12.4) with, as far as possible, a particular attention paid to the reliability, accuracy and degree of confidence (Sections 12.4 and 12.6).

12.2 WHISPER Instrument Design and Performance The WHISPER experiment results from a collaboration between experimenters organised within the context of the Wave Experiment Consortium, WEC, which includes five instruments ([15, 26] esp. Sects. 1 and 4). The WHISPER instrument [6,7] consists basically of a receiver, a transmitter, and a wave spectrum analyser, associated with parts of two other WEC instruments: the sensors of the Electric Field and Wave, EFW, experiment and data processing functions of the Digital Wave Processing, DWP, experiment. WHISPER is devoted to the survey, active and natural, of the WEC electric signal in the frequency range from about 2–82 kHz, which includes electrostatic and electromagnetic natural emissions of interest to the CLUSTER objectives, in particular in the vicinity of the plasma frequency from which the total electron density may be determined. Let us note that one of the main functions of the WHISPER experiment is to provide reference values of this fundamental plasma parameter. Two modes of operation are used alternatively. The natural wave mode, when the transmitter stays on stand-by, and the sounding mode during which the transmitter triggers plasma resonances prior to reception. In both cases, a pair of EFW sensors is used as a double-sphere electric dipole antenna, whose potential difference is band-pass filtered, digitised, multiplied by a windowing function (to make the signal periodic), and finally analysed in frequency by an onboard FFT processor which computes a full frequency spectrum every 13.33 ms (40 ms/3) in normal telemetry operation (normal mode, NM), and in most burst mode (BM) usually every 26.6 or 40 ms (two or three times the basic WHISPER recurrence period of 13.33 ms). The spectrum covers the 0–83 kHz band in 512 bins or 256 bins, with a frequency

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

187

resolution of 162.8 Hz (Fech= 1024 with the sampling frequency Fech = 1 MHz/6) in the 512-bin FFT option or 325.5 Hz in the 256-bin FFT option. The modulus of the complex amplitude in each bin is calculated, while the associated phase information is discarded. These spectral amplitude (moduli) are used for all subsequent onboard calculations. This choice of amplitude (rather than power) may appear curious, but there was a strong technical constraint: the dynamic range of the spectral power exceeds the dynamic range (limited by the word length) of the processors used by WHISPER, and also by DWP which, in some modes of operation, performs processing for WHISPER. The difference in the results obtained by summing the squares and summing the amplitudes is not large (less than 2 dB) for a wide-sense-stationary random process. A few bins are not transmitted to the CLUSTER telemetry (due to limited TM resource and input signal filtering), which explains why the 0–83 kHz band is not fully covered in the WHISPER data (as the WHISPER spectrum data size is always 512 in CAA, the missing bins are set to the -1.0 fill value). In natural wave mode, successive spectra are accumulated onboard over intervals of 0.213 s (16 times 13.33 ms) to 0.853 s (64 times 13.33 ms) when in NM mode and of 0.107 s (8 times 13.33 ms) to 0.213 s (16 times 13.33 ms) when in BM mode, thus smoothing sporadic features and improving the signal to noise ratio. Moduli of the same frequency are thus summed onboard while averages are computed on ground. Each accumulated modulus is compressed quasi-logarithmically to a 6-bit word (or 8-bit word). At regular time intervals, one of these accumulated spectra is selected and transmitted to ground: between one out of ten and one out of two in NM mode, resulting in a time resolution in the range 1.7–3.4 s; and one out of three in BM mode, resulting in a time resolution of 0.320 or 0.640 s. In all natural telemetry modes, the measured integrated power in the band 2–80 kHz, prior to windowing and FFT processings, is accumulated over each time acquisition, forming a quantity called “Energy”, available every 13.33 ms. In one option, available in BM telemetry, time series of “Energy” are transmitted to ground at full time resolution (13.33 ms), after a quasi-logarithmic compression of the “Energy” values to 8 bits (3 bits for the mantissa and 5 bits for the exponent; mantissa and exponent are described in Section 12.3.1). Otherwise, these “Energy” values are accumulated aboard over the same time interval as for the onboard accumulated-spectra. These values, also quasi-logarithmically compressed to 8 bits, are transmitted to the ground without further selection, i.e. with a time resolution better than that of the transmitted-to-ground accumulated spectra, by a factor between two and ten in NM telemetry mode and up to 48 in BM. In sounding mode, a signal acquisition is performed a few milliseconds (4.9 or 5.2 ms) after transmission of a sounding pulse, the listening duration is at least 6.14 ms, which is compatible with a sampling frequency of 166.66 kHz and a 512-bin FFT. The sounding pulse is a short sinusoidal wave train whose frequency sweeps the 4–82 kHz band, or less depending on the operating mode, in steps of 976.6 Hz (1 MHz/1024) when the wave train duration is 1.024 ms and of 1953.1 Hz when the wave train duration is 0.512 ms. The spectral width of the emitted pulse is the inverse of its duration, and consequently the pulse can trigger all the characteristic frequencies of the medium within a band of 1 or 2 kHz around the pulse

188

J.G. Trotignon et al.

frequency. The high spectral resolution of the FFT receiver allows the triggered frequencies to be identified as resonances, i.e. as very intense echoes which are strongly monochromatic and correspond to frequencies for which the group velocity is very small and close to the spacecraft velocity [3, 10]. In natural wave mode, the measured antenna potential difference is processed by FFT to compute a full frequency spectrum every 13.33 ms, while in sounding mode this computation is performed after every emission of a sounding pulse, at intervals equal to the step duration, 26.6 or 40 ms in most cases (two or three times the basic WHISPER recurrence period of 13.33 ms). As the telemetry resource does not allow all the onboard computed spectra to be downloaded the following compression strategy has been applied. It consists in selecting, in each full (0–83 kHz) spectrum associated to each sounding wave train, only the bins whose frequency is inside the frequency band (1 or 2 kHz) of the emitted pulse. Those selected frequency bins contribute to a global “active/sounding” spectrum, obtained once the 4–82 kHz frequency sweep has been completed, which is downloaded. Let us note that 80 frequency steps of 976.6 Hz frequency bandwidth are for instance necessary to cover the full WHISPER frequency range. This corresponds to sweep duration of 3.2 s (80 40 ms) in the simplest mode, available as a back up operation option fully managed by the WHISPER electronic module. This back up option has been used on C4 CLUSTER satellite since September 2003, when part of the onboard WHISPER memory failed, probably due to severe environmental conditions. In other operational modes, DWP performs much of the WHISPER data processing. The most frequently used option has the emitter stepping by 976.6 Hz (pulse duration 1.024 ms) in the lower part of the frequency range, and by 1,953.1 Hz (pulses of 0.512 ms) at higher frequencies. With a step duration of 26.6 ms at all frequencies, the overall sweep duration is less than 1.5 s, which permits two sweeps within 3 s of a total sounding time interval. Other modes are available, with sweep duration between 0.5 and 10 s. Active spectrum values (moduli) are compressed quasi-logarithmically to 8-bit words before being transmitted to ground. In addition to the frequency bins retained in each frequency step to construct a sounding spectrum (usually 6 bins, 162.8 Hz in bandwidth), bins whose frequency is far away from the sounding frequencies may also be retained as part of a ‘passive’ spectrum. As a consequence, whenever a sounding sweep is complete, both a sounding spectrum and a passive spectrum may be available. In addition, provided the retained passive bins (again usually 6 bins, 162.8 Hz in bandwidth) are those whose frequencies are the same as the ones that will be covered, by the next transmitting pulse frequency, i.e. in the next frequency step of the current weep, the time delay, for a given frequency, between passive and sounding measurements becomes very short (less than 40 ms). The WHISPER relaxation sounder has been designed in order to avoid, or at least minimize, any corruption between natural and sounding measurements, and also between sounding measurements. In particular, transmitted frequencies are swept in appropriate orders, described in tables chosen by command, in such a way that several consecutive transmitting pulses never have central frequencies close together. Passive measurements associated with sounding measurements are therefore not

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

189

influenced by long lasting resonances and no perturbations are expected between two successive sounding steps. Analysis on the ground of the resonance pattern observed in active spectra (plus comparison with the associated passive spectra when available) allows the identification of characteristic frequencies of the surrounding plasma (see Section 12.4.1 and 12.4.2), in particular the plasma frequency Fpe , and hence the total electron density Ne D Fpe2 =’ where ’ is a constant: ’ D e 2= .4 2 ©o m/ D 80:7 Hz2 =cm3 . The sounding mode is operated alternately with the natural wave mode, with all four spacecraft following closely the same time line. Typically, 3 s (or 4 s) of sounding mode (two sweeps in frequency) are followed by 49 s (or 100 s) of natural wave mode, leading to a standard 52 s (or 104 s) sounding recurrence time. After September 2003, C4 CLUSTER spacecraft operates always at a 104 s sounding recurrence. In the standard (most frequently used) sounding cycle, the four spacecraft sound simultaneously every 104 s, and C1, C2 and C3 also sound 52 s later, but C4 does not transmit, it continues to monitor natural waves. Such a duty cycle is used for all WEC operations, both in nominal (NM) telemetry mode and in burst (BM) mode, when the time resolution of natural wave measurements is, on average, improved by a factor 5. WHISPER data are usually presented in the form of dynamic spectrograms, in which the electric-field intensity is plotted as a function of time (abscissa) and frequency (ordinate) using a colour code. The spectrograms bear important information about explored regions. The characteristic signature of natural or actively triggered waves indicates the nature of the ambient plasma regime and, combined with the spacecraft position, reveals the position of key magnetospheric boundaries encountered during a specific time interval. Fig. 12.1 shows typical signatures encountered by CLUSTER. The yellow ellipses indicate natural emissions: (a) faint escaping non-thermal continuum radio waves, (b) continuum enhancement [8], (c) Type III solar radio burst and (d) magnetosheath low frequency turbulence. White arrows show characteristic resonances triggered by the sounder: (1) the plasma frequency WHISPER 3, 9 November 2002

1 a

c

LPCE/CNRS, Orléans

3

2 b

d

Fig. 12.1 One day E-field spectrogram showing typical signatures of natural waves and resonances seen by the CLUSTER/WHISPER relaxation sounder (see text)

190

J.G. Trotignon et al.

at the plasmapause boundary, (2) the electron gyrofrequency and its harmonics and (3) the plasma frequency in the magnetosheath. The plasmapause is recognized by the plasma frequency increasing as the spacecraft enters the plasmasphere near 00.50 UT, then decreasing near 01:50 UT as it exits. The magnetopause position (crossed at 18:50 UT) is indicated by a rapid change in both the plasma frequency and the level of low frequency turbulence.

12.3 WHISPER Data Products in CAA This section describes the WHISPER data products to a level adequate to be able to understand and start using them. Further information may be found in the CAA Interface Control Document for WHISPER [13], and other documents available on the CAA site. All WHISPER data products are validated at LPC2E (Laboratoire de Physique et Chimie de l’Environnement et de l’Espace Orl´eans, France) before being delivered to the archive as CEF files (CLUSTER Exchange Format). Eleven WHISPER datasets are archived at CAA. They fall into four categories: natural waves (one science and one support dataset), energy (two science datasets), electron number density (one science dataset), and sounding (three science and three support datasets), as explained in the following sections.

12.3.1 Natural Waves Related Datasets Most of the time WHISPER observes natural waves in the (2, 80 kHz) band, using one of the two EFW antenna pairs: Ey or Ez. Frequency spectra are obtained by onboard FFT (generally in 512 or 256 frequency bins, but 64 and 128 bins are also possible) with a nominal resolution of 162.8 Hz in 512 bins. Only accumulated bin moduli are transferred to the telemetry. Three gains are available: 12, 24, 36 dB. The 24/12 and 36/24 modes switch automatically from high gain to low gain whenever too many overflows are encountered, overflows being signal samples exceeding the dynamical range of the analog-to-digital converter. Natural wave spectra cannot all be transferred due to TM limitations: compression and selection are therefore required. A number of successive spectra called “average number”, are accumulated bin by bin, and accumulated spectra are transmitted at a rate selected to optimize the use of the TM rate available at the time. Finally, spectral values are compressed logarithmically from 24 bits (DWP word size) to 8-bit or 6-bit words. There are two types of natural wave related datasets: CP WHI NATURAL for natural (accumulated) spectra and CT WHI NATURAL EVENT indicating processing mode changes. Only the first is briefly presented below. The second is an event table, which contains engineering parameters needed to fully understand the associated physical data: for a description please refer to the ICD (2009).

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

191

CP WHI NATURAL files contain: the spectrum (central) time; the time uncertainty which is due to accumulation, in s; the FFT size; the EFW receiving antenna actually used; the electric-field spectral power densities computed on ground from accumulated moduli, expressed in V2 m2 Hz1 ; the FFT bin frequencies, given in kHz; an overflow coded value which can vary between 0 (no saturation, i.e. no time samples out of limits before A/D conversion) and 1 (all time samples are out of limits, thus clipped during conversion); the average number, i.e. the number of accumulated spectra; the number of gain changes during acquisition; and the mantissa and exponent sizes of compressed data (usually adjusted to the observed signal dynamical range), which indicate the uncertainty in spectral line moduli due to onboard compression. It is important to recall that electric-field spectral power densities are not quantities directly delivered by the instrument, but are elaborated on ground. We will now follow each step of the conversion from TM words to physical quantities, in order for the user to appreciate the precision and reliability of the data. The parameter measured onboard is the potential difference between the two probes used to form the dipole antenna, which is then analysed spectrally by FFT of successive segments of the waveform, the moduli of the complex Fourier transform being accumulated bin by bin to obtain the values transmitted to the ground. A telemetry word (of 6 or 8 bits) allocated to a bin is divided into two parts, the exponent and the mantissa. Their respective size (for instance 4 bits for the exponent and 2 bits for the mantissa) is the same for all words in the spectrum. Exponent indicates the dynamical range of a bin modulus in a given spectrum. For instance a value of 4 indicates that the number of bit shifts from the largest to the smallest values need 4 bits to be coded, meaning that all values are within a 96 dB range, which accommodates the largest dynamical range possibly encountered in the WHISPER instrument. The mantissa bit size gives the number of significant bits allocated to each bin modulus, thus the uncertainty; for instance a value of 2 yields an uncertainty below 25%. When the dynamical range of bin moduli is small, the exponent size is small (for instance, 2 bits for a range of 18 dB) and the mantissa size allows a high precision (a 4 bits mantissa leads to an uncertainty below 6%). The shift value common to all bins, i.e. the position of largest shift, is indicated via the bin size information (exponent and mantissa sizes) present in the data record of each spectrum. The receiver transfer function (gain in modulus versus working frequency) is known with a similar uncertainty, of order 1 dB, to what is due to onboard compression. The transfer function has been measured on ground with great detail, on each spacecraft and in various environmental conditions. A possible evolution of this quantity exceeding tolerable thresholds would immediately be known to the WHISPER operation team, via results of onboard calibrations activities, run every orbit. Nothing of the kind happened since the start of CLUSTER mission. The same calibration functions have thus been applied on ground since then, translating telemetry words to physical quantities. The total uncertainty attached to the electric field spectral modulus is estimated to be about 2 dB. To determine the electric field, i.e. the physical quantity of interest, the potential difference must be divided by the effective (electrical) length of the antenna. The latter has been assumed to be equal to 100 m tip-to-tip which is the rounded

192

J.G. Trotignon et al.

geometrical length (actually 88 m). While the geometrical length is used widely to convert potential differences to electric-field values, it is only a crude approximation of the effective length. It may be indeed significantly wrong close to characteristic frequencies of the ambient plasma, and possibly too in conditions, natural or instrumental (EFW), which produce an unusual plasma sheath around the antenna. This point is discussed further in Section 12.6. The final operation transforms the electric field spectral moduli to the equivalent spectral power density, while simultaneously taking account of the shape of the Blackman-Harris windowing function applied prior to FFT analysis.

12.3.2 Energy Related Datasets In each natural spectrum acquisition interval, the large band (2–80 kHz) electric potential is sampled over 512 or 1024 time samples (depending on FFT size) which are squared, summed, and transmitted. A compression from 48 to 8 bits is applied onboard, with fixed bit allocations for the exponent and for the mantissa. The dynamical range of the energy parameter, corresponding to the ratio between the largest and smallest measurable values of large band electric-field power, is only 48 dB, due to instrumental limits. The energy parameter is thus especially interesting in case of bursty signals of large amplitudes, and when the selection rate of transmitted spectra is severe. In such cases, it allows to survey the variability of the electric field during the time intervals when spectra are not transmitted. There are two datasets related to ‘energy’ like parameters: CP WHI WAVE FORM ENERGY; and PP WHI. The latter gives, together with an extremely crude estimate of density, the energy in three wide frequency bands (2–10, 10–20, and 20–80 kHz), calculated on ground from available spectra, and the normalized standard deviation of large band electric field power from time samples, calculated on ground.

12.3.3 Sounding Related Datasets As for the natural wave spectra described in Section 12.3.1 not all the active/sounding spectra can be transferred due to TM limitations: compression and selection are thus required. Different strategies can be selected such as: compression of the values (to 8-bit or 6-bit words); reduction of the frequency sweep range; active bins grouped by pairs, with only the greater of each pair being kept, and the missing spectral values being filled by linear interpolation during ground processing prior to archiving; active-to-passive ratios coded on 2 bits and transmitted rather than sending passive data. As described in Section 12.2, whenever a frequency sweep is complete, both an active spectrum and a passive spectrum are available. They are constructed with frequency bins selected in each of the frequency steps that compose the sweep, so that

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

193

the time difference, for a given frequency, between a passive bin and an active bin is less than 40 ms. Let us recall that passive spectra have a frequency/time distribution which is significantly different from that obtained in the natural wave modes described in Section 12.3.1. The constructed ‘passive’ spectrum associated with the ‘active’ spectrum can be sent to ground or used to compute active-to-passive ratios that can also be sent to ground, depending on the chosen compression strategy. Comparison of ‘passive’ and ‘active’ conditions at a given frequency during a single frequency sweep is possible because the succession of the frequency steps in a sweep are ranged in such an order as ‘passive’ measurements cannot be perturbed by previous soundings (see Section 12.2). This is a great advantage compared to previous sounder concepts, because the time delay between these passive measurements and the related active ones is here very short (a few tens ms instead of a few s). The 2-bit coding of the active-to-passive intensity ratio is as follows: if (active/2 passive) return 0; else if (active/4 passive return 1; else if (active/16 passive return 2; else return 3). The types of sounding related archives are six in number: CP WHI ACTIVE including sounding (active) spectra CP WHI PASSIVE ACTIVE for associated passive spectra CP WHI ACTIVE TO PASSIVE RATIO, which contains the active to passive ratios (with only four possible values for each bin) CT WHI ACTIVE EVENT to list sounding mode changes CP WHI HK, which gives elementary sounding events extracted from the WEC housekeeping files CP WHI SOUNDING TIMES, which is a simplified version of the above mentioned CP WHI ACTIVE file where only the sweep central sounding time is kept The CP WHI ACTIVE content is recalled below, for the other files please refer to the ICD for WHISPER document (2009). In the CP WHI ACTIVE file, we find the following information: spectrum (central) time; time uncertainty due to acquisition and sweep, in s; the EFW receiving antenna used; the active electric-field spectral power densities, in V2 m2 Hz1 ; FFT bin frequencies, in kHz; scanned frequency table, which gives the transmitted frequencies and the sweep order; sounding pulse length, in ms; time between 2 successive emissions in a sweep; emission/reception delay i.e. the time between emission and reception; number of gain changes; and compression parameters i.e. the mantissa and exponent sizes.

12.3.4 Electron Density Related Dataset Compared with the three previous categories of WHISPER data products, created at CNES, validated at LPC2E, and directly delivered by CNES to CAA, the CP WHI ELECTRON DENSITY file is not a product which can be generated by routine processing (decommutation and conversion into physical units) of WHISPER data. As explained in Section 12.4, the electron density is determined from either the characteristic plasma frequencies at which resonances are triggered when

194

J.G. Trotignon et al.

WHISPER is in sounding mode, or from natural wave signatures, or a combination of both [Chapter 18 of this book, 22, 24]. Density files result from complex procedures which require human intervention and scientific expertise, and therefore they are produced at the laboratory and delivered to CAA on a best effort basis. Are included in the CP WHI ELECTRON DENSITY file: the central time of the spectrum from which the density is derived; the uncertainty time, in s; the spectrum type (Active or Natural); the computational method, i.e. the algorithm and code used; information on the external data used, for example E whenever EFW data are used; the derived electron density value, expressed in cm3 ; the electron density uncertainty, expressed in cm3 , if available (depends on the applied algorithm); a “contrast” factor, between 0 and 1, if available (based on the local contrast of the peak/cut-off, as we will see later).

12.4 Electron Density Determination Process The electron density is determined semi-automatically with the help of a specific tool called FPTool whose working principle is shown in Fig. 12.2. The first step is to visualize input parameters from the four spacecraft, such as the WHISPER active/passive spectrograms, the EFW spacecraft potential and FGM magnetic-field strength. The user is then invited to select a frequency band that includes the plasma frequency signature, a resonance or a clear cutoff, from which the density will be derived [25]. Knowledge of the region (plasma regime) where the spacecraft is located is also highly desirable for a reliable determination; this information may come from the ESOC Predicted Event Times or the dedicated software (MAGLIB) developed at CNES. The next step is to choose and apply the appropriate algorithm (FPTool box in Fig. 12.2) to look at the electron density file (Fp file box in Fig. 12.2) that

Binary WHI Data

FPTool (interactive)

Config. files

FP Computation (automatic)

EFW Archives/PP (Preview) FGM Archives/PP

EDI CRF files

Fp files Archive Generation (automatic)

CAA Act. & Nat. files

CAA Ne files

Fig. 12.2 FPTool flow chart describing how the CAA WHISPER density files are produced

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

195

is produced and, if necessary (Preview box in Fig. 12.2), remove erroneous values or select another algorithm. The loop indicated in Fig. 12.2 (FPTool – FP Computation – Preview) is driven until a satisfactory result is obtained. This process has been conceived to minimize large errors in the density estimate, to confirm that the most appropriate algorithm has been applied, and to minimize the time necessary to validate products automatically created.

12.4.1 Density Determination in the Solar Wind and Magnetosheath Only one strong resonance is usually observed at, or close to, the plasma frequency Fpe from which the density Ne is directly derived (see also Section 12.2): Ne Œcm3 D Fpe2 ŒkHz =80:7 In addition, lower cut-off of thermal noise recorded whenever WHISPER is used as a natural wave detector is also a valuable way to determine Ne . Cut-off indeed occurs very close to Fpe . Another way to derive Ne is to combine WHISPER active observations with S/C potential measurements made by EFW and to adjust, at best, the two time series, interval by interval. Pattern recognition processes are therefore applied. Thus, the density (plasma frequency) determination process may be decomposed into three steps: a preprocessing of active spectrograms; a reliable identification of plasma-frequency resonances on active spectra (only peaks with high signal-tonoise ratios are kept); and finally a density extraction from active and/or passive spectra in a frequency band centered on the general trend of the plasma frequency variations. This trend is defined using the previously determined resonance frequencies and interpolation frequencies coming from any cut-off signature and/or a recalibration of the EFW S/C potential. It is worth noting that some of the resonances that were first rejected due to their low signal-to-noise ratios may be finally reselected provided they are in the search frequency band. The two last steps are illustrated in Fig. 12.3. The active spectrogram pre-processing consists in removing the interference signals (perturbations from EDI, spurious signals from the onboard power converter, or DSP bug in the FFT transform for WHISPER). These signals are detected by converting active spectrograms into “peak images” where only spectral peaks are kept [19]. A morphological “opening” operator is applied on the obtained binary images along the “time” dimension. Whenever three peaks are at least located at the same frequency in five contiguous spectra, they are considered as interference signatures, and the five values related to this frequency are then removed and replaced by interpolated values. How to compute the plasma frequency from active spectra? Fpe is usually characterized by a strong resonance. It is therefore given by the tallest peak in each

196

J.G. Trotignon et al.

WHISPER 1, 06 June 2002, Earth’s Magnetosheath

Frequency (kHz)

80 60 40 20

Frequency (kHz)

80 60 40 20

05:00

05:20

05:40

06:00

05:00

05:20

05:40

06:00 UT

Fig. 12.3 Density determination in magnetosheath: top-left panel, WHISPER active spectrogram; bottom-left panel, identified plasma frequency resonances (white full circles) superimposed on active spectrogram; top-right panel, plasma frequency derived from the EFW S/C potential (white line) superimposed on WHISPER natural/passive spectrogram; bottom-right panel plasma frequency (white line) deduced from natural wave lower cut-off superimposed on passive spectrogram

spectrum, provided its intensity (in dB) is at least 0.85 times higher than the intensity (in dB) of the second tallest peak. The Fpe uncertainty is nominally 163 Hz, the best WHISPER frequency resolution. The values obtained are shown in the lower left panel of Fig. 12.3. The “contrast” factor, which is also given in the dataset, reflects the local contrast of the peak. Let P be the tallest peak intensity relative to the background level and M the mean intensity in the 20 frequency bins around the peak, the contrast factor C is simply given by: C D min.P = M 1; 1/ The last step in this example is the use of cut-off signatures and/or EFW S/C potential recalibration. A rough approximation of Fpe may indeed be derived from the S/C potential SP (which is negative) using the empirical relation: Fpe D 9.˛.SP /ˇ /1= 2 ; with, by default, ˛ D 200 and ˇ D –1:85 (if necessary, these values can be adjusted by the operator during the validation process). The resultant Fpe curve is then fitted to the above mentioned extracted resonances by a simple affine transformation. There are now two possibilities: whenever a clear cut-off arises in natural/passive

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

197

spectra, its frequency is automatically extracted inside a search band (about 30 frequency bins in width) defined around a Fpe curve that links up the already extracted resonances or the Fpe curve coming from the S/C potential recalibration; conversely, if no natural signature is present at all, the EFW fit itself is given as the plasma frequency. Let us note that the contrast factor is set to the fill value of 1 when the density value is obtained from EFW data. Otherwise, the contrast is an indicator of the contrast of the low cut-off range in the search band. Let M , U , and L be the mean signal magnitudes in respectively, the whole search band, the upper part of the search band (from the cut-off frequency bin returned by the algorithm to the upper limit of the search band), and the lower part of the search band (from the lower limit of the search band to the cut-off frequency bin returned by the algorithm), the contrast factor C is given by: C D max.min..U L/ = M; 1/; 0/ Finally, it may be noted that the cut-off seen in the natural wave spectrograms of Fig. 12.3 is that of the continuum which differs from the thermal noise that has a much narrower bandwidth, within which there is no structure. The fact that the cut-off is close to the plasma frequency can be confirmed here by comparison with the plasma frequency resonances shown in the bottom left panel, this is a great advantage compared with only passive experiments: natural waves may be indeed generated at a considerable distance from the spacecraft and their frequency signatures, such as cut-offs, may thus not reflect local plasma characteristics. Let us also note that in the solar wind and magnetosheath the electron gyrofrequency Fce is too low compared with the plasma frequency Fpe for the upper-hybrid frequency Fuh (if actually triggered) to be distinguished.

12.4.2 Density Determination in the Plasmasphere, Cusp and Tail In the Earth’s magnetosphere, and in particular in the plasmasphere, resonances arise at the electron cyclotron frequency Fce , its harmonics nFce , the electron plasma frequency Fpe , the upper-hybrid frequency Fuh .Fuh D .Fpe2 C Fce2 /1=2 /, and Fqn that correspond to Bernstein electrostatic cyclotron waves with group velocity close to zero .Fqn > 2 Fce and nFce < Fqn < .n C 1/Fce /. A detailed plasma diagnosis can therefore be made from the identification of these resonances, which depend on the magnetic field strength and the total electron density. Two populations of electrons, cold and hot may also be revealed [23]; nevertheless this kind of analysis is too intricate and time consuming to be performed automatically. Consequently, we decided to take advantage of the higher spectral energy density close to Fuh , from which the plasma frequency (and thus the electron density) may be directly derived. It has been indeed established from many observations in the magnetosphere (on ISEE-1 GEOS 1 & 2, Viking) that the resonances, including the n Fce series, are more intense close, and above, the plasma frequency and that the spectral energy

198

J.G. Trotignon et al.

density reached its maximum very close to Fuh [11,21]. The magnetic field strength, and hence Fce , are known either from FGM or from nFce observed by WHISPER (Fce D 28 B, where Fce is expressed in Hz and B in nT). Active spectrograms are first pre-processed to enhance the contrast of resonances: a morphological top-hat operator [20] is first applied to each spectrum and spectra are not normalised [19]. In other words, as usual in image processing the objective is to reduce the noise and to detrend the image while preserving edges in order to highlight peaks. Spectra are then smoothed (Butterworth filter) and the peak frequency of each smoothed spectrum is retained as Fuh , the uncertainty being the half-width at the half-height of smoothed peaks. Let also note that only maxima with intensities at least five times higher than the around background level observed in close natural/passive spectra (10 frequency bins around) are retained. Are also discarded spectra where two or more peaks are observed with similar intensities. The kept points are indicated in green and those who are rejected in black, in the bottom-left panel of Fig. 12.4 The next step is the EFW S/C potential recalibration on the “green” points as described in Section 12.4.1. A search band may then be defined around the obtained Fuh values (the upper and lower limits of this frequency band, shown as red lines in the bottom-left panel of Fig. 12.4, are respectively 0.5 and 1.5 times these values derived from the recalibrated potential). Finally, a new WHISPER 1, 12 January 2003, Earth’s Plasmasphere 6

80

40

40

3

0

0

20

Arbitrary Unit

60

dB

Frequency(kHz)

80

80 60 40

40

dB

Frequency(kHz)

80

20 0 04:00

05:00

06:00

07:00 04:00

05:00

06:00

07:00 UT

Fig. 12.4 Upper-hybrid frequency Fuh determination in plasmasphere: top-left panel, WHISPER active spectrogram; top-right panel, reprocessed active spectrogram to bring out resonances; bottom-left panel, search band (red lines) around the Fuh calculated from the re-calibrated EFW density estimate, Fuh derived from WHISPER resonances (points kept during the first recognition step in green, rejected points in black, new points obtained during the last step of the recognition process in blue); bottom-right panel, final set of identified Fuh (white line) superimposed on active spectrogram

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive Y Y

Signatures

N

Y

Resonances

Cutoff

N EFW

No Data

Upper cut-off

Active Spectra

N

L

Y

EFW Resonances & EFW filtering

Signatures

N U

199

Lower cut-off around EFW

Y

N EFW

Lower cut-off

Recalibrated Potential

No Data

Natural Spectra

Fig. 12.5 Flow charts of plasma frequency/density recognition algorithms that are applied to WHISPER active (on the right) and natural/passive (on the left) E-field spectra, depending on the availability of EFW S/C potential data and upper or lower cut-off signatures of natural waves

search of spectrum maxima is made in this frequency band, which gives the final set of Fuh values: they are represented in green and blue. The last step of processing calculates Fpe from the pair Fuh and Fce , and finally the density. Figure 12.5 shows the different strategies that can be used to determine the electron density, depending upon the available inputs: resonances, natural wave cut-off signatures, and EFW spacecraft potential measurements It applies to all plasma regimes.

12.5 Cross-calibration Activities WHISPER determines the total electron density from the propagation characteristics of radio waves in a large region of space surrounding the spacecraft, and consequently the satellite potential and the plasma sheath surrounding the spacecraft have negligible effect on the measured density. The PEACE experiment measures the flux of electrons at down to thermal energies, where the correction for the spacecraft floating potential is significant. Therefore there is a strong demand from PEACE for high reliability electron density data from WHISPER, which regularly receives lists of time intervals of particular interest. As these intervals are well in advance of the foreseen WHISPER date for delivering these data to CAA, it was agreed not to interrupt routine WHISPER density production, but to provide for PEACE cross calibration work specially derived accurate and reliable density files which, for convenience, are in a simple ASCII format. The main problems arise from the fact that the requested time intervals, particularly in the cusp and tail regions, are often difficult to interpret and process due to the absence of triggered resonances, the absence of clear signatures in the natural/passive data, and/or strong interferences of various natures. Anyway, the process is not without success and PEACE still continues to ask for new time intervals. For cross-calibration of the magnetic field with FGM and EDI resonances observed by WHISPER in the plasmasphere have been used. These resonances occur

200

J.G. Trotignon et al.

at the electron cyclotron frequency Fce and its harmonics nFce , from which the magnetic field strength may be derived (see Section 12.4.2). An algorithm has been developed [19], leading to a sub-pixel accuracy for the Fce determination, i.e. less than 163 Hz, the highest WHISPER frequency resolution. It is based on the Radon transform technique, commonly used in image processing. This algorithm has been validated on synthetic spectrograms. WHISPER measures Fce (and hence the magnetic field strength) directly, and the only limitation comes from the measurement accuracy. By using the nFce the Fce determination can be much more accurate provided the number of available harmonics is sufficient. Particle instruments have great difficulty in analyzing particle fluxes at thermal energies, which are comparable with the spacecraft floating potential. WHISPER should be of great help for internal calibration of such measurements. Particular attention has therefore to be paid to comparing density evaluations from WHISPER and CLUSTER particle experiments. The same is true for EFW and WHISPER density products [16].

12.6 Discussion and Conclusion The WHISPER relaxation sounder has been designed in such a way that natural wave (passive) and even active measurements cannot be perturbed by long lasting echoes received after transmitted wave trains. We indeed know, since the eighties, that sounder-stimulated resonances can last up to a few 100 ms (see, for example, Figure 12.2 in [21]), it is therefore required not to perform any natural and even active measurements at, or close to, the same frequency of the last transmitted pulse before the echo intensity fades away. To solve this problem, natural wave measurements are always made before active measurements and in active modes the transmitted frequencies are not swept in a regular order, they follows an appropriate order, which is described in a table and chosen by command, where two successive frequencies are always well separated. Finally, let us note that the minimum time separation between WHISPER active and passive operations is 215 ms, i.e. the time separation between the last sounding pulse and the next natural waveform recording: so even for very long lasting resonances no perturbations may be detected on natural spectra due to the averaging process applied before TM delivery. One appreciable advantage of WHISPER instrument, shared by most wave instruments, is that calibration files are stable over the mission lifetime. Conversion of raw spectra to physical units can be performed at an early stage of data handling, immediately after telemetry decommutation. The calibration procedure is still executed once per orbit, to check the instrument’s health. But during the 8 years since launch, the calibration files have never needed any revision. We may note that amplitudes of waves from distant radio sources (type III solar bursts), as measured from the four different WHISPER instruments, are equal to within 1 dB, less than the experimental uncertainty which is estimated to be 2 dB. Furthermore, the conversion of bin number to frequency is well defined due to the high stability of the onboard oscillators, and does not evolve with time.

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

201

Contrast Factor

However, a word of caution must be given concerning signal amplitudes. To convert the potential difference measured by the antenna to an electric field a fixed effective antenna length has been used. In reality, the CLUSTER antenna effective length depends on the plasma regime (see [1]). Scientific studies requiring precise signal amplitudes may need a better estimation of the antenna effective length as, for example, in the study of electrostatic waves as a possible cause of the precipitation of the particles associated with diffuse auroras [9]. The intrinsic quality of the main WHISPER data products (high resolution electric-field active/passive spectra, power spectral density, and large band intensity) is considered to be good, except when: (1) There are gaps in the data from one or more of the spacecraft during segments of the orbit where data coverage had been planned (as indicated in the Master Plan). Some gaps affect easily identifiable time intervals, other are irregular and random; (2) WHISPER data are corrupted on one or more spacecraft, possibly due to DWP/EFW/WHISPER bugs; (3) EDI strongly perturbs WHISPER data. EDI Event files indicate time intervals that are potentially affected, and the probable extent of the perturbation. The quality control of WHISPER density products is performed during the production of the plasma frequency. If the plasma frequency determination seems flawed it is corrected when possible, or else discarded. An automatically-computed “contrast” index is associated with each determination (see Section 12.4.1, for example). Examples of contrast factors for density determinations from active and passive data respectively are shown in Figs. 12.6 and 12.7. As Fuh resonances shown in the

0.6

Frequency (kHz)

0.2 Plasmasphere 60

80 Fce

40

dB

Fig. 12.6 Example of the contrast factor associated with the density derived from WHISPER active measurements in the plasmasphere: bottom panel, the derived plasma frequency (white full circles) superimposed on WHISPER combined active/passive spectrogram; top panel, the associated contrast factor computed during the recognition process

1.0

Fuh Fpe

40

20 0 15:00

17:00

19:00

WHISPER 1, 13 April 2005

21:00 UT

202

J.G. Trotignon et al.

Fig. 12.7 Example of the contrast factor associated with the density derived from WHISPER passive/natural measurements in the cusp region: bottom panel active spectrogram; middle panel plasma frequency (white line) superimposed on natural spectrogram; top panel, the associated contrast factor computed during the recognition process

bottom panel of Fig. 12.6 are, in this plasmasphere event, strong monochromatic peaks compared with the surrounding background level, the recognition process worked satisfactorily. The derive plasma frequencies are represented as white full circles in the active spectrogram (bottom panel) of Fig. 12.6 and, as can be seen in the top panel, the computed density contrast factors are very high. Figure 12.7 is an example of density determination results in the cusp region. Here the natural wave cut-off (middle panel) and resonances (bottom panel) were both used to obtain the plasma frequency profile shown as a white line in the natural spectrogram (middle panel). The contrast factor of density values derived from natural wave cutoff signatures, between 06:05 and 06:26 UT, is given in the top panel of Fig. 12.7.

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

203

Although the latter density values are very close to those derive from resonances, the contrast factor values appear to be equally distributed between 0.4, the value of the threshold used here, and 1. If the plasma frequency is found to be too “noisy” the user can either check the contrast factor and plasma frequency uncertainty and possibly remove unreliable values, or keep only density values derived from active spectra, with a consequent lower time resolution. It is worth recalling that electron density files are high level products that require human intervention and scientific expertise, their production is time consuming, and therefore they are produced and delivered on a best efforts basis. Anyway, the policy is to give as complete and as good a dataset as is possible. Two frequently asked questions are: What is about the density determination in non-maxwellian plasmas, such as a

bi-maxwellian with 10 eV and 10 keV populations? What are the connections with the boom and Debye lengths?

To tentatively answer the first question we can consider the Earth’s plasmasphere where two series of Bernstein Fqn resonances are often observed (two resonances instead of one are seen between two successive gyroharmonics, see for example Fig. 12.8 in [22]). Theoretical calculations [2] have shown they may be explained by the presence of a hot electron component superimposed on the dominant cold one. One of the two observed Fqn series is aligned in an appropriate plot, the so-called Hamelin diagram (The frequency separation between a Fqn and the gyrofrequency nFce , that is just below, is plotted as a function of the plasma frequency). This series is due to the cold plasma component and the corresponding density may therefore be simply deduced from the plasma frequency read on the plot abscissa. The other series can be used to estimate the hot to cold density and temperature ratios. The total density which can be derived in this way – after a time consuming analysis – is generally not very different from the density of the cold population. We consider nevertheless that the parameter to be archived is the total density. Hopefully, additional resonances related to the total density are often detected: one is the so-called Fx [2], which turns out to be the total electron plasma frequency here, the other is the upper-hybrid frequency Fuh . In the solar wind and magnetosheath Fpe and Fuh almost merge due to the weak intensity of the embedded magnetic field, leading to a single resonance observed in the active spectrum. There is thus no risk of a wrong identification of the characteristic plasma frequency, except in case of significant spurious noise, which is taken care of as explained in Section 12.4.1: in case this risk is present, a contrast factor linked to each density determination would be significantly lower than 1. The uncertainty on the plasma frequency resonance being the FFT frequency resolution, usually 162.8 Hz, the accuracy of the total electron density is thus a few percents for low densities (4% for an electron plasma frequency Fpe D 9 kHz, i.e. a density of Ne D 1 cm3 ) and less than 1% for high densities (0.5% for Fpe D 63:6 kHz, i.e. Ne D 50 cm3 ). In magnetized regions, like the plasmasphere, the resonance pattern is more complex. Resonances are indeed observed at Fce (which is proportional to the magnetic

204

J.G. Trotignon et al.

Fig. 12.8 Bryant plots showing the averaged density (top) and its “accuracy/uncertainty” (middle) during almost 8 months. In Bryant plots, the horizontal axis is absolute time or orbit number while the vertical axis is time since last perigee; thus each orbit is represented by a sloping line. The reference S/C position is given at the beginning, middle, and end of the time interval (bottom)

field strength), its harmonics, Fpe (proportional to the square root of the electron density), Fuh (the square root of Fpe squared plus Fce squared), and Fqn (which are higher than 2Fce , higher than Fuh , thus also higher than Fpe , and observed between two successive gyroharmonics). As said in Section 12.4.2, the challenge presented to the processing algorithm is the correct identification of Fuh , which is considered as the most intense and reliable resonance from which the total electron density may be derived, Fce being computed either from FGM magnetic measurements or nFce resonance frequencies. Section 12.4.2 also explains how the algorithm works, identifying Fuh as the frequency at highest intensity in the smoothed active frequency spectrum. The Fuh value returned by the algorithm may unfortunately differ from the true Fuh (when recognized via the Fpe , Fuh , Fce relationship). This is because the

12

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

205

amplitudes of the resonances that may be observed close to the computed smoothed spectrum maximum (Fpe , Fuh , the first Fq , possibly double, and even gyroharmonics) vary with the actual plasma composition (hot to cold density ratio) and also with antenna orientation with respect to B-field orientation, such that the highest peak may be not always at Fuh . However our practical expertise, based on numerous detailed analysis of case event studies, indicate that most of those resonances are grouped within a not too large frequency interval in the regions explored by CLUSTER. Diffuse plasma D resonances are currently observed in the ionosphere but have not been observed by the WHISPER relaxation sounder [24], probably due to the low level of the emitter. A crude estimate of the uncertainty range linked to the density data obtained in a magnetized region can be provided. Let’s take, as illustration, a case in the plasmasphere where Fce D 10 kHz. Whenever Fuh =Fce < 2, the resonance observed between Fce and 2Fce is Fuh , with a frequency resolution of 162.8 Hz, the FFT frequency resolution. The relative uncertainty on the density varies hence between C= 30% (Fuh near Fce , Fuh D 10:55 kHz, Ne D 0:14 cm3 ) and C= 2.3% (Fuh near 2Fce , Fuh D 19:6 kHz, Ne D 3:5 cm3 ). In case Fuh =Fce is in the range 4–6 and with a 2 kHz uncertainty on Fuh position (12 frequency bins) leads to a density uncertainty from C= 10% (Fuh near 4Fce , Fuh D 41:5 kHz, Ne D 20 cm3 ) to C=7% (Fuh near 6Fce , Fuh D 57:8 kHz, Ne D 40 cm3 ). To summarize, the density which is archived can reasonably be considered to be the total electron density in all regions. The uncertainty about the density is primarily due to the FFT frequency resolution, except in magnetized regions, where it is usually below 10% (up to about 30% when Fuh is close to Fce ). Concerning the relative sizes of the wavelength of the excited waves and the length L of the physical antenna, we know that whenever is close to or shorter than L, the amplitude of the received signal varies as for spherical probes, and the signal fades if L is a multiple of [12]. However, waves excited by relaxation sounders are electrostatic in nature. Their wavelengths are therefore, most of the time, very large compared with the antenna length. Resonance frequencies are not affected. On comparative sizes of the Debye length D and the physical antenna length L, let us recall that the minimum wavelength in a collective plasma is actually D , so the antenna must be longer than D in order to sense the plasma presence instead of responding as if it were in free space. A hot population with an associated Debye length much larger than L could therefore be ignored by the sounder. Unfortunately, no quantitative estimate of the real limit has been made. However, qualitative information is available (comparison of the signatures of sounder and mutual impedance experiments, theoretical models of the mutual impedance response: [4, 5, 17]). Let us finally mention that visualization tools have been developed to validate data a posteriori and to overview density products. Two outputs of one of them are presented in Fig. 12.8. The Bryant plot [14] format has been used to see the averaged density and its uncertainty over large time intervals. Predictions of various boundary crossings, including the neutral sheet, magnetopause and bow shock, are indicated

206

J.G. Trotignon et al.

by coloured symbols. WHISPER operates in all regions but it becomes clear that the favourable regions for the density determination (i.e. where resonances are triggered in the instrument’s frequency range at a high enough level and natural wave signatures, such as cut-offs, are observed and identified) are: solar wind, most of magnetosheaths, cusp and clefts, parts of dayside magnetosphere, parts of nightside plasmasheet. Most of lobes are unfavourable. As a conclusion, the sounder is certainly a sophisticated instrument, but it is robust and it produces data whose quality is both very stable over the long-term, and from which it is possible to derive, in most of the key regions of the Earth’s ionized environment, reliable and accurate measurement of the total ambient electron density in a volume of space large enough for local plasma sheath effects to be negligible. Plasma densities (as archived in CAA) from 0.14 to 80 cm3 are derived from WHISPER relaxation sounder data with a good accuracy: a few percents for low densities and less than 1% for high densities, in non-magnetized or weakly magnetized regions; and, for instance, below 10% when Fpe is above 0.7 Fce down to about 30% when Fpe is close to 0.34 Fce , in magnetized regions with Fce around 10 kHz). WHISPER densities are therefore currently used for calibrations of CLUSTER instruments, in particular particle instruments, which have difficulties to measure the colder part of the population, and EFW instrument, because spacecraft potential variations, currently used as a proxy of density variations, depend not only on the density but also on the energy of electrons. Finally, it is worth recalling that WHISPER also allows the spectral energy distribution of natural waves to be monitored from 2 to 80 kHz. Natural and active measurements bear complementary information about plasma regimes that are actually used in the electron density determination process. Acknowledgements The authors thank all those who are in charge of CAA development, maintenance, and operations for their friendly assistance and efficiency. They also wish to express their great thanks to people from CNES who produce and deliver most of the WHISPER data products. Thanks are also due to all the scientists, research assistants, and engineers who have contributed their time to make the WHISPER experiment such a great success.

References 1. B´eghin C, D´ecr´eau PME, Pickett J, Sundkvist D, Lefebvre B (2005) Modeling of CLUSTER’s electric antennas in space: application to plasma diagnostics, Radio Science, 40, RS6008, 1–18, doi:10.1029/2005RS003264. 2. Belmont G (1981) Characteristic frequencies of a non-Maxwellian plasma: a method for localizing the exact frequencies of magnetospheric intense natural waves near fpe, Planet. Space Sci., 29, 1251–1266. 3. Belmont G, Canu P, Etcheto J, de Feraudy H, Higel B, Pottelette R, B´eghin C, Debrie R, D´ecr´eau PME, Hamelin M, Trotignon JG (1984) Advances in magnetospheric plasma diagnosis by active experiments, in Proc. Conf. Achievements of the IMS, 26–28 June 1984, Graz, Austria, ESA SP-217, 695–699. 4. D´ecr´eau PME (1983) Fonctionnement d’une sonde quadripolaire sur satellite magn´etosph´erique (exp´erience GEOS). Contribution a` l’´etude du comportement du plasma froid au

12

5.

6.

7.

8.

9.

10. 11.

12.

13. 14.

15.

16.

17.

18.

19.

The WHISPER Relaxation Sounder and the CLUSTER Active Archive

207

voisinage de la plasmapause e´ quatoriale, state doctorate thesis, Orl´eans University Orl´eans, France. D´ecr´eau PME, Hamelin M, Massif R, De F´eraudy H, Pawela E, Perraut S, Pottelette R, Bahnsen A (1987) Plasma probing by active wave experiments on the Viking satellite, Ann. Geophys. Ser. A, 5(4), 181–186. D´ecr´eau PME, Fergeau P, L´evˆeque M, Martin Ph, Randriamboarison O, Sen´e FX, Trotignon JG, Canu P, de F´eraudy H, Bahnsen A, Jespersen M, M¨ogensen PB, Iversen I, Dunford C, Sumner A, Woolliscroft LJC, Gustafsson G, Gurnett DA (1993) ‘WHISPER’, a Sounder and High-frequency Wave Analyser Experiment, ESA SP-1159, pp. 51–67. D´ecr´eau PME, Fergeau P, Krasnosel’skikh V, L´evˆeque M, Martin P, Randriamboarison O, Sen´e FX, Trotignon JG, Canu P, M¨ogensen PB, and WHISPER Investigators (1997) WHISPER, a resonance sounder and wave analyser : performances and perspectives for the CLUSTER mission, Space Sci. Rev., 79, 157–193. D´ecr´eau PME, Ducoin C, Le Rouzic G, Randriamboarison O, Rauch JL, Trotignon JG, Valli`eres X, Canu P, Darrouzet F, Gough M, Buckley A, Carozzi T (2004) Observation of continuum radiations from the CLUSTER fleet: first results from direction finding Ann. Geophys. 22, 2607–2624. El-Lemdani F, Rauch JL, D´ecr´eau PME, Trotignon JG, Suraud X, Valli`eres X, Darrouzet F, Canu P (2009) Wave emissions at half gyroharmonics in the plasmasphere region: CLUSTER observations and statistics, Adv. Space Res., 43, 253–264, doi:10.1016/j.asr.2008.06.007. Etcheto J, de Feraudy H, Trotignon JG (1981) Plasma resonance simulation in space plasmas, Adv. Space Res., 1, 183–196. Etcheto J, Belmont G, Canu P, Trotignon JG (1983) Active sounder experiments on GEOS and ISEE, in: Proc. Symp. on Active Experiments in Space, 24–28 May 1983, Alpbach, Austria, ESA SP-195, 39–46. Gurnett DA (1998) Principles of space plasma wave instrument design, in Measurement techniques in space plasmas: fields, edited by Pfaff RF, Borovsky JE, Young DT, American geophysical Union, USA, pp. 121–136. Interface Control Document for WHISPER, ICD (2009) ESA, CAA-WHI-ICD-0001, issue 1.6 (http://caa.estec.esa.int/documents/ICD/CAA WHI ICD V16.pdf). Lepine DR, Beharrel I, Bryant DA, Ely RJ, Macdougall JR, Gershuny EJ, Norris AJ (1985) Data Handling for the UKS Spacecraft, IEEE Transactions On Geoscience And Remote Sensing, 23, 221–229. Pedersen A, Cornilleau-Wehrlin N, De La Porte B, Roux A, Bouabdellah A, D´ecr´eau PME, Lefeuvre F, Sen´e FX, Gurnett D, Huff R, Gustafsson G, Holmgren G, Woolliscroft L, HStC. Alleyne, Thompson JA, Davies PNH (1997) The Wave Experiment Consortium (WEC), Space Sci. Rev., 79, 157–193. Pedersen A, Lybekk B, Andr´e M, Eriksson A, Masson A, Mozer FS, Lindqvist PA, D´ecr´eau PME, Dandouras I, Sauvaud JA, Fazakerley A, Taylor M, Paschmann G, Svenes KR, Torkar K, Whipple E (2008) Electron density estimations derived from spacecraft potential measurements on CLUSTER in tenuous plasma regions, J. Geophys. Res., 113, A07S33, doi:10.1029/2007JA012636. Perraut S, de Feraudy H, Roux A, D´ecr´eau PME, Paris J, Matson J (1990) Density measurements in key regions of the Earth’s magnetosphere: cusp and auroral region, J. Geophys. Res., 95, 5997–6014. Perry C, Eriksson T, Escoubet P, Esson S, Laakso H, McCaffrey S, Sanderson T, Bowen H, Allen A, Harvey C (January 2006) The ESA CLUSTER Active Archive, in Proc. CLUSTER and Double Star Symposium – 5th Anniversary of CLUSTER in Space, 19–23 September 2005, Noordwijk, The Netherlands, ESA SP-598. Rauch JL, Suraud X, D´ecr´eau PME, Trotignon JG, Led´ee R, El-Lemdani Mazouz F, Grimald S, Bozan G, Valli`eres X, Canu P, Darrouzet F (January 2006) Automatic determination of the plasma frequency using image processing on WHISPER data, in Proc. CLUSTER and Double Star Symposium – 5th Anniversary of CLUSTER in Space, 19–23 September 2005, Noordwijk, The Netherlands, ESA SP-598.

208

J.G. Trotignon et al.

20. Serra J (1982) Image analysis and mathematical morphology, Vol. 1, Elsevier Academic Press London, XVIII-610 p. 21. Trotignon JG, J. Etcheto J, Thouvenin JP (1986) Automatic determination of the electron density measured by the relaxation sounder on board ISEE 1, J. Geophys. Res., 91, 4302–4320. 22. Trotignon JG, D´ecr´eau PME, Rauch JL, Randriamboarison O, Krasnoselskikh V, Canu P, Alleyne H, Yearby K, Le Guirriec E, S´eran HC, Sen´e FX, Martin Ph, L´evˆeque M, Fergeau P (2001) How to determine the thermal electron density and the magnetic field strength from the CLUSTER/WHISPER observations around the Earth, Ann. Geophysicae, 19, 1711–1720. 23. Trotignon JG, D´ecr´eau PME, Rauch JL, Le Guirriec E, Canu P, Darrouzet F (2003a) The WHISPER relaxation sounder onboard CLUSTER: a powerful tool for space plasma diagnosis around the Earth, Cosmic Res., 41(4), 345–348, Translated from Kosmicheskie Issledovaniya, 41 369–372. 24. Trotignon JG, Rauch JL, D´ecr´eau PME, Canu P, Lemaire J (2003b) Active and passive plasma wave investigations in the earth’s environment: the CLUSTER/WHISPER experiment, Adv. Space Res., 31, (5)1449-(5)1454, doi: 10.1016/S0273-1177(02)00959-6 25. Trotignon JG, D´ecr´eau PME, Rauch JL, Suraud X, Grimald S, El-Lemdani Mazouz F, Valli`eres X, Canu P, Darrouzet F, Masson A (January 2006) The electron density around the Earth, a high level product of the CLUSTER/WHISPER relaxation sounder, in Proc. CLUSTER and Double Star Symposium – 5th Anniversary of CLUSTER in Space, 19–23 September 2005, Noordwijk, The Netherlands, ESA SP-598. 26. WEC Instrument User Manual, WEC UM (2000) ESA, CL-WEC-UM-002, issue 1.03 (http://caa.estec.esa.int/caa/dwp docs.xml).

Chapter 13

ESOC Data Products in the CAA J. Volpp and D. Sieg

Abstract This document intends to help users to find information provided by the Cluster Flight Control and Flight Dynamics teams. Both teams deliver to the Cluster Active Archive a wide range of operations related information, which could be helpful as background information for science data processing. The products range from high level summary reports to specific detailed analysis of spacecraft and ground segment anomalies, including operations timelines, list of all manual platform and payload operations, and summary plots of key platform parameters such as evolution of temperature and time correlation. This document provides for each product the information, when it is ingested into the Cluster Active Archive, what it contains and if applicable why this information could be useful for users.

13.1 The Cluster Mission 13.1.1 Ground and Space Segment The Cluster mission operations are carried out by ESA from its European Space Operations Centre (ESOC) located at Darmstadt/Germany. Most of ESA’s science and Earth observation missions are controlled from ESOC. It is also in charge of maintaining and operating ESA’s ground stations, the European Space Tracking Network (ESTRACK). The core network encompasses the 15 m antennas at Villafranca and Maspalomas (both Spain), Redu (Belgium), Kiruna (Sweden), Kourou (French Guiana) and Perth (Australia) and the two deep space antennas New Norcia (Australia) and Cebreros (Spain). The routine operations of the Cluster mission involves the continuous cooperation between the Cluster Flight Control Team (FCT) and the Flight Dynamics Team, J. Volpp European Space Operations Centre, Robert-Bosch-Strasse 5, D-64625 Darmstadt, Germany e-mail: [email protected] D. Sieg Hewlett-Packard GmbH at European Space Operations Centre e-mail: [email protected] H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 13, c Springer Science+Business Media B.V. 2010

209

210

J. Volpp and D. Sieg

both at ESOC, the Joint Science Operations Centre (JSOC) located at the Rutherford Appleton Laboratory (RAL) near Didcot/United Kingdom and the instrument teams. The Cluster Flight Control Team has the overall responsibility for the commanding of the spacecraft platform and payload and requests the ground station passes from the ESTRACK Scheduling Office. The Flight Dynamic Team prepares the planned orbit and attitude manoeuvres and monitors and evaluates the execution. JSOC merges the Principal Investigators’ (PI) requests into coordinated, regular scientific mission planning input for the mission planning system at ESOC. As a detailed spacecraft design description was given by Credland et al. [2] only a few basics will be mentioned here. The main structural element are the inner cylinder, the main equipment platform (MEP) and the six curved solar array panels forming the outer cylindrical shape of the spacecraft body. The inner cylinder with the inner platform houses part of the propulsion system and the five batteries and their regulators. The experiments are accommodated either on the upper part of the MEP around the rim, on the two rigid booms or on the four wire booms. The platform subsystems are mounted on the lower surface part of the MEP. The six propellant tanks are located below the main equipment platform and attached to the inner cylinder. Four radial and two axial 10 N thrusters of the Reaction Control Subsystem (RCS) are accommodated at the lower ends of the cylinder. Two axial thrusters are mounted on the upper face of the spacecraft. The spacecraft temperature is controlled in sun light with the excess power from the solar arrays, i.e. how it is split between the internal and external power dumpers (IPD/EPD). Most of the heaters are used during eclipse only. The average MEP–temperature (MEP-T), calculated from the reading of several temperature sensors, is used as a reference temperature. Routine instrument operations are performed via time tagged commands with a time resolution of 1 s. The on-board time provided with the science data enables a time resolution down to 10s. The drift between the on-board clock and the Universal Time (UT) is monitored. Periodically a time correlation between on-board time and the UT at the ground station is performed and the resulting correlation parameters are automatically added to the auxiliary data.

13.1.2 Mission Operations Concept The underlying spacecraft operations concept gives the frame for the various ESOC data products provided in the CAA. A detailed mission operations concept applicable for the initial Cluster mission in 1996 was given by Ferri and Warhaut [5]. As significant changes were made for the Cluster mission launched in 2000 the new mission operations concept is summarized below including the evolution of the mission from 2000 until 2009. Cluster is a non-real time mission. However, constellation change manoeuvres, routine attitude slews and eclipse operations are performed mostly as manual and special operations and not via the mission planning system. Payload operations in real time are the exception. It is limited to special requests from the instrument

13

ESOC Data Products in the CAA

211

teams such as calibration, software patches, manual restarts and trouble-shooting after contingency situations. All routine payload operations are planned 3 weeks in advance for each planning period. The FCT generates the composite command timeline for both system and instruments with the input from JSOC. JSOC is merging on a weekly basis the Principal Investigators’ requests into coordinated, regular scientific mission planning inputs, the Observation Request (OBRQ). At each pass the FCT performs regular health checks and uplinks the time tagged command schedule. The visibility of the Cluster fleet changed considerably during the past years: The visibility from the Villafranca station decreased from 40% in 2001 to 25% in 2005 while the visibility from Perth increased from 42% in 2001 to 55% in 2009. As the Perth ground station supports also the XMM-Newton mission Cluster can effectively use only about 28% of its orbit period to schedule a pass. The visibility from Maspalomas decreased from 40% to 30%. As the four Cluster spacecraft have to share two ground stations since June 2002 this leaves about 8 h contact time to each spacecraft per orbit revolution. The science and house keeping data are stored on board in the Solid State Recorder (SSR) outside visibility. Real time housekeeping and real time science telemetry are down linked during each ground station pass, together with playback data from the SSR. The data are then transferred to ESOC and stored on the Cluster Data Dissemination System (CDDS). This server provides a quick-look facility for the instrument teams and a selection of these data is plotted and made publicly available on the JSOC server. In addition, Wide Band Data experiment data are transmitted directly to the ground stations of NASA’s Deep Space Network and transferred to the PI institute in the United States. The house keeping, science and auxiliary data on the CDDS were copied on to CD-ROM until end 2005 and distributed by mail to the Cluster community. Since January 2006 the data are transferred as a ‘CD-ROM image’ via Internet to the CAA at ESTEC where it is available to the community. The Cluster FCT performs regularly trend analysis of key parameters, e.g. solar array output power and MEP-T, keeps track of all special operations and events in various data bases, the spacecraft event log, the special operations requests (SPO), data bases for anomaly reports of the platform, payload and the ground segment. The weekly mission operations reports are issued. They summarize the major events including the cause of data gaps larger than 2 min and the resulting science data return ratio and lists other reports/documents issued. Beside all activities related to manoeuvres the flight dynamics experts support the Cluster user community in the decision which constellation to be selected and perform routinely the orbit and attitude determination and provide the results as auxiliary data to the CDDS.

13.1.3 Evolution of the Cluster Mission The Cluster mission and in particular the ground segment evolved since the launch in 2000. The initial mission was planned to last only 2.5 years. One dedicated ground station, Villafranca-1 (Spain), had to be shared by all four spacecraft. This constraint

212

J. Volpp and D. Sieg

Fig. 13.1 Inter-spacecraft separation distances in the past. Constellations manoeuvres were used to change from one separation distance to the other approximately every 6 months

in downlink capability limited the science operations to 50% of the orbit. In 2002 the mission was extended twofold: A second ground station, Maspalomas on the Canary Islands, was made available allowing continuous scientific measurements throughout the entire orbit. This science data extension phase started on 2 June 2002. Additionally, the mission itself was extended by 3 years up to end 2005. For the second mission extension, from January 2006 until end 2009, the ground station usage for Cluster was changed: Maspalomas was kept but the station at Perth (Australia) replaced Villafranca-1. Also the constellation strategy had been changed during the mission. The separations between the spacecraft achieved so far (October 2008) are ranging from 40 up to 10,000 km (see Fig. 13.1). The strategy during the first 5 years was to keep a perfect tetrahedron at two points around the orbit (purple line). Then, since 2005, multi-scale measurements are achieved by drifting the spacecraft around their orbit. This is saving fuel but is limiting the constellation that can be achieved: Cluster C1, C2 and C3 form a 10,000 km triangle that only the orientation can be changed and on the other hand, C3 and C4 have approximately the same orbit and by drifting one spacecraft with respect to the other they can be moved to a few tens of km together or separated by 10,000 km or more. A special operation was conducted in May 2008 to measure for the first time the third component of the electric field by tilting spacecraft 3 with respect to spacecraft 4 by 45ı , while the two spacecraft were separated by 40 km.

13.2 Reports and Technical Notes These reports and documents as listed in Table 13.1 were generated during operations and ingested into the Cluster Active Archive (CAA) in the following year.

13

ESOC Data Products in the CAA

213

Table 13.1 Summary list of reports and technical notes File Format Monthly reports Portable Document Format (PDF) Weekly mission operations reports Portable Document Format (PDF) Spacecraft anomaly reports Portable Document Format (PDF) Ground segment anomaly and observations reports Portable Document Format (PDF) Spacecraft system technical notes and reports Portable Document Format (PDF) Manoeuvre reports Portable Document Format (PDF) Eclipse and battery reports Portable Document Format (PDF) AGC fluctuation reports Portable Document Format (PDF) Miscellaneous technical notes Portable Document Format (PDF)

Monthly Reports The monthly operations reports describe on a high level the major spacecraft operation events and ground segment activities including the overall mission performances. On one page the operations and status of platform, payload and ground segment are summarized including the list of anomalies occurred. Weekly Mission Operations Reports The weekly Mission Operations Report (MOR) is a summary of the spacecraft and ground segment operations performed in this week. It addresses separately the ground segment, the spacecraft platform and the payload. It provides information about ground segment operations and events, configuration changes, platform and instrument anomalies and special/manual operations performed. All spacecraft and ground segment anomalies occurred in that week are addressed. If applicable details about eclipses and manoeuvres are provided. Most useful for the users might be the identified data gaps larger than 2 min, manual instrument operations performed, and information about non-nominal behavior of the time correlation. On a quarterly basis the detailed instrument status as received from the instrument teams is attached. Spacecraft Anomaly Reports The Spacecraft anomaly report is extracted from the Anomaly Report Tracking System (ARTS) at ESOC which is used by all missions. The naming convention is CL AR-nnn. Each anomaly report documents the symptoms of the anomaly, its operational impact, including data loss, and recovery actions taken. If applicable an analysis about its cause and preventive action taken are included. Related files generated during the analysis are attached. Also reference to related anomaly reports is made, if similar events had happened before. Ground Segment Anomalies and Observation Reports The ARTS at ESOC is also used to record and track anomalies, failures and operational mistakes occurred in the ground system. For the Cluster project the naming convention for ground segment anomalies is CLU-nnnn. These reports cover events which occurred in the period between mission planning up to the data delivery to the archive at ESTEC. Of special interest for the user could be anomalies causing increased errors in the time correlation, e.g. due to configuration changes in the ground stations.

214

J. Volpp and D. Sieg

The Cluster Anomaly Report includes a general description of the event, its impact on data retrieval and the actions taken. Spacecraft System Technical Notes and Reports During the ongoing Cluster mission reports and documents were issued either (a) quasi periodically or (b) dedicated to a specific problem. The first covers seasons with eclipses or manoeuvres, battery handling and regular trend analysis of the solar array performance. The second cannot be characterized, it includes instrument operations constraints during eclipse season, the process of time correlation, up to re-sizing the time tagged command buffer in the spacecraft platform and test reports. A screening through the titles should help to indicate whether it could help the user. Manoeuvre Reports These reports document the operations for the constellation change or phasing manoeuvre seasons. The documents give full details, time and duration, of all manoeuvres performed, including observations/anomalies observed and lessons learnt. One report for each constellation change or phasing season is provided. The amount of fuel burned during a manoeuvre could impact the instruments performance or measurement results. Eclipse and Battery Reports These reports present all details about the eclipses, the performance of the various spacecrafts batteries after the short or long eclipse seasons or after Moon eclipses. The battery health assessment at the beginning of the eclipse season is included. Eclipse reports were issued up to spring 2006. With degrading batteries and increasing length of the eclipse season no dedicated eclipse season reports were produced any more in favor of a detailed, immediate reporting after each eclipse and a lessons learnt document after the eclipse season. The reports contain information about the spacecraft temperature evolution which could be of interest for the instruments. AGC Fluctuation Reports Strong fluctuations of the Automatic Gain Control (AGC) of the receivers on-board and at the ground station showed up regularly in autumn disturbing the up- and downlink to the spacecraft since 2000. In 2002 the cause could be linked to Rayleigh-Taylor-instabilities in the low latitude ionosphere. This so called Equatorial Spread F phenomenon occurs during hours after sunset above the magnetic equator or during magnetic sub-storms. Two ground stations were affected: Maspalomas and Villafranca. These reports summarized the observed effects and the investigations performed since autumn 2001. The annual reports list all AGC events, their time, the elevation and azimuth of the spacecraft wrt. the ground station. During periods with strong disturbances also the weekly mission operation reports include all the data gap information and the operation performed. Miscellaneous Technical Notes with Relevance to Science Data These documents were issued in course of the mission to document analysis made about specific problems, as time correlation, effect of leap second, on-board monitoring schemes etc. A description which would cover all documents cannot be provided. A screening through the titles of the documents is recommended.

13

ESOC Data Products in the CAA

215

13.3 Spacecraft System Data The annual spacecraft system data provided are summarized in Table 13.2. They are retrieved about 1 year later and ingested then into the CAA, i.e. in the year N C 1 the data on year N are compiled. Spacecraft Event Log and Major Platform Event Lists The Cluster Event Log is an Excel book and includes all activities, which were not scheduled by the Mission Planning System, e.g. special operations requested by instrument teams or Engineers. It includes the detection of platform or payload anomalies and Out Of Limits (OOL) and their recovery. In addition to the main event log sheet the book contains several worksheets compiling the occurrence of frequent events/anomalies: all CTU switch-over and re-boots, all re-occurrences of the anomalies AR-008, AR-013, and AR-052 (Bus Mis-Match Error), all SSR failures and all un-commanded orbit changes. The event file includes all the details concerning the special activity: execution time, the spacecraft identifier, procedure applied and/or input provided by the instrument teams. Special Payload and Platform Operations (SPO) The Special Operations Request (SPO) is a form used by the Cluster Flight Control Team to document and execute manual operations outside the mission planning activities in a controlled way. Each SPO includes all the instructions to be performed during special payload and/or platform activities and in case of instrument operations it refers to or even includes the input received from the instrument team. TDA Mode Changes Telemetry Definition Acquisition (TDA) modes define roughly speaking the data rates for telemetry transmission and/or recording. The Cluster Acquisition Byte (CAB) defines basically the telemetry type, i.e. normal science data or Burst Mode data. A concise specification of both. TDA mode and CAB is given in ESA [3]. The TDA mode changes as executed on board were retrieved from the housekeeping telemetry (VC0) and compiled together with the selection

Table 13.2 Summary list of spacecraft system data provided File Format Spacecraft event log Special payload and platform operations TDA mode changes Evolution of the solar array performance Battery conditioning plan Battery conditioning history data IPD, EPD power and MEP temperature Thermal eclipse data and bus current Thruster summary and manoeuvre history SSR bit error counts Examine file

Microsoft Excel Portable Document Format (PDF) Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel ASCI text file

216

J. Volpp and D. Sieg

of the Cluster Acquisition Byte. As this could be in some cases, eclipse seasons and recovery operations, different from the operations planned via Joint Science Operations Centre (JSOC) this could be helpful to reconstruct the operations on board. Evolution of Solar Array Performance The overall Solar Array (SA) performances and prediction data is covered by a single Excel file with respect to the spacecraft orbit position and orbital elements. The file includes the actual SA current measurements, the graph and a statistical model to predict the SA power for the future years. The model includes normalisation of the data to 1AU, 23ı C solar array temperature and 90ı solar aspect angle using the actual temperatures and solar aspect angle. The SA power output has no direct impact on the science data, however, the degradation rate could give an indication about the (long-term) radiation environment of the spacecraft. Battery Conditioning Plans During the mission operation phase the five batteries on each single spacecraft undergo a conditioning cycle with four states: empty storage, charging, full storage and discharging. The conditioning strategy applied has changed during the mission considerably, as the mission was extended and the batteries started to degrade. The battery-conditioning plan includes all scheduled activities specifying the status of each single battery. Each file covers a few months period. The length is governed by the occurrence of eclipse seasons. This plan should have no impact on instrument operations. However, as it turned out that some batteries cracked during charging or discharging process and released gas the conditioning plan could help to rule out an unexplained signal seen in the data. Battery Conditioning History Data The Battery capacity evolution at the end of each conditioning season is outlined in several worksheets of a single Excel file including plots, comments and power estimation for the next eclipse season. These data are most likely not of relevance for science data analysis. Each file covers the period from launch up to the year of the last ingestion. It contains the data in Excel sheets and the associated graphs. The file is updated every year. IPD, EPD Power and MEP Temperature The Internal Power Dumper (IPD) data show the power disseminated in the internal heaters. The External Power Dumper (EPD) data show the excess power radiated into space. As it had been observed that large changes in the IPD/EPD could influence the readings of other analogue parameters via cross talk these parameters could have an effect on instruments data. The excess power into the IPDs is the main driver to regulate the temperature of Main Equipment Platform (MEP) and with this the temperature of the instruments mounted onto the MEP. Long term trends in the MEP average temperature can be seen on this plot. Each file covers an entire year with the data in Excel sheets and the associated graphs. A new file is added every year.

13

ESOC Data Products in the CAA

217

Thermal Eclipse Data – Permanent and Switchable Bus Current The files provide for the long eclipses the temperature evolution of spacecraft units/ subsystems, e.g. Main Equipment Platform (MEP), the High Power Amplifier (HPA), the Reaction Control System (RCS) and the heaters. The power load of the two power busses, permanent and switchable is included. Each file covers an entire eclipse season with the data in Excel sheets and the associated graphs. A new file is added every year up to end 2008. Beyond 2008 in most cases no house keeping data will be available during the eclipses as the On-Board Data Handling (OBDH) has to be switched off. Thruster Summary and Manoeuvre History The file summarizes the individual thrusters usage for all manoeuvres conducted during the lifetime of the mission and provides a table with the overall manoeuvres history. From the burn duration of each thruster the amount of propellant used can be calculated. This could be of interest if the amount of exhaust could influence the instrument performance. This table is updated every year. SSR Bit Error Counts The Solid State Recorder (SSR) performs continuous background memory scrubbing in which the SSR corrects any 1 symbol (4-bit) Single Event Upsets (SEU) failure and detects any 2 symbol (2 4-bit) SEU failure per 64-bit data word. The numbers of these correctable and non-correctable errors are named Bit Error Counts. They are accumulated while the SSRs are operating and are dumped to ground and reset at each ground station pass. The signal of radiation events such as large Coronal Mass Ejections can be identified (and correlated with Proton flux obtained from GEOS satellites) on top of the background level most likely produced by the cosmic rays as the long term trend from 2001 to 2008 suggests. Each file contains the total SSR bit error counts obtained on-board between the start of two passes. It covers an entire year with the data in Excel sheets and the associated graphs. A new file is added every year. Examine File The Examine File is generated by the Mission Planning System based on the required platform commands with input provided by Joint Science Operations Centre for the instrument operations, Flight Dynamics System and the ESTRACK Scheduling Office. The file is the concise operations plan including the payload and platform operations in terms of command sequences and their time tag, the passes scheduled over the nominal ground station and the forecasted SSR fill levels.

13.4 Flight Dynamics Data All files listed in Table 13.3 are submitted by Flight Dynamics once per year and are thus in sections covering one calendar year. The files are in ASCII text format and the complete history can be obtained by concatenating the files of subsequent

218

J. Volpp and D. Sieg Table 13.3 Flight Dynamics data files File DDID acronym Orbit STOF/LTOF Covariance matrix COVM Eclipse Event STEF/LTEF Spacecraft attitude and spin rate SATT Reports

Format ASCII text; access routine needed ASCII text; access routine needed ASCII text, human-readable ASCII text, human-readable ASCII text, human-readable Portable Document Format (PDF)

years. Each file contains a sequence of data records. Some of them span intervals of several hours. If the interval starts in one year and ends in the next year then the record is included only in the archive of the new year. Therefore some hours at the end of the year might not be covered by the file for a specific year. Descriptions of the file formats are given by ESA/ESOC in the Cluster Data Delivery Interface Document (DDID), ESA/ESOC [4]. Data is provided to the archive only after the final assessment of all relevant raw measurements. Thus the archive never contains data records flagged as predicted. Orbit File Using the corresponding FORTRAN access routine the user can retrieve state vectors, i.e. positions and velocities, of the satellite at any desired epoch from the orbit file. Then further geometric information like the distances and directions between satellite pairs can be derived. The archive orbit file is in the same format as known from the Long Term Orbit File (LTOF) and Short Term Orbit File (STOF). The original data refers to an inertial coordinate system. If orbit data is available in other coordinate systems like solar or geomagnetic systems then the data was already post processed by other parties. Covariance Matrix File Together with the corresponding FORTRAN access routine the user gets estimated covariance matrices which correspond to the orbit state vectors retrieved from the orbit file at the same epochs. The covariance matrices of the individual spacecraft state vectors have to be combined when deriving errors of spacecraft configurations or spatial gradients as done by Chanteur and Harvey [1]. Typically the position error at apogee is around one kilometer and gets smaller around the orbit towards perigee. Especially when flying in very small formations one might have to consider errors in the inter-satellite directions and distances when interpreting instrument measurements. Eclipse File One eclipse file for each Cluster spacecraft lists all start and end times (penumbra and umbra) of Moon and Earth eclipses and the minimum angle between the two lines of sights seen by the spacecraft. Event File The archive event file contains the same type of events as the Long Term Event File (LTEF). All events are calculated from the final orbit data as available in the archive. Most of the events were mainly of interest at the stage of planning the actual spacecraft operations. Concerning the eclipse and manoeuvre events there are other

13

ESOC Data Products in the CAA

219

sources available (eclipse files and reports) which provide more information. Of major interest for the archive user are probably just the perigee crossing events listed as “START REVOLUTION” as they contain the orbit revolution number. Spacecraft Attitude and Spin Rate File (SATT) This file contains a complete history of the determined attitudes including spin axis, right ascension, declination and the spin rate. Extract of Flight Dynamics Reports This document contains reports about the manoeuvre optimization, command generation, orbit and attitude determination tasks performed by Flight Dynamics. As the intended readership are flight dynamics experts, not everything might be understandable to other readers. Nevertheless the document is considered to contain useful information for everybody in particular the introduction and the summary of constellations sections. These provide a good overview about all flown constellations. To get more details about a specific constellation one can then also take a look at the manoeuvre optimization part which contains plots of the satellite formation around the orbit.

13.5 Ground Segment Data All the annual spacecraft operations system data in Table 13.4 described in the remaining part of the paper can be grouped into three categories: (a) ground station pass information, (b) timing information about the on-board time and the ground segment and (c) data recovery and data gaps. These data products are retrieved/generated about 1 year later and ingested then into the CAA, i.e. in the year N C 1 the data on year N are compiled. Request to ESTRACK Scheduling These files cover all ground station passes as requested to be scheduled based on the Cluster Mission Planning System. This plan identifies the Ground Station, the Spacecraft and the scheduled time of the Acquisition Of Signal (AOS), i.e. the time of first telemetry frame expected, and the scheduled Loss Of Signal (LOS), the time of the last telemetry frame.

Table 13.4 Summary list of ground segment related data provided File Format Request to ESTRACK scheduling Actual ground station pass times Drift of the on-board clock Time correlation coefficients Internal delays for time correlation Data base changes Data recall log-file and data gap analysis Science data return ratio of frames/formats Science data volume and data return ratio summary

Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel Microsoft Excel

220

J. Volpp and D. Sieg

Actual Ground Station Pass Times The so called Ground Station Utilization log (GSU-LOG) covers all the ground station passes as actually performed including the AOS/LOS times. The GSU-LOG is maintained by the spacecraft controllers. It includes all extra passes, not shown in the request to ESTRACK scheduling and catches all deviations from planned AOS/LOS times. This information could be helpful to reconstruct from which ground station the data were received in case of a problem with the time correlation. Drift of the On-Board Clock The drift of the on-board time against the ground station time is monitored by the Cluster Mission Control System during each pass by calculating the so called DIFF-time parameter. This is the difference between the on-board time as derived from the time correlation using the current applicable time correlation coefficients and as calculated by using the Earth reception time and the known delays. Each file of the DIFF-time covers for each spacecraft a three months period containing the individual data in an Excel sheet and the graph. The DIFF-time plot includes comments and a summary graph. It is identical to the plots distributed during the mission on a quarterly basis to all instrument teams. Since October 2004 the sign of the DIFF-time parameter has been added. Since 23 November 2007 a time correlation is performed periodically at each nominal pass, not waiting any more for the DIFF-time to reach 2 ms. It helps to estimate the timing error caused by the drift of the on-board clock. This information could help to achieve a much better timing accuracy than the required 2 ms as described by Yearby et al. [6]. Time Correlation Coefficients The time correlation table provides the date and time when for each spacecraft a new time correlation was performed and accepted and lists the coefficients to be used. This information is actually already part of each daily CD-ROM data set image in the CAA. No additional information is added to this table. Internal Delays for Time Correlation The Spacecraft Event Time (SCET) of a measurement is calculated independent from the time correlation coefficients by using the Earth Reception Time (ERT), the propagation delay spacecraft-to-ground and the on-board delays. The ERT is the time stamp from the ground station. It has to be corrected by the ground station internal delays. The one-way-light-time requires the knowledge of the geographical position of the antenna. The spacecraft position is continuously updated by the time correlation software using the latest released Short Term Orbit File. The on-board internal delays from measurement to inserting the on-board time into each frame had to be corrected on ground. The table provides the geographical position of all ground stations and all bit-rate dependant internal delays at each ground station and spacecraft used by the time correlation software in the Cluster ground system. The user should be aware that a significant change occurred with the ground station internal delays as used by the Cluster specific software: With the migration from X.25 to TCP/IP communications protocol the Telemetry Processors (TMP)

13

ESOC Data Products in the CAA

221

were replaced at the ground stations by the Telemetry & Telecommand System (TMTCS). The new TMTCS takes the ground station internal delays already into account before applying the ERT stamping. Therefore the table used by the Cluster specific software shows 0 (zero) ground station delay. As the time correlation is essential for the data processing these data could be of relevance for the users. Data Base Changes The Cluster data base changes have been recorded as “Data Base Change Requests” and implemented as a Microsoft Access application. The data base includes all the telemetry and the telecommands changes performed at platform and payload level over the lifetime of the mission. Any change of threshold for soft and hard out-of-limits or calibration curves could be traced with this data base. Data Recall Log-File and Data Gap Analysis The Cluster Data Dissemination System (CDDS) software includes a data gap tool that identifies missing frames within the science and housekeeping data streams. When the transfer of data from the ground station to the CDDS was interrupted for whatever reasons it is always possible to recall the data from the ground station. Based on the gaps identified by a gap check tool the Flight Control Team initiates a data recall if possible. The tool identifies gaps greater than 10 Frames for housekeeping data – VC0 data 100 Frames for nominal science data – VC2 data 700 Frames for burst science data – VC3 data

The data file provides lists for each spacecraft about (a) all recalls performed, differentiating between the various virtual channels and (b) the data gaps remaining which could not be filled. The resolution is greater than 10, 100 or 700 frames for VC0, VC2 or VC3 data respectively. If known the cause of a data gap is mentioned. Science Data Return Ratio for Frames and Formats Each file covers one planning period for all four spacecraft as defined by the Observation Request (OBRQ) file. For each individual acquisition interval (Normal Mode, Burst Mode or Wide Band Data operations) within this OBRQ the number of expected and actual received frames and formats is given and their ratio is calculated. Each file also contains the summary tables:

The total science data return The fraction of Normal Mode and Burst Mode The fraction of live and recalled data The number of ground station passes for Wide Band Data (WBD) operations

The first available science data return ratio file is for the planning period PP74, 14Oct-2001, when the above software tool was made available for the calculation. This information could be used to identify which observation requests were affected by a data loss.

222

J. Volpp and D. Sieg

Science Data Volume and Data Return Ratio Summary This file provides an overview about the science data recovery, i.e. the data volume expected per planning period and the achieved data return ratio. Based on the OBRQ-file received from the JSOC the data volume, i.e. the amount of Normal Mode and Burst Mode data is derived first. Using the science data return ratio file described above secondly the data return ratio for each of the 4 spacecraft is calculated Normal Mode and Burst Mode. Combining these results the overall science return computation provides for each planning period:

The OBRQ data volume split for Normal Mode and Burst Mode The minimum possible simultaneous data return ratio The maximum possible simultaneous data return ratio The average data return ratio for all four spacecraft The worst single spacecraft data return ratio

The data are available as Excel sheet and graphs. The science data return ratio summary file starts with planning period PP74 on 14-Oct-2001.

13.6 Summary Beside a brief description of the Cluster mission and its evolution during the last 8 years this paper explains all data products provided by ESOC to the CAA. These data can be categorized in four groups: Regular operations reports, anomaly reports and specific technical notes. Spacecraft platform data such as summary of operations and trend analysis of relevant parameters. Flight dynamics data such as orbit and attitude information, manoeuvres and eclipses. Ground segment related ground station passes, drift of the on-board clock with reference to the ground and last not least all science data gaps and the overall science data recovery.

References 1. Chanteur, G., and C.C. Harvey (1998), Spatial interpolation for four spacecraft: Application to magnetic gradients, in Analysis Methods for Multi-Spacecraft Data, edited by G. Paschmann and P. W. Daly, ISSI Sci. Rep. SR-001, pp. 371–393, Int. Space Sci. Inst., Bern. 2. Credland, J., Mecke, G. and Ellwood, J.: The Cluster Mission: ESA’s spacefleet to the magnetosphere, Space Science Reviews 79: 33–64, (1997). 3. ESA (1993); Cluster Experiment Interface Document (EID) Part A; chapter 3.3 On-board data handling interfaces; CL-EST-RS-0002. 4. ESA/ESOC (2000); Data Delivery Interface Document (DDID); CL-ESC-ID-2001. 5. Ferri, P. and M. Warhaut (1997), Cluster Mission Operations, P.475–485 in The Phoenix and Cluster Missions, edited by C.P Escoubet, C.T. Russel and R. Schmidt, Kluwer Academic Publishers, 1997, ISBN 0-7923-4411-1. 6. Yearby et al., this issue.

Part II

Tools for the CAA Data Analysis

Chapter 14

QSAS, QM Science Analysis Software A.J. Allen

Abstract QSAS is a powerful tool for manipulating, combining and displaying multi-spacecraft and multi-instrument data. It uses the CAA metadata to its fullest potential to allow efficient handling of units and reference frames. It can also save and restore calculations and plot designs for reuse on different data intervals.

14.1 Overview QSAS treats all data as objects that may be dragged and dropped between its various windows. These objects include all the relevant metadata and time tags for the data values and include key quantities necessary if the data is to be combined with other data. In particular QSAS understands the notion of frames of reference for vectors (and tensors), knows about the various vector representations (Cartesian, polar, . . . ) and transparently offers such services to the user. It further understands the units of data objects and can convert units on the fly when necessary. Handling in situ space plasma data requires careful attention to the time-series nature of most of the data. Time plays the dual role of event identification and a continuous variable, and any time series data object in QSAS can be used to specify a time interval. The main components of QSAS are: Data Ingestion. QSAS can ingest data via its own database management system

that transparently concatenates files. QSAS can also open up data files directly, including CDF, CEF and other ASCII files. ASCII files not conforming to Cluster or ISTP standards can usually be ingested by the provision of a simple, handedited header file. Data Export. QSAS can export the data it ingests, together with the result of any analysis, to CSDS-compliant CDF files, CAA standard CEF files or to ASCII files conforming to Cluster standards. This facilitates the exchange of data and analysis results with other teams and other CAA aware software environments. A.J. Allen () Blacket Laboratory, Imperial College, London, SW7 2AZ, UK e-mail: [email protected] H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 14, c Springer Science+Business Media B.V. 2010

225

226

A.J. Allen

Data Analysis. The results of all analysis operations are available either for

further analysis, plotting or export. Analysis operations supported include: – Join. This joins multiple time series onto a common set of time tags. Options are available to control the join algorithm and gap handling, while the target time tags can be either regularly spaced (to facilitate power spectral analysis) or the time tags from a data object. – Calculator. The QSAS calculator provides a powerful interface in which users can build up complex expressions out of arithmetic unary and binary operators, all of which exploit the QSAS data object paradigm to transparently deal with different kinds of data (scalars, vectors, matrices, and sequences of these types). Additionally, the calculator is built around the QSAS Join interface, so that joining and arithmetic operations are performed in a self-contained interface. Results from one operation can be used as inputs to another to allow construction of processing chains. – Other built in operations. Simple filters are also available for removing or changing fill values, removing values greater or less than a specified value and for changing units. – Object sub-sampling. When a data object is dropped into an operation, it is always possible to sub-sample it via a simple pull-down menu. This allows talking a subset or average of a sequence, as well as component of a vector or a slice from an array. Interactive selection of data slices and time intervals is thus directly available in analysis or plot windows. – Plug-ins. Generic software can never meet 100% of all users’ needs. The QSAS plug-in interface permits user-written modules to be dynamically loaded, and many specialised modules have been written by other teams for inclusion in the QSAS distribution. Display. QSAS provides an interactive, flexible display capability for time series,

spectrograms and XY plots. Plot output can be directed either to the screen or plot files. The user has powerful control over the layout and plot style to generate publication quality plots. Save/Restore. QSAS can save and restore individual window settings (e.g. plot layouts, calculator algorithms, etc.) or an entire session, both with and without the currently held data objects. This protects users against errors or crashes, facilitates repeated use of analyses and plot layouts, and allows users to return to a full QSAS session. Save sessions are portable, and thus also provide a useful bug-reporting element.

14.2 Case Study On 5 July 2001 the Cluster spacecraft were in the flank moving in and out of the Magnetopause. This contact discontinuity must be in pressure balance with dominant contributions from the magnetic field and ions, and smaller contributions from

14

QSAS, QM Science Analysis Software

227

electrons, bulk turbulence and flow deflection. A useful case study is therefore to determine the total ion plus magnetic pressure as a function of time during Magnetopause crossings. Using CAA data in CDF format we can ingest the ion density, velocity, perpendicular temperature and 3D ion distribution from CIS and magnetic field vector from FGM using the QSAS Data Selector, Fig. 14.1. The calculator can then be used to construct the particle pressure nkT? and magnetic pressure B2 =20. However, using the CAA data to generate the ion pressure in this way results in units of 1012 m3 J while the magnetic pressure comes out in nT2 H1 m. To combine these would require converting both from the CAA units into a common unit, and QSAS will do this automatically. When creating the sum of two quantities, QSAS will ensure that the second argument is converted on the fly into the units of the first using the attached SI conversion attribute. In this example we first convert the magnetic pressure into nPa and then the calculator automatically converts the ion pressure into the same units before addition. Nevertheless, in order to plot the ion pressure on the same graph, we also convert it explicitly into nPa, Fig. 14.2. The resulting plot, Fig. 14.3, shows the magnetic field components and magnitude, proton density and bulk flow velocity followed by the proton energy spectrum and the three pressures. The total pressure is shown in black, with the magnetic contribution in blue and the ion contribution in red. It is easy to see the total pressure remaining fairly steady across most of the interval, even though the relative contributions from magnetic fields and protons is changing dramatically. It is also possible to discern that there are periods when a high proton pressure corresponds to hot, lower density ions (for example, from 17:50 to 18:00) as well as periods of denser, cooler plasma (for example, from 15:50 to 16:00). Interestingly, the total pressure remains steady across the hot plasma regions but appears unsteady and a little lower during intervals where denser lower energy ions dominate the total pressure. This indicates that in the latter regions one of the neglected pressure contributions is non-negligible. It is now trivial to repeat the analysis of the first event on a new time interval. If one clears the QSAS working list and drags a new time interval object to the Data Selector window and presses “Get Data”, the same selection of variables that we used in the first case study will be ingested for this new event. It is then only necessary to press the “Do” button on the Calculator to repeat the sequence of calculations already set up, and then press “Plot” in the plot window, to display the results from the new calculation. This ability to save a QSAS calculation or plot design, or even a whole session, permits an analysis chain to be kept for repeat use.

14.3 Discussion QSAS is a science analysis and display tool that is designed to use the CAA provided metadata intelligently. It understands about the different reference frames and representations of vectors and tensors, while also ensuring that data are always handled

Fig. 14.1 Data Selector window

228 A.J. Allen

Fig. 14.2 The calculator window with the QSAS main window and working list of data behind

14 QSAS, QM Science Analysis Software 229

230

A.J. Allen SC 1 20

Bxyx (nT)

0 Bx (GSE) B_y B_z

–20 30 ⎪B⎪(nT) 20 10 6 Np (cm –3)

4 2 300

⎪v⎪(km/s) 200 100 15h40m 16h0m 2001-07-05

20m

40m 17h0m Universal Time

20m

40m

106 particles/s

Energy (eV)

CIS–CODIF 3D HS Proton distribution 10m 1000 100

105 104 1000

15h40m 16h0m 2001-07-05

20m

40m 17h0m Universal Time

20m

40m

20m

40m

18h0m

P_B + nkT_P 0.6 P (nPa) 0.4 0.2 15h40m 16h0m 2001–07–05

20m

40m

17h0m

Universal Time

Fig. 14.3 Plot for sample event showing some ingested CAA parameters and the pressure values calculated in QSAS. In the bottom panel the total, magnetic and proton pressures are in black, blue and red respectively

in the correct units. It ensures the results of analysis also retain the correct frame and units information for further manipulation. It can read and write data files in CAA, CSDS, ISTP and other syntax in both CDF and CEF formats. The results of analysis can themselves be exported for

14

QSAS, QM Science Analysis Software

231

ingestion into other packages, while externally written analysis modules can also be attached to QSAS on the fly through its plug-in interface. QSAS is available for Mac OSX, Unix and Linux platforms. It can be downloaded from http://www.sp.ph.ic.ac.uk/csc-web/csc.html as either source or binary distribution, and is available under a GNU public licence.

Chapter 15

Flexible Tools for Accessing the Cluster Archives E. Gamby, J. De Keyser, and M. Roth

Abstract Nowadays, multi-spacecraft analysis is becoming common practice. The lack of homogeneity in data archive structure, access types and data formats results in a lot of time spent to search, retrieve and reformat data. We have developed a software model dealing with these issues. In this model, we integrate an abstraction layer for accessing and caching archive data from different providers. A second abstraction layer allows converting the specific data formats into a common working format. This model has been implemented in the MIM software, developed by the Belgian Institute for Space Aeronomy. In particular, it provides a framework for accessing the CAA (Cluster Active Archive) and for exploiting Cluster data with multipoint analysis tools such as, e.g., gradient computation and magnetopause reconstruction.

15.1 Introduction Space science data are usually stored in large, centrally managed archives. Scientists that access these data must deal with a plethora of access types, archive structures and data formats. This lack of homogeneity makes it more difficult to implement algorithms based on data from multiple sources. We present a software model that we have developed to handle this situation.

15.2 Software Infrastructure The whole sequence of activities that scientists perform on data can be seen as a single workflow. In this workflow, we identify three main processes: 1. Accessing the remote archive and retrieving data 2. Converting these data into a common format 3. Data analysis and visualization E. Gamby (), J. De Keyser, and M. Roth Belgian Institute for Space Aeronomy, Brussels, Belgium e-mail: [email protected] H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 15, c Springer Science+Business Media B.V. 2010

233

234

E. Gamby et al.

The output of the second process provides a uniform data model to the scientific algorithms of the third one, so that these algorithms don’t have to deal with data heterogeneity. We base our software model on this three-fold decomposition.

15.2.1 The Retrieval Process This process (see Fig. 15.1) is responsible for downloading data from remote archives. Each connection to such an archive is called a channel. A channel specifies where and how to find spacecraft data for a given time interval. The process keeps a queue of pending user requests, or jobs. A channel has two software components: 1. A data source, which encapsulates the communication protocol used by the remote archive and which hides the access details from the other processes. 2. A local cache, in which the previously downloaded data are stored. This caching strategy aims at saving time and bandwidth when the same data are requested several times.

15.2.2 The Formatting Process The formatting process (see Fig. 15.2) merges the different formats into the same data model and organizes these data into datasets. Each conversion is steered by a format description that specifies how to extract the data and metadata from the source format. A dataset is a collection of instrument data, such as measurements of the magnetic field, of ion densities . . . and of the associated ancillary data, such as the spacecraft position and velocity. These quantities are called data items in our software model.

Fig. 15.1 Retrieval and caching of data from various data sources. Black arrows represent associations between concepts; white arrows are specializations of concepts

15

Flexible Tools for Accessing the Cluster Archives

235

Fig. 15.2 Importing science data into the MIM internal format

Fig. 15.3 Analyzing and visualizing data. Algorithms can both consume and produce datasets, making the system very flexible

15.2.3 The Analysis and Visualization Process This process includes the scientific algorithms and visualization techniques (see Fig. 15.3). As all algorithms share the same data model, the result of one of them can be used as input of another one. Moreover, adopting such a flexible model eases the combination of data issued from different sources (spacecraft or numerical simulations) in various algorithms. For instance, multi-spacecraft data can be fed into algorithms specifically tailored for multi-point analysis and be compared to numerical predictions.

15.3 The MIM Implementation The above model is implemented into the MIM (Manager of Interactive Modules) software, a data management and modeling tool developed by the Belgian Institute for Space Aeronomy. It provides an interactive environment for accessing and processing data from a multitude of experiments. MIM is a MATLAB application, written in an object-oriented way. It consists of a set of modules, amongst which the channel manager and the dataset formatter, respectively

236

E. Gamby et al.

implementing the retrieval and formatting processes. Other modules are dedicated to scientific algorithms, such as multi-spacecraft gradient computation and magnetopause reconstruction. In the channel manager, we define channels for accessing the CAA (Cluster Active Archive) and CSDS (Cluster Science Data System) http-based archives, as well as the Themis ftp site. Since data download can lead to significant wait times, cache hits are important. A high cache hit rate can be achieved as scientists often work for a prolonged time with a limited set of events. The dataset formatter is the module responsible for creating format descriptions and applying them to retrieved data. We have defined templates to import CDF (Common Data Format), HDF (Hierarchical Data Format) and CSV (Comma Separated Values) files. In particular, we can import CDF data from the Cluster and Themis repositories. Amongst the scientific modules, algorithms are available to process multi-point data, e.g. gradient calculation [2], magnetopause reconstruction [1], time delay analysis . . . An example of a calculation of the least-squares gradient during a Cluster pass through the inner magnetosphere is shown in Fig. 15.4.

Fig. 15.4 Calculation in MIM of the least-squares gradient of plasma density obtained from the WHISPER instruments on the four Cluster spacecraft during a pass through the inner magnetosphere on 7 August 2001; perigee is encountered at about 4 RE around 08:05 UT. The increasing and decreasing density before and after the perigee crossing reflect the entry into the outer regions of the plasmasphere. The model assumes the gradient to be locally constant. The distance over which the gradient can safely be considered constant is estimated automatically. The output is the computed gradient vector with its error margins

15

Flexible Tools for Accessing the Cluster Archives

237

15.4 About the Importance of Metadata In the software model introduced in the previous sections, extensive use is made of metadata. Indeed, meta-information is very useful to automate data processing. For example, normalizing data from heterogeneous sources requires information such as the units or the coordinate system in which data are expressed. Moreover, data visualization can benefit from metadata such as scale type (linear or logarithmic), minimum and maximum values, axis labels . . . Unfortunately, so far, no standard dictionary of metadata has been defined. However, with projects such as ISTP/IACG, CSDS, SPASE and the Cluster Metadata Dictionary, there is an ongoing effort towards standardization: The ISTP/IACG (International Solar-Terrestrial Physics/Inter-Agency Consulta-

tive Group for Space Science) Guidelines describe a standard way of structuring data in CDF files [3]. The CSDS (Cluster Science Data System) is an extension of the ISTP guidelines including science tool driven metadata that is machine readable. The SPASE (Space Physics Archive Search and Extract) data model provides a rigorous hierarchy of metadata but is so far essentially oriented to data access and retrieval [5]. The Cluster Metadata Dictionary, which has been specifically developed for the Cluster mission, borrows concepts from the two preceding dictionaries [4] to provide a complete set of metadata accounting for both the retrieval and the science processing in advance of their adoption by SPASE. One of the main advantages of the latter dictionary is that it provides metadata down to the level required by science tools, while very often, standards tend to stop at metadata needed for search and delivery.

15.5 Conclusion Manipulation and interpretation of data from multiple sources require specific functionalities to deal with heterogeneities in data formats and archive structures and with the lack of standardization amongst metadata. These functionalities are often not available in standard data processing environments. We have proposed a software model that integrates at least some of the required functionalities. This model has been successfully implemented in the MIM application (downloadable from http://www.spaceweather.eu/software/mim). Acknowledgements The authors wish to thank the Belgian Science Policy (BELSPO) and the Solar Terrestrial Center of Excellence (STCE) for their financial support.

238

E. Gamby et al.

References 1. De Keyser, J., Roth, M., Dunlop, M.W., R`eme, H., Owen, C. J., Paschmann, G.: Empirical reconstruction and long-duration tracking of the magnetospheric boundary in single- and multispacecraft contexts. Ann. Geophys. 23, 1355–1369 (2005) 2. De Keyser, J., Darrouzet, F., Dunlop, M.W., D´ecr´eau, P.M.E.: Least-squares gradient calculation from multi-point observations of scalar and vector fields: methodology and applications with Cluster in the plasmasphere. Ann. Geophys. 25, 971–987 (2007) 3. Goddard Space Flight Center: ISTP/IACG guidelines. Space Physics Data Facility Website. http://spdf.gsfc.nasa.gov/sp use of cdf.html (2006). Accessed 2 June 2008 4. Harvey, C.C., Allen, A.J., D´eriot, F., Huc, C., Nonon-Latapie, M., Perry, C.H., Schwartz, S.J., Eriksson, T., McCaffrey, S.: Cluster Metadata Dictionary, version 2.3. The Cluster Active Archive Website. http://caa.estec.esa.int/documents/DataDic.pdf (2008). Accessed 2 June 2008 5. SPASE Consortium: A Space and Solar Physics Data Model, version 1.2.1. SPASE Website. http://www.spase-group.org/data/doc/spase-1 2 1.pdf (2008). Accessed 2 June 2008

Chapter 16

AMDA, Automated Multi-Dataset Analysis: A Web-Based Service Provided by the CDPP C. Jacquey, V. G´enot, E. Budnik, R. Hitier, M. Bouchemit, M. Gangloff, A. Fedorov, B. Cecconi, N. Andr´e, B. Lavraud, C. Harvey, F. D´eriot, D. Heulet, E. Pallier, E. Penou, and J.L. Pinc¸on

Abstract We present AMDA (Automated Mutli-Dataset Analysis), a new service recently opened at CDPP. AMDA is a web-based facility for on line analysis of space physics data coming from either its local database or distant ones. This tool allows the user to perform on line classical manipulations such as data visualization, parameter computation or data extraction. AMDA also offers innovative functionalities such as event search on the content of the data in either visual or automated way, and the generation, use and management of time-tables. These time-tables can be seen as a brick of up-coming virtual observatories in space physics, and could be used as an input to extract data from other databases, Cluster Active Archive in particular.

16.1 Introduction For in depth studies of plasma objects such as the Earth’s magnetosphere or the solar wind, it is necessary to analyse multi-points and multi-instruments measurements. In practice, that means that researchers have to exploit together data coming from many C. Jacquey (), V. G´enot, M. Bouchemit, M. Gangloff, A. Fedorov, N. Andr´e, B. Lavraud, C. Harvey, E. Pallier, and E. Penou CDPP/CESR, CNRS/Universit´e Paul Sabatier, 9, avenue du colonel Roche, 31028 Toulouse, France e-mail: [email protected] E. Budnik Noveltis, 2, Avenue Europe, 31520 Ramonville Saint Agne, France R. Hitier Co-Libri, Cremefer 11290 Montr´eal, France B. Cecconi LESIA, Observatoire de Paris-Meudon, 5, place Janssen, 92195 Meudon F. D´eriot and D. Heulet CNES, Centre spatial de Toulouse, 18 avenue E. Belin, 31401 Toulouse, France J.L. Pinc¸on LPCE, Laboratoire de Physique et Chimie de l’Environnement, 45071 Orl´eans, France H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 16, c Springer Science+Business Media B.V. 2010

239

240

C. Jacquey et al.

sources which can initially be heterogeneous in their internal organisation, their description or their coding format. At the scale of the individual researcher, or even of a laboratory, this requires an important investment and consumes a large amount of time and energy. This became critical when the ISTP program and the CLUSTER mission were launched. Facing this challenge, several data centres emerged with the support of the research institutions and the space agencies, such as in USA (SPDF, Space Physics Data Facilities), in Japan (DARTS, Data Archives and Transmission System) and in Europe (CAA, Cluster Active Archive). The CDPP (http://cdpp.cesr.fr/, Centre de Donn´ees de Physique des Plasmas) is the French national centre for space physics data. It was jointly created by CNES (Centre National d’Etudes Spatiales) and CNRS (Centre National de la Recherche Scientifique) in 1998. Recently, the CDPP has opened a new web-based service called AMDA (Automated Multi-Dataset Analysis, http://cdpp-amda.cesr.fr/). This service offers “classical” functionalities such as data extraction, merging or visualisation but also innovative ones such as user-edited automated search or computation on the content of the data as well as the creation and the use of time-tables. The aim of this paper is to present this service. We first describe what are the needs met in space physics data analysis. Then we present the functionalities of AMDA and illustrate them with a use case of the study of the cross-tail current dynamics.

16.2 Analysis of the Needs for Data Exploitation in Space Physics As schematically depicted in Fig. 16.1, case studies or statistical studies in space physics are generally performed as follows: (1) The first step is generally the search of cases of interest. This search can be performed through examination of quicklooks or using quantitative criteria on the content of the data. (2) When an event or a set of events has been identified, the second step consists of retrieving and extracting the numerical data. (3) Before analysis with the user preferred software, the data may need to be formatted depending on their original format. This step benefits of the more and more common use of standard formats as CDF, CEF or NetCDF. (4) Then, basic treatments generally need to be applied e.g. bad value filtering, gap interpolation or merging data at the same baseline, . . . (5) In this step, the user performs various manipulations on the data. Many of them are specific but commonly

1

2

Event Search

Collecting the data

3 Formatting the data

4

5

Preparing the data

Data analysis

Fig. 16.1 Schematic illustration of the successive steps followed in a study based on data analysis in space physics

16

AMDA, Automated Multi-Dataset Analysis

241

include visualisation, parameter computation and comparison to standard models such as magnetic field models. It is only when these steps have been accomplished that the interesting work, i.e., the interpretation of the observations, starts. Because there are many datasets available, most of which coming from different origins, formatted in various ways, the work required before interpreting the observations often consumes large amount of time and energy from both researchers and software engineers. This has direct impact on the data exploitation, i.e. on the scientific return of the experimental investment.

16.3 AMDA (Automated Multi-Dataset Analysis) The main objective of AMDA is to help the user in performing the different steps described above. In order to enable complex operations on data, AMDA has been built around elementary objects: the parameter and the time table. A first and immediate difference relatively to most databases is that the concept of data files does not exist in AMDA at the user level. The user manipulates parameters without taking care of the files where they come from. Inside the AMDA system, the parameter is associated with properties (scalar, vector, tensor, units, . . . ) and with corresponding options (decomposition, frame of reference, . . . ). The second useful object in AMDA is the time table which is basically a collection of time intervals. The functionalities of AMDA allow to use and to couple these two classes of objects. They are interactively activated through web-interfaces by the user who edits requests and collects the results in his/her workspace. All the requests and results may be saved for further sessions as soon as the user has registered. We now briefly describe these functionalities. Parameter builder. In the “My parameters” interface, the user can compute

new parameters by editing mathematical expression combining existing parameters. Heterogeneous time bases are handled by AMDA transparently to the user. He/she freely chooses the time resolution of his/her final new parameter and can also manage data gaps. Once a new parameter has been created it can be used as any pre-existing parameter. Remote data access. Following the Virtual Observatory paradigm AMDA goes further in giving a direct access to parameters hold in distant databases (noticeably CDAWeb is available through AMDA). In the “External Data” interface, the user can browse through the parameters of the connected distant databases, select the desired ones and finally save them in his/her own external data tree. This data tree, with the usual hierarchy mission/instrument/dataset/parameter, will then be added to the existing local parameters. The parameters referenced in the external data tree can be used thereafter like any pre-existing ones. Data download. In the “Download Data” interface, newly created, remotely accessed or local parameters can be equally downloaded (or exported) by the user in different formats (plain ASCII, CEF, CDF) and with a chosen resolution.

242

C. Jacquey et al.

Time series may optionally be joined on a common time basis before exporting. The data merging is performed either by averaging or interpolating, depending on the time resolution of the original datasets. As it was mentioned above, the user does not need to care about new data downloads. It is fully automated according to the requested time interval and the parameters needed. Data visualisation. In the “Plot Data” interface, the user can edit a figure combining any available parameter with the desired options (reference frame, scaling, . . . ). The request can be saved in such a way that the edited figure can be applied to any time interval chosen by the user. Note that the interface provides the option to apply time shifting for solar wind propagation. Widgets also allow to shift the figure to the following/preceding time interval. This functionality also accepts time tables as input. In this latter case, widgets allow to skip the figure from a time interval to the next/previous one. Visual search. While browsing the requested figures, the user can retain intervals when a special feature, visually determined, occurs. He can do so by double clicking at the start/stop of the event (see Fig. 16.2). Then an interface is provided to store this interval in a list; that is, the user can create a time table by visual inspection. Automated conditional search. A second way to create time tables is by selecting time intervals when a particular mathematical condition applied on given parameters is fulfilled. In the “Conditional Search” interface (see Fig. 16.3), the user can edit his/her condition with mathematical functions and logical operands (>, <, &, j). This condition is applied on a given time window and only subintervals fulfilling the condition are retained to populate the time table. The time window may itself be a single time interval or a time table, offering the possibility to perform successive automated searches. Time-table manager. In the “My Time Tables” interface (see Fig. 16.4), time tables may be subsequently edited and it is possible to apply operations on a single time table: extend/shrink/shift by a given duration; or on multiple time tables: union or intersection. The time table can be exported in ASCII format or in VOTables compliant to the IVOA standards.

16.4 A Use Case To illustrate the use of AMDA, let us consider the case of a user who wants to study from the CLUSTER data the local properties of the cross-tail current inside the neutral sheet with respect to interplanetary conditions and geomagnetic activity. He wants to focus on events when the current sheet is encircled by the CLUSTER satellites. As illustrated in Fig. 16.5, a possible way to render this situation is to consider that this condition is fulfilled when two spacecraft are above the neutral sheet (i.e. BX > 0) and the two other ones are below (i.e., BX < 0).

16

AMDA, Automated Multi-Dataset Analysis

243

Fig. 16.2 Partial view of the “Plot Data” interface when the visual event selection functionality is activated. Magnetic field and electron spectra of THEMIS-B and THAMIS-D are represented here. A plasmoid event has been selected and recorded in the time table THEMIS PLASMO

In the “Conditional Search” interface of AMDA (Fig. 16.2), the user first performs an automated search of the time intervals when the CLUSTER spacecraft encircled the neutral sheet as defined above. He/she edits the following criteria: BX1 BX2 BX3 BX4 > 0 AND min fBX1 ; BX2 ; BX3 ; BX4 g < 0 AND max fBX1 ; BX2 ; BX3 ; BX4 g > 0 AND XGSM1 < 10 He/she applies them to the CLUSTER data during the tail seasons, for example the one of the year 2002 (August to October). To avoid selecting too brief events, the

244

C. Jacquey et al.

Fig. 16.3 Partial view of the “Conditional Search” interface. The criteria edited in the “Search Condition” window corresponds to the use case presented in Section 16.4

filter is applied by computing on the fly 1 minute-averaged FGM data. AMDA then generates a first time table (hereafter TT1) listing all the time-periods when this neutral sheet encirclement condition is fulfilled. In the case of the 2002 tail season, 216 events are found. Starting from this time table TT1, the user can then undertake several possible actions. Some examples: Extraction of sub-database corresponding to CLUSTER neutral sheet

encirclement events. In the “My parameter” interface, the user can formulate additional parameters such as the plasma beta, the total pressure for example. Then, in the “Download Data” interface, he/she can select the parameters of interest for his/her study and then download them in merged files corresponding to the time-intervals recorded in the time-table TT1. So, he/she constitutes a

16

AMDA, Automated Multi-Dataset Analysis

245

Fig. 16.4 Partial view of the “My Time Tables” interface

Fig. 16.5 Illustration of the neutral sheet encirclement by the CLUSTER constellation. The profile of the BX component through the current sheet is shown on the right side of the right panel. BX is positive above the neutral sheet and negative below

sub-database of the CLUSTER neutral sheet encirclement events that he/she can statistically treat off-line with his/her own tools and software. In view of analysing the context of these events, the user can generate a new time table TT2 by extending the time intervals of TT1 and download the corresponding complementary sub-database.

246

C. Jacquey et al.

Extraction of corresponding sub-database of solar wind data. To download

the interplanetary data obtained by ACE, the user can first extend the Time-Table by 60 min and shift it by 45 min, corresponding to a rough averaged time-delay of solar wind propagation. (The on-the-fly computation of the solar wind propagation time-delay is an up-coming functionality of AMDA). Filtering the event collection with additional criteria. As explained in Section 16.3, the “Conditional Search” interface allows to perform successive searches. The user can use this possibility if for instance he/she wants to: (i) Find the events when all the CLUSTER spacecraft were embedded in the central plasma sheet (with a condition on the plasma beta parameter) (ii) Classify the events as a function of their location in the XYGSM plane, the observed flows, the wave activity level, the geomagnetic activity . . . (iii) Find the events when GEOTAIL and/or POLAR were also imbedded in the plasma sheet (with condition on their location and their plasma measurement) in order to perform multi-scale studies These automated searches would generate new time tables which could be thereafter used to constitute new sub-databases or to produce figures via the “Plot Data” interface.

16.5 Conclusion and Perspectives AMDA is a web-based service developed with the aim to help researchers in space physics. AMDA offers functionalities to access and analyse multi-dataset in a transparent fashion and allows perform event search. This tool has already been used in various studies [1–3]. AMDA is in continuous development. Future developments concern new analysis tools, new products, enhancement of interoperability and the creation/management of catalogues. AMDA is evolving in the Virtual Observatory paradigm. It gives a direct access to data from distant databases and includes a connection layer compliant with the SPASE (http://www.spase-group.org/) standards. AMDA produces, manipulates and uses time tables. The time tables can be seen as one of the primary brick to be used for the interoperable exchanges in space physics. This idea encouraged to start a collaboration between databases (AMDA and CAA) on one hand, and generic tools (QSAS and CL) on the other hand. Acknowledgements The authors would like to thank P. Louarn, C. Perry, T. Allen, S.F. Fung and D. Bilitza for helpful discussion and the CDPP User Committee for having tested AMDA.

16

AMDA, Automated Multi-Dataset Analysis

247

References 1. G´enot V., E. Budnik, C. Jacquey, I. Dandouras, E. Lucek (2008a), Mirror mode events observed with Cluster in the Earth magnetosheath: statistical study and IMF/solar wind dependence, submitted to Advances in Geosciences (2008) 2. G´enot V., E. Budnik, P. Hellinger, T. Passot, G. Belmont, P. Travnicek, P.-L. Sulem, E. Lucek, and I. Dandouras (2009), Mirror structures above and below the linear instability threshold: Cluster observations, fluid model and hybrid simulations, Ann. Geophys., 27, 2, 601–615 3. Louarn P., C. Jacquey, E. Budnik, and V. Genot (2006), The CDPP, FGM, CIS and STAFF teams, The active plasma sheet: definition of ‘events’ and statistical analysis, Proceedings of the 8th International Conference on Substorms, March 27–31, 2006, Banff Centre, Editors M. Syrjaeso and E. Donovan

Chapter 17

Cluster CAA Module for PaPCo ˚ J. Faden, A. Asnes, R. Friedel, M. Taylor, S. McCaffrey, C. Perry, and M.L. Goldstein

Abstract A PaPCo module for visualization of data from the CAA has been developed. This module retrieves data from the CAA web interface, and allows for discovery and plotting of new datasets. PaPCo is modular, open source IDL software that uses plug-in modules to bring new datasets on to a stack of time series plots (www.papco.org). PaPCo includes modules for plotting data from Cluster/PEACE and Cluster/RAPID, CDA Web data which includes Cluster Prime Parameters, and various modules from CRRES, POLAR, GPS, and many other spacecraft. The Cluster CAA module is presented, as well as a brief description of PaPCo’s use and installation procedure.

17.1 Introduction A CAA Module for the IDL plotting software “PaPCo” has been developed, allowing data from the CAA to be plotted and used for analysis. This article has two audiences: those who simply want to use the software to access data, and those who want to have full access to the capabilities and need developer documentation as well as user documentation.

J. Faden () Cottage Systems, Iowa City, IA, USA e-mail: [email protected] ˚ A. Asnes, M. Taylor, and S. McCaffrey ESA/ESTEC, Noordwijk, The Netherlands R. Friedel Los Alamos National Labs, Los Alamos, NM, USA C. Perry Rutherford Appleton Laboratory, Chilton, Didcot, Oxon, UK M.L. Goldstein NASA/GSFC, Greenbelt, MD, USA

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 17, c Springer Science+Business Media B.V. 2010

249

250

J. Faden et al.

17.1.1 What Is PaPCo? PaPCo is modular, IDL-based software for making a stack of panels for analysis of data from various sources on a common time axis. Software modules created at many different institutions plug into the “PaPCo Core” to draw graphic data products on panels of the page. PaPCo was first developed for the CRRES mission in 1993 and has roughly 80 modules for instruments on over 15 spacecraft, and data sources like CDAWEB. PaPCo has been improved specifically for use with Cluster data over the past three years. Additionally, PaPCo has mature code for page layout, time axis controls, event list generation, batch processing, reformatting data and many other useful features. Most PaPCo modules now use PaPCo’s conventions for storing data. A dataset in this “internal data model” is a structure with the data itself stored in a multidimensional array, other datasets that tag the data dimensions and conventional name-value pairs to store metadata such as labels, physical units, typical ranges. Using this uniform model, modules can use PaPCo infrastructure for reducing and slicing data, caching, and accessing one another’s data.

17.1.2 CAA Module for PaPCo The CAA is a large store of useful datasets in a common data model. This allows for a PaPCo module to be created that can leverage off this store to provide a powerful tool for doing Cluster analysis. The module tries to achieve three goals. First, to provide a discovery capability that allows the user to discover useful data stored in the CAA. The analyst can browse list of roughly 800 available data sources, filtering the list by spacecraft, instrument and product type components of dataset identifiers. The dataset identifier and description are searchable using a simple regular expressions search. Also, we provide a “parameters search” which searches through descriptions and metadata for each plottable parameter within the datasets. Second, we automatically download data for the user using the Sinead McCaffrey’s “Command Line Interface,” an http-based interface for requesting data from the CAA. PaPCo’s File Cache Tool is used to show what is remotely and locally available, and availability plots can be made summarizing data availability. Third, data from the CAA is adapted to PaPCo’s internal model for plotting and slicing. High dimensionality datasets like PEACE 3DX can be viewed as TimeEnergy spectrograms, and then sliced to view pitch angle distributions. Data plotted by the module is available to other modules for further analysis, and codes written for the data model can be used.

17

Cluster CAA Module for PaPCo

251

17.2 Using the Module The process of installing PaPCo is described later in this article, and for now we’ll assume that the software is installed with a version of the CAA module from the time of this writing. This section is intended to show the basic idea of how the module works, then in later sections we will look in depth at the module controls and installation process. The analyst launches PaPCo, and the PaPCo main widget appears. At the top left, a button labeled “Cluster CAA” (A) provides access to the “panel editor” for the CAA module (E), which is a GUI that adds and configures the plot panel (Fig. 17.1). The analyst operating the software clicks on Cluster CAA button (A), and the panel editor pops up, showing all available datasets. There are roughly 800 datasets at the time of this writing. The analyst then filters the list to locate the dataset for plotting, and selects the parameter to plot. For example, the analyst selects the parameter “N p C1 PP CIS,” which is CIS density data. The time axis range is controlled using the time control on the PaPCo main widget (B). For example, “9/13/2002” is entered to set the time axis to display data for that day. The analyst hits the “Draw” button (C) and several things will happen at this point. If this is the first time he is accessing the CAA, and he will need to enter his username and

Fig. 17.1 PaPCo Main Widget with CAA module panel editor. Pushing the module’s button (A) brings up its panel editor (E). This adds panels to the stack of panels to be plotted (D). Time is controlled on the main widget (B), and the Draw button (C) draws the page

252

J. Faden et al.

Fig. 17.2 PaPCo plot with data from PEACE, FGM, and CIS. A vertical slice at 14:58 shows slices of selected panels

password. The module checks to see what data is available, and if data is available, then a request is made to the CAA to download the data. The CAA may satisfy the request immediately and send over the data. In this case, data is sent over and copied into the local store of files. PaPCo routines for drawing line plots and spectrograms are used to display the data. Or, the CAA can “stage” the data request and module will automatically download the data later. The analyst can proceed while the request is staged and a empty panel indicates status. This process can be repeated to make a stack of plot panels (D) to compose a plot product (Fig. 17.2, right). Axis controls allow the analyst to scan along in time and zoom in on a feature of interest. The plot product can then be saved in a .papco file, printed, run as a batch process, etc, as with any PaPCo product. A complete description of PaPCo’s use and features is available on the PaPCo website, http://www.papco.org.

17.2.1 Slicing Data PaPCo modules can define “slicers” that allow the data to be inspected more deeply at a given time. For example, the analyst can click on a spectrogram to

17

Cluster CAA Module for PaPCo

253

see an individual spectrum line plot for the time selected. Slicing can be thought of as throwing out the contextual dimension to inspect the details of the remaining dimensions. Slices can be quite complex. For example, slicing on a highly-dimensional PEACE spectrogram will show a pitch angle distribution of the data. PaPCo is able to slice multiple panels at once and several slices can be displayed side-by-side or plotted on the same axes (Fig. 17.2, left). Any dataset can be sliced, in general a reasonable view of the slice of data is displayed, using the axis ranges from the sliced plot. When this default view is too limited, the custom slicers can be written to create improved displays. Please refer to the PaPCo website for updates to this documentation, and demonstration videos of the software.

17.2.2 Panel Editor Clicking on the Cluster CAA button causes brings up its panel editor (Fig. 17.3). With the panel editor, the analyst adds new panels and edits existing panels. Each

Fig. 17.3 CAA Module Panel Editor is used to add and edit panels

254

J. Faden et al.

CAA panel has a dataset identifier and parameter identifier specifying what should be plotted, plus axis settings and plot style settings. The panel editor consists of four areas. At the top of the panel editor is a filter panel (F) that reduces the number of datasets listed in the dataset selector (D) below. Once a dataset is selected in the dataset selector, the plottable quantities within the dataset are displayed in the parameters selector (P) below. Remaining panels (S) control axis ranges and plot style. Note also, the terse descriptions of datasets and parameters are displayed below each selector (I). The “info” button displays a more complete dataset description. Once selections are made, the “Add and Close” button of the actions panel (Q) adds the panel to the stack of panels PaPCo plots. The filter panel allows the list of dataset IDs to be constrained by S/C, dataset type, and instrument. These are the three droplist widgets in the upper left of the filter panel (F). For example, setting these to C1, , PEA would constrain the list to PEACE datasets from the Cluster 1 spacecraft. The text widget labeled “search” performs a keyword search on dataset IDs and their terse descriptions. Only datasets containing the string are displayed. For example, the analyst might search for moments datasets by setting this to “moment” or use part of a dataset ID like “DEFlux.” This is a regular expression, and case is ignored. The checkbox labeled “limit to listed datasets” will limit the list of datasets to those that have already been downloaded. This allows the analyst to see just the datasets that he uses, without having to dig through hundreds of datasets that are not of interest. The filter panel text widget labeled “parameters search” performs an exhaustive search for parameters within the list of datasets. This further constrains the datasets list to those containing parameter descriptions that match the search. This can be a little slow, since a search of all datasets requires about 80,000 lines of text to be processed to complete the search. Matching parameters are marked with an asterisk ( ) in the parameters list. Last, (X) is an empty panel in the figure that is filled with additional widget controls for more complex datasets. These are called plug-in extensions and are described more below. For example, when vector quantities such a BGSE are selected, controls for selecting component or magnitude are inserted here.

17.2.3 Transferring Data from the CAA The Cluster CAA module automatically downloads data from the CAA for plotting. This is one of the more complex and useful aspects of the module. The CAA has a “Command Line Interface” that is a simple web interface, so we can then use PaPCo’s “wget” functionality to interact with the CAA. Note PaPCo can be put in off-line mode to prevent interaction with the server. PaPCo has a “File Cache” idiom, which is a set of remote data files that are parameterized in time, cached locally. Typically the remote source is an FTP or web site, and the file cache downloads files as requested into a local mirror of the

17

Cluster CAA Module for PaPCo

255

Fig. 17.4 File Cache Tool showing locally and remotely available data. The “Get Selected” button is used for batch downloads

remote source. The File Cache Tool (Fig. 17.4) is a GUI that manages the file cache, allowing the analyst to browse the local and remote stores of data and download a number of files at once. We adapt the CAA Command Line Interface to make it appear to PaPCo as a remote store of data so that the file cache codes can be used. To enter the File Cache Tool, click on the “File Cache” button in the panel editor. Initially, the tool shows just the local files for the dataset, and the “list remote” button will perform a listing of the data in the CAA. The software may prompt for credentials at this point, and enter a username and password for the CAA. Note these can be stored permanently by using the “Config” button on the panel editor. After a short while, the CAA returns an availability listing for the dataset, and the File Cache Tool displays the listing. This listing is cached locally along with the data, and the “reset” button of the tool will retrieve a new listing. Note the CAA Command Line Interface allows clients to request arbitrary spans of data, and the module currently works with the CAA using day-long granules, without regard for how data was ingested into the CAA. The analyst can download data by selecting several remote files, then click on the “Get Selected” button. This posts a request for each day to the CAA. Its response

256

J. Faden et al.

will be either the files, or a message saying where the data will be when ready and a job ID. In this case, the staged request is marked with a “stage” file, and later attempts to get the data will check the stage file for the job ID and query CAA to see if the data is ready. When the data is ready, the data is then downloaded. If not, a message will be displayed. Queries to the CAA are limited to once per 120 s, because we don’t want to issue a query each time the page is redrawn. Data is downloaded as a .zip file, unpacked and then renamed to follow the conventions of the file cache. The tool can be used to delete data files once they are no longer useful.

17.3 Internal Mechanics PaPCo is open-source software and has a long history of being customized for particular applications. This section describes briefly technical aspects of the module for those who might consider using code or modifying the software for their own purposes. PAPCO BASE and PAPCO USER are configuration parameters that are set during the PaPCo installation process.

17.3.1 Reading and Plotting the Data Chris Perry’s IDL CEF reader is used to parse the CEF file. Data is read in and converted to PaPCo’s internal data model. Typically there is no information loss in the conversion, and using PaPCo’s model makes writing code for handling the data much simpler. Some datasets will necessarily have their dimensionality reduced to fit into more canonical data forms. For example, high-rank PEACE data is reduced to Flux (Time, Energy, Pitch Angle) so that distribution plots can be made using existing PaPCo software.

17.3.2 Datasets Database The dataset list comes from a local database of plottable parameters. Presently there is no mechanism to update this database, other than downloading a new version of the software. CAA is expected to add dataset discovery to its machine interface, and a future version will reflect changes in the available datasets. The database is distributed along with the module in a .zip file, and this file is unpacked into the user’s PAPCO USER area when the panel editor is first invoked. The database is implemented as a flat text file of line-separated records containing the dataset Ids and information about the dataset. This is PAPCO USER/caa catalog/caa datasets.tab. Each dataset has an XML description of its plottable parameters, and these are in PAPCO USER/caa catalog/db/.

17

Cluster CAA Module for PaPCo

257

17.3.3 Plug-in Extensions Some datasets are not immediately useful for plotting. For some, calibrations need to be applied, a component must be selected, or the dimensionality is too high for plotting. This may be the most challenging element of the module, and this is where “power users” may want to write special plug-in code that extends the module to handle special cases. The plug-in is like a mini module within the CAA module that defines a widget controller and is called by the CAA module as data is loaded and plotted. Plug-ins are IDL objects that allow more complex datasets to work with the module. Complete documentation would be beyond the limits of this article, but example codes can be found within the module at PAPCO BASE/papco modules/cluster caa/ plugins.

17.3.4 Using PaPCo and the CAA Module to Access Data Within the module is code that can be used for IDL analysis outside of PaPCo. PaPCo must be initialized for the downloading to work. On the PaPCo website there are example codes showing how the module can be used read data from the CAA and perform queries.

17.4 Installing PaPCo PaPCo’s home on the web is http://www.papco.org. Instructions for downloading and installing PaPCo can be found there, as well as demonstration videos and more complete documentation. PaPCo is intended for use on Windows, Linux, Mac, and Solaris platforms. The full version runs on IDL 5.6 and newer. It will run under Runtime (unlicensed) IDL with minor limitations, mainly that only the core modules can be used since new code cannot be compiled at runtime. PaPCo is open-source software that is freely available for use and modification. The source and distributions can be found on SourceForge.net.

17.5 Conclusion The CAA Module for PaPCo provides a convenient and powerful tool that handles the details of finding data in the CAA, downloading and storing files, reading a new data format, and other minutia so the analyst can focus on science. Further,

258

J. Faden et al.

power users are enabled by providing access to the internals of the module. Data is provided in a uniform and flexible data model to support further analysis and future modules based on data from the CAA. Acknowledgments The author wishes to thank NASA/Goddard for continued support for PaPCo and the CAA Module.

Part III

Measurement Techniques and Calibration Routines

Chapter 18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach A. Masson, O. Santol´ık, M.G.G.T. Taylor, C.P. Escoubet, A.N. Fazakerley, ˚ J. Pickett, A. Asnes, X. Valli`eres, H. Laakso, and J.-G. Trotignon

Abstract Electron density is a key physical quantity to characterize any plasma medium. Its measurement is thus essential to understand the physical processes occurring in the environment of a magnetized planet, both macroscopic and microscopic. Since 2000, the four satellites of the European Space Agency (ESA) Cluster mission have been orbiting the Earth from 4 RE to 20 RE and probing the density with several types of instruments. In the magnetotail, this rare combination of experiments is particularly useful since the electron density and the temperature fluctuate over several decades. Two of these experiments, a relaxation sounder and a high-time resolution wide-band receiver, have rarely been flown together in the far tail. Such wave data can be used as a means to estimate the electron density via the identification of triggered resonances or the cutoffs of natural wave emissions, typically with an accuracy of a few percent. For the first time in the magnetotail .20 RE /, the Z-mode is proposed as the theoretical interpretation of the cutoff observed on spectrograms of wave measurements when the plasma frequency is greater than the electron gyrofrequency. We present examples found in the main regions of the magnetotail, comparing simultaneous density estimation from active and passive wave measurements with a particle instrument and calibrated spacecraft-to-probe potential difference data. With these examples, we illustrate the benefit of a multi-instrument approach for the estimation of the electron density in the magnetotail and the care that should be taken when determining the electron density from wave data. ˚ A. Masson (), M.G.G.T. Taylor, C.P. Escoubet, A. Asnes, and H. Laakso ESA/ESTEC, D-SRE, Noordwijk, The Netherlands e-mail: [email protected] O. Santol´ık Institute of Atmospheric Physics and Charles University, Faculty of Mathematics and Physics, Prague, Czech Republic A.N. Fazakerley UCL, Mullard Space Science Laboratory, Surrey, UK J. Pickett University of Iowa, Iowa, USA X. Valli`eres and J.-G. Trotignon LPCE/CNRS and Universit´e d’Orl´eans, Orl´eans, France H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 18, c Springer Science+Business Media B.V. 2010

261

262

A. Masson et al.

18.1 Introduction A precise estimation of the electron density in the Earth’s magnetic tail is needed to understand key physical phenomena at all spatial scales: from the microphysics of magnetic reconnection to the mechanisms of plasma entry into and within the tail. However, this estimation is not straightforward since the magnetotail is far from being a uniform medium, with electron density values ranging from 0.001 to 10 cm3 and electron temperature from a few 10’s of eV to several keV. To cover such a large range of plasma conditions, several scientific instruments are usually carried onboard a magnetospheric satellite to probe this key physical parameter. In the case of the ESA Cluster mission [8], four different types of instruments can be used to estimate it on each spacecraft: an electron experiment, an electric field instrument, a wide-band radio receiver and a relaxation sounder. Depending on the plasma conditions, each has their pros and cons (Section 18.2) but used in concert, they offer a rare opportunity to derive one of the most accurate measures of the electron density in the magnetotail. Section 18.3 focuses on the theoretical interpretation of the cutoffs observed on the spectrograms of active and passive wave measurements performed in the distant tail in order to derive the electron density. A rare event in the southern lobe region is then presented where density estimates from natural waves and calibrated spacecraft-to-probe potential difference measurements are compared. Typical examples found in the main regions of the magnetotail are finally presented in Section 18.4, comparing simultaneous density estimates from the electron instrument and active wave data.

18.2 Cluster Instruments for Electron Density Estimation On each Cluster satellite, the electron density can be derived from the following four instruments: the Plasma Electron And Current Experiment (PEACE) [18], the double-probe Electric Field and Waves (EFW) instrument [15] and two plasma wave instruments: the Wide-Band Data (WBD) experiment [14] and the Waves of HIgh frequency and Sounder for Probing of Electron density by Relaxation (WHISPER) instrument [5]. As detailed below, each of them has its advantages and disadvantages depending on the plasma conditions encountered.

18.2.1 PEACE The electron density is the zero-th order moment of the electron velocity distribution function measured by an electron experiment. A tutorial on the measurement of such distribution functions may be found in Fazakerley et al. [11] for particles of a few eV to a few tens of keV energy. A companion paper on plasma moments derived

18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

263

from a measured electron velocity distribution function [28] details their definition, computation, accuracy and use. An overview of the PEACE data contained in the Cluster Active Archive (CAA) with many aspects of their calibration can be found in [Fazakerley et al. Chapter 8, 19]. For completeness, we include some specifics here of the PEACE instrument with respect to the measure of the electron density. Accurate estimates of the electron density can be obtained in most regions of the magnetotail from the ground analysis of three dimensional (3D) data collected by PEACE on each Cluster satellite. These data have a good data coverage along the Cluster orbit compared to WBD (see Section 18.2.3) but don’t always have spin by spin cadence, except on Cluster 2 (C2). There are a couple of drawbacks inherent to this type of experiment including: the aging of the micro-channel plates (MCP) and the non-linear response of the MCP depending on the gain level used and the energy of the electrons detected. To counter balance this latter issue and fine-tune the calibration of the PEACE experiment, the absolute sensitivity is determined using comparisons of density measurements between PEACE and the plasma wave instruments WHISPER and WBD. Moreover, due to the inherent measurement technique used by the instrument (only a finite energy range is sampled), PEACE only provides a partial moment of the full electron distribution. As a result, in certain plasma regimes, where a proportion of the distribution is outside the range of the instrument, the density will be under estimated. These measurement challenges reinforce the importance of inter-calibrating all Cluster instruments able to estimate the electron density.

18.2.2 EFW The EFW instrument onboard Cluster [15], Chapter 6 estimates the potential difference between the main body of the satellite and the electric field probes every 0.2 s in the normal mode. In order to sample the pristine unperturbed plasma, the probes are attached to long booms to escape the sheath of plasma perturbations induced by the spacecraft. This goal is achieved on Cluster by having four spherical probes at the end of four 42 meter long antennas attached to each conductive spacecraft platform. Moreover, the probes are kept close to the local plasma potential by balancing their photoelectron current with a bias current from the spacecraft. In the magnetosphere, the spacecraft potential is determined by the balance between the current related to the escaping photoelectrons, the current related to the collected ambient electrons and the bias current. Thus, the knowledge of the photoelectrons escaping the spacecraft as a function of its potential, also called spacecraft photoelectron characteristic, enables the derivation of the ambient plasma density from the spacecraft potential measurements. Several papers have detailed this procedure for data collected by different magnetospheric satellites (e.g., [7, 19, 20, 24, 30–32, 36, 40]). Cully et al. [4] have carried out numerical calculations of the potential around Cluster, including the booms and the probe systems. They found that the potential

264

A. Masson et al.

in the vicinity of the probes (or probe potential) is driven by the positive boom tips and is approximately 19% of the spacecraft potential when the Debye length is larger than the probe-boom tip separation of 3 m. Taking into account this effect, yearly relationships (2001–2007) between the EFW spacecraft-to-probe potential difference measurements and the ambient electron density have been recently reported for plasma conditions found in the tail lobes and the polar caps [33, 38]. These recent developments are discussed further in Section 18.4.1.

18.2.3 WBD The NASA Wideband (WBD) plasma wave instrument consists of a digital wideband receiver onboard each Cluster spacecraft. Each of them provides very high-rate electric field waveforms either between 100 Hz and 9.5 kHz, 100 Hz to 19 kHz or 700 Hz to 77 kHz. These waveforms can also be sampled in three frequency conversion modes that provide frequency offsets of approximately 125, 250, and 500 kHz. For a complete description of the WBD instrument, see Gurnett et al. [14] and for the Wideband data products delivered to the CAA, see [Pickett et al. Chapter 11]. In the magnetosphere, WBD data can be used as a means to obtain accurate estimates (to a few percent at most) of the electron density via the cutoffs of plasma waves (see Section 18.3). However, the WBD primary mode of operation is the transmission of the measured waveform data in real-time to the NASA Deep Space Network or to the ground station in Panska Ves, Czech Republic. The consequence of this mode of operation is a non-continuous coverage of data along the Cluster orbit, unlike most of the other instruments like PEACE, EFW or WHISPER (see Section 18.2.4). After 8 years of operations, the average amount of data recorded by WBD is around 9 h per 57-h orbit for each spacecraft. In order to fulfill its many scientific objectives, the Wideband data recorded in the distant magnetotail are scarce. A recent survey of the WBD database found less than a dozen cases where the electron density could be accurately deduced in the distant tail. For all these events, WBD was operating in the mode up to 9.5 kHz which provides continuous waveforms with a 27.4 kHz sampling rate. This frequency range enables estimates of the electron density to be made between 0.001 and 1 cm3 . Despite their paucity, these case studies have played a key role in deducing the mode of propagation of the waves whose frequency cutoff can be used to derive the electron density. In combination with WHISPER active soundings, these cutoffs can be related to the local plasma density signatures.

18.2.4 WHISPER In most of the regions crossed by the Cluster mission, the electron density can be derived accurately from the active soundings and passive measurements of the

18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

265

WHISPER experiment. The hardware of WHISPER mainly consists of a pulse transmitter in the 4–80 kHz range, a wave receiver (2–80 kHz) and a wave spectrum analyzer. The principle of a relaxation sounder may be described as follows. A radio transmitter sends a wave train of frequency f during a short time period (1 or 0.5 ms for WHISPER). Soon after, a radio receiver connected to an electric sensor records the electric field signal around f for a short while. In the case of WHISPER, the electric sensor is one double-sphere dipole probes of the EFW experiment. The frequency is then shifted for a new sounding. The process is repeated until the full frequency bandwidth is covered. When a transmitted wave train is close to a characteristic frequency of the local plasma, very intense echoes can be received. Some of these stimulated signals are called resonances. They are almost monochromatic and mainly electrostatic. These plasma resonance signals enable the accurate derivation of the electron density via the identification or the deduction of the electron plasma frequency. A complete description of the WHISPER hardware can be found in D´ecr´eau et al. [5] and WHISPER data products delivered to the CAA are described in [Trotignon et al. Chapter 12]. The sounding approach has been considered the standard in the ionosphere (e.g. [6]) to which other techniques are compared and is also capable of making reliable magnetospheric electron density measurements using either relaxation sounding [9] or high-power sounding designed to produce long-range echoes [34]. Three main advantages favour the use of a relaxation sounder to derive the plasma density. First, it can trigger plasma resonances locally when the medium does not always show them naturally, as in the distant magnetotail (see Section 18.4.2). Second, it guarantees in situ measurements, while cutoffs of the plasma waves captured by a radio receiver might reflect the density at some unknown distance from the spacecraft. Last but not least, the instrument is not subject to ageing phenomena and needs no real calibration, only integrity checks. The main limitations are inherent to the technique and the hardware characteristics. The density is derived indirectly and the resonance recognition process is not 100% efficient especially in complex plasmas [5]. On the hardware side, the frequency range chosen for WHISPER limits the analysis to plasma regimes with densities between 0.2 and 80 e-/cc in the sounding mode and 0.05 to 80 e-/cc in the passive mode. The electron temperature range where plasma densities can be derived from WHISPER data is related to the length of the EFW booms and will be discussed in Section 18.4.3. Finally, recall that each Cluster satellite only transmits to the ground the on-board calculated Fast Fourier Transform (FFT) of the digital electric waveforms. This prevents some data analysis techniques from being used, like those performed on the data collected by the ISEE-1 relaxation sounder [42]. Unlike WBD, WHISPER benefits from full data coverage along the Cluster orbit but the time and frequency resolutions of WHISPER are much smaller than WBD. In the normal mode, a passive spectrum is recorded every 2.2 s and an active spectrum every 52 s or 104 s, with a minimum frequency resolution of 162 Hz.

266

A. Masson et al.

18.3 Z-Mode in the Distant Tail Relaxation sounders have been rarely flown in the distant tail. The first spacecraft carrying a sounder above 10RE was ISEE-1 in the late 1970s [16, 17]. Using the measurements of this sounder, Etcheto and Saint-Marc [10] reported the first observation of a high-density cold plasma region (5 cm3 , 30 eV) just outside the plasma sheet boundary layer (PSBL). Parks et al. [27] later named this new region the LowEnergy Layer (LEL). More recently, a number of papers have presented examples of spectrograms recorded by WHISPER in the distant tail (e.g. [5, 43]). Each time, the low-frequency cutoff of the triggered emissions and sometimes the natural waves observed were considered to be close to the electron plasma frequency (fpe ), thus enabling the derivation of the density, Ne from p fpe ŒkHz D 8:98 Ne cm3

(18.1)

But overall, few studies made use of the electron density measurements derived from active soundings performed in the deep tail. Moreover, whereas the derivation of the electron density from sounder data collected in the solar wind, magnetosheath and plasmaspheric regions is common knowledge, the interpretation of active soundings and natural wave cutoff in the far tail is much less clearly linked to the cold plasma theory. For instance, Etcheto and Saint-Marc [10] consider that the strongest signal to be “in the vicinity of the plasma frequency”. The plasma frequency is the characteristic frequency at which plasma oscillates when its constituents are displaced relative to one another. In addition to this mode, a number of characteristic frequencies and speeds associated with the medium exist. These modes are derived from the dispersion relation of that medium. According to the cold plasma theory (e.g. [37]), where one ignores the thermal speeds of the ions and the temperature of the electrons, the so-called Z mode with L-X polarization exists below the electron plasma frequency (fpe ). Above fpe , three modes exist, the X polarized Z-mode and the two free space L-O and R-X modes when fpe is greater than fce with fce the electron cyclotron frequency (see left panel of Figure 1.2. of Benson et al. [2]). Local Z-mode signatures have been recorded by many ionospheric spacecraft and more recently in the magnetosphere by the NASA IMAGE satellite (polar orbit, apogee 7RE ). See Benson et al. [2] for a recent review. These data confirmed the idea that a more complicated relationship exists between the characteristic modes of the plasma and the observed cutoffs. In particular, the cutoff corresponding to fpe is accompanied by a lower cutoff (fZ ) corresponding to the cutoff at L D 0 of the Z mode with L-X polarization and given by fZ D

. 0:5 fce 2 C 4 fpe 2 fce 2

(18.2)

The importance of a proper wave-mode identification was well illustrated by the classic work on the auroral plasma cavity by Calvert [3] where a whistler-mode interpretation by Gurnett and Green [12] was reinterpreted as a Z-mode emission,

18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

267

and by Gurnett et al. [13], who used DE-1 satellite data to reinterpret the Hawkeye satellite continuum data of Gurnett and Green [12] as Z-mode emission. An extensive review on auroral Z-mode emissions was done by LaBelle and Treumann [21]. The work of Lefeuvre et al. [22] stressed the benefit of multi-component wave measurements to distinguish between wave modes. Our paper stresses the benefit of multi-instruments in resolving difficult interpretation situations. In this section, we present two rare case studies, where a multi-instrument approach enables us to propose a new theoretical interpretation of these low frequency cutoffs observed on active and natural wave spectra far in the tail; data were collected by WBD, WHISPER, PEACE, EFW and the FluxGate Magnetometer (FGM) [1].

18.3.1 Event 1: 28 July 2006 Figure 18.1a displays a spectrogram of WBD electric field data recorded onboard Cluster 1 (or C1) in the outer plasma sheet (Te 660 eV) on 28 July 2006 between 09:25:30 and 09:26:00 UT. A clear steady cutoff of wave emissions is observed at 8:7 kHz called fZ (see white arrow). This cutoff can be extracted automatically

a dB

−70 −80 −90

No data

Frequency (kHz)

10

5

−100

fz

−110 −120 RE MLAT

30

40

50

00

2.9 21.0

Time (s)

−5

10

Vrms.Hz

−1/2

b

19.6 -14.6

MLT L

0

09:25:30 (28 July 2006) Cluster – C1

WBD

fp

WHISPER fz~ fp > fce~ 0.5 kHz

−7

10

5

11

17

23

30

36 kHz

42

49

55

61

Fig. 18.1 (a) WBD electric field measurements performed on 28 July 2006 onboard Cluster 1 from 09:25:30 to 09:26:00 where a clear low frequency cutoff is observed at fZ 8:7 kHz; (b) WHISPER active spectra recorded at 09:25:31 showing the exact same cutoff

268

A. Masson et al.

from the WBD data by adapting the automatic cutoff recognition procedure developed for auroral hiss by Santol´ık et al. [35]. An active spectrum recorded by WHISPER on C1 at 09:25:31 is displayed in Fig. 18.1b where the same frequency value of the cutoff is observed (thick white arrow pointing up). Note that the chosen frequency cutoff (white arrow) was picked up in the middle of the decreasing slope and not at the foot of it. The active/passive comparison (not shown), however, does not indicate the presence of significant emissions stimulated by the sounder. Thus, we assume that both signatures are relevant to a local signature of the medium. Due to the high-time resolution of WBD, a dimmer but detectable secondary cutoff is present around 0.25 kHz above fZ . We suggest that this secondary cutoff corresponds to the electron plasma frequency (fpe ) while the sharp lower cutoff (fZ ) corresponds to the Z-mode frequency defined in equation (18.2). If true, the 0.25 kHz frequency difference between fpe and the lowest cutoff (fZ ) should correspond to half the value of the electron gyrofrequency as follows from Eq. (18.2) when fpe >> fce fz fpe fce =2 . Since fce ŒHz D 28 B ŒnT and the magnetic field strength estimated by FGM at this time is B 18 nT, hence fce =2 0:25 kHz. Thus, this interpretation leads to an estimate of the plasma frequency of fpe 8:95 kHz hence Ne 0:99 cm3 . If the lower cutoff had been interpreted as the electron plasma frequency (fpe 8:7 kHz), this would have led to an estimate of 0:94 cm3 . During the same time interval, PEACE measures a density varying between 0.92 and 0:84 cm3 using current calibrations. Knowing that the spacecraft potential is around 10 V, we suggest that the tiny difference between density estimates from the wave cutoff and the electron detectors is due to the electrons with energy below 35 eV not measured by PEACE during this time interval. In other words, PEACE data cannot be used to discriminate between the two cutoffs observed on WBD for this specific case study. It is fair to say that for such plasma conditions, fpe >> fce , the density difference between the new interpretation (fZ ) and previous interpretation (fpe ) is almost negligible (5%). In other words, previous published density estimates from active sounder data in the outer plasmasheet were good approximations. Moreover, since the nominal frequency resolution of WHISPER is 162 Hz but often 324 Hz is used, we cannot fully reject the possibility that the lower cutoff observed on active spectra in the far tail might be the plasma frequency instead. However, in other plasma conditions (e.g. the lobes), the interpretation of the lower cutoff of WBD passive data as either Z-mode or fpe can lead to a significant density difference, as we will now demonstrate.

18.3.2 Event 2: 07 September 2006 On 07 September 2006, the Cluster spacecraft were crossing the southern lobe region. Figure 18.2 displays the spectrogram of the electric field recorded by WBD on C1 between 07:38:30 and 07:39:00 UT where a clear cutoff is visible

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

WBD

fpe>fce~0.9 kHz

dB 10 Frequency (kHz)

269

−70 −80 −90 −100 −110

5

−120 RE MLAT

fz

MLT L

0 30

40

50

07:38:30 (07 Sept 2006) Cluster – C1

18

18.2 –36.5 0.6 28.1

00 Time (s)

Fig. 18.2 WBD electric field measurements performed on 07 September 2006 onboard Cluster 1 from 07:38:30 to 07:39:00 where a clear low frequency cutoff is observed at fZ D 1:5 kHz

at fZ 1:5 kHz. Since such a frequency is below the lower limit of the WHISPER soundings (4 kHz), we need to assume that this natural wave cutoff corresponds to a local signature. At this time, the magnetic field strength measured by FGM was 32nT. Using the newly proposed interpretation of the frequency cutoff, fpe 2:01 kHz and Ne 0:05 cm3 . The spacecraft-to-probe potential difference recorded by EFW on C1 is perturbed by the Electron Drift Instrument [29] operating on both C1 and C3 during this time period. In such a tenuous and cold plasma environment (Te < 100 eV), EDI has sometimes the opposite effect of the Active Spacecraft POtential Control [41]: pushing the spacecraft potential up and perturbing PEACE electron measurements. However, EDI was not operating on C2 and C4 at the time of the WBD measurements on C1. The spacecraft-to-probe potential difference measured by C2 and C4 match very well for hours before and after the WBD time period. Since C2 and C4 were the more spread apart satellites within the Cluster space fleet while C1 was flying in between, we assume that the plasma was homogeneous within the Cluster constellation in the lobe region crossed that day. In other words, we make the assumption that the spacecraft-to-probe potential difference on C1 fluctuated between 35.55 V and 36.16 V from 07:38:30 to 07:39:00 UT, according to C2 and C4 measurements. Thanks to a recent calibration study of the spacecraft-to-probe potential difference measurements for such plasma conditions (see Sections 18.2.2 and 18.4.1), we are able to derive a density estimate of Ne 0:053 ˙ 0:01 cm3 from the average EFW potential value (35:7 V). This estimate is in good agreement with the WBD estimation. However, it would have been in disagreement if the WBD lower cutoff frequency would have been interpreted as the plasma frequency since fpe 1:5 kHz gives Ne 0:028 cm3 .

270

A. Masson et al.

PEACE estimates of the density on C2 and C4 exceed or are close to 0.028. These estimates take into account electrons with energy between 50 and 200 eV but discard lobe electrons in the same energy bin as the photoelectrons (below 50 eV) resulting in an underestimation of the electron density. Thus, PEACE data on C2 and C4 also support the Z-mode interpretation during this time interval.

18.4 A Multi-instrument Approach The Earth’s magnetotail is composed of three main regions, characterized by distinct electron density and temperature ranges: the lobe regions (Ne < 0:1 cm3 , Te 100 eV), the central plasma sheet (0:1 < Ne < 10 cm3 , Te 1 keV) and sandwiched between these two regions the Plasma Sheet Boundary Layer (PSBL). The PSBL is characterized by high-speed streaming and counter-streaming beams of particles of ionospheric and magnetic reconnection origin. Deriving the best density estimate in these regions requires a careful selected combination of Cluster instrument measurements. We illustrate this below with recent results and open questions showing how an accurate estimation of this parameter is relevant to assess critical geophysical issues.

18.4.1 Lobe Regions The magnetospheric lobes are the regions of space located tailward of the polar cap and delimited by the mantle and the plasmasheet regions of the magnetotail. They are usually characterized by tenuous cold plasma with density below 0:1 cm3 [38]. We note that the lowest density value that can be derived from WHISPER soundings is 0:2 cm3 . In such plasma conditions, the potential attained by the spacecraft is so high (20–70 V) that most of the ambient ions are not able to reach the ion sensors. Thus, ion density measurements in this region are of little relevance. Electron measurements can be affected by photoelectrons (see [? ? ]) and in certain low temperature plasma environments are limited by the finite energy bandwidth of the instrument. However, since the measure of the spacecraft-to-probe potential difference does not have these intrinsic limitations, a proper calibration of these measurements can provide accurate estimates of the ambient plasma density with a high time resolution (up to 0.2 s on Cluster) for all lobe crossings as long as the ASPOC instrument is turned off. The calibration procedure of the measured EFW spacecraft-to-probe potential difference (Vscp ) into density is detailed in Pedersen et al. [33] for plasma conditions found in the lobes and over the polar cap regions. Pedersen et al. [33] provides, for the first time, yearly relationships between Vscp and electron density from 2001 to 2004 assuming Te < 100 eV. An update to 2007 has been recently published by Svenes et al. [38]. Using these relationships, Ne can be derived from Vscp with an

18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

271

error bar of ˙ .10–20/ %. For spacecraft potential above 40 V .Ne < 0:05/, only an order of magnitude accuracy of Ne is claimed. Thanks to these relationships, Svenes et al. [38] was able to study the nearEarth magnetospheric lobes throughout 7 years of solar cycle 23 covered so far by Cluster. All derived densities are distributed between 0:007 cm3 and 0:5 cm3 but two-thirds of them are below 0:1 cm3 with a median value of 0:047 cm3 . The density in this region was found in general to be independent of the solar wind conditions or the geomagnetic activity. However, the high-density tail of the population (Ne > 0:2 cm3 ) seemed to decrease with the waning of the solar cycle, pointing to a source region influenced by the diminishing solar UV/EUV intensity. As the quiet time polar wind has such a characteristic, they conclude that polar wind is most likely the dominant source of plasma in the lobes.

18.4.2 PSBL/Outer Plasmasheet The plasma sheet boundary layer and the outer plasmasheet are characterized by electron densities of the order of a few electrons per cc and temperatures of a few hundreds eV. In these regions, the electron density is accurately sampled on each Cluster satellite by the PEACE experiment. However, as discussed in Section 18.2.1, this instrument requires in-flight calibration due to the aging of the micro-channel plates (MCP) and the non-linear response of the MCP to the gain level and the energy of the electrons detected. As detailed in Fazakerley et al. [? ], the post-launch calibration has been enhanced by the comparison with other experiments estimating the electron density in an independent way. Contrary to the lobe regions, no yearly relationships between EFW spacecraft potential and density have been derived yet while WBD datasets are too scarce to provide reference densities through the PEACE energy range. Very often triggered emissions are observed on WHISPER active spectra in these regions, which enable the provision of accurate reference densities over a wide range of electron temperatures. Figure 18.3 presents data collected by C1 during a typical outer plasmasheet crossing (17 August 2003, 15:00–16:20). The top panel displays a passive WHISPER spectrogram of the electric field between 2 and 40 kHz, where almost no wave activity is detected above 10 kHz. The second panel from top presents an active WHISPER spectrogram between 5 and 40 kHz during the same time period with a 52 s time resolution. In this panel, the most intense emissions are observed between 15 and 30 kHz. These emissions are sounder stimulated since almost no wave activity is observed on the passive receiver data above 10 kHz during the same time period (see top panel). A clear resonance, nearly continuous, is observed around 15 kHz in the second panel. This resonance is pointed out with three white arrows.The third panel also displays the active WHISPER spectrogram but this time, the PEACE density dataset, re-sampled at WHISPER sounding times is over-plotted in pink using Eqs. 18.1 and 18.2. Meanwhile, the magnetic field modulus estimated by FGM fluctuates between

A. Masson et al. 70 55 50 45 40 35 30 25 20 15 10 5 0

Frequency (kHz)

40 30 20 10 2

WHISPER resonance 30

17 August 2003 Cluster – C1

Frequency (kHz)

40

Signal Level (dB) Reference 0dB: 1.00E-07 Vrms.Hz−1

272

20

10

Frequency (kHz)

40 PEACE At 16:20 RE 17.8

30

MLAT MLT

20

L

14.6 1.2 18.6

10 15:00

15:20

15:40

16:00

16:20

1000

1 100

10 15:00

15:20

15:40

16:00

Electron density (cm−3)

Parallel electron Temperature (eV)

5

0.1 16:20

Fig. 18.3 Top: WHISPER passive spectrogram of the electric field between 2 and 40 kHz recorded on 17 August 2003 from 15:00 to 16:20 UT, 2.2 s time resolution; second panel from top: WHISPER active soundings performed between 5 and 50 kHz every 52 s during the same time period where clear continuous triggered resonance is pointed out with three white arrows; third panel from top: same as the second panel with PEACE densities converted to Z-mode frequencies overplotted. An almost perfect match is found with the continuous resonance pointed out in the second panel; bottom panel, PEACE estimation of the parallel temperature in electron-volts (eV)

18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

273

10 and 35 nT, corresponding to an electron gyrofrequency, fce , between roughly 0.3 and 1 kHz. An almost perfect match is found between PEACE densities converted to Z-mode frequencies and WHISPER resonances. The bottom panel displays the parallel electron temperature (black curve) and the electron density (pink curve) determined by PEACE. The temperature slowly evolves between 100 and 200 eV while the density fluctuates between 1 and 5 cm3 . For each Cluster spacecraft, analysis of tens of cases such as this one have been performed in the outer plasmasheet and the PSBL by members of the intercalibration working group of the Cluster Active Archive. Each time, an agreement better than 90% was found between PEACE densities converted to Z-mode frequencies and WHISPER resonances.

18.4.3 Central Plasmasheet Most of the time, the plasma number density in the central plasmasheet is of the order of a few tenths of electrons per cubic centimetre (e =cc) with a temperature around a few keV [23]. However, during periods of extended northward Interplanetary Magnetic Field (IMF) with low geomagnetic activity, a Cold Dense Plasmasheet (CDPS) can be observed. The density of the CDPS can be as high as several e =cc with a plasma temperature dropping well below 1 keV. The CDPS is observed more often near the flanks, but at times extends throughout the entire plasmasheet. Several mechanisms have been proposed to explain the CDPS, in particular plasma diffusion across the magnetopause [39], plasma mixing due to KelvinHelmholtz vortices [25], and dual lobe magnetic reconnection [26]. Understanding the CDPS formation is important because the plasmasheet provides the seed populations for the ring current and the radiation belts, which can become substantially enhanced when the IMF turns southward after a CDPS episode, and when the ensuing magnetic storm pushes the plasma into the inner magnetosphere. The central plasmasheet is also the place of major macro-scale magnetospheric phenomena such as bursty bulk flows and magnetic reconnection related to substorms. For all these reasons, an accurate estimate of the electron density is crucial. The PEACE instrument on each Cluster satellite provides an accurate estimate of the electron density in most cases. However, regular comparisons with an independent estimate of the density help to fine-tune the calibration of PEACE. In the central plasmasheet, the analysis of WHISPER active soundings on tens of case studies has revealed that accurate densities can be derived from active spectra. One of these case studies is presented in Fig. 18.4. It corresponds to the crossing of the central plasmasheet by Cluster 1 on 30 July 2002 between 18:00 and 19:20 UT. In the top panel, the parallel electron temperature derived from PEACE measurements every 4 s is displayed as a black line. The blue asterisks indicate when an estimate of the electron density was derived from WHISPER active soundings. In the second panel, the black line corresponds to the estimation of the electron density by PEACE while the blue asterisks are the densities derived from WHISPER active

274

A. Masson et al.

Electron parallel Temp. [eV]

Cluster 1-20020730 5000 1000

100

10 18:00

18:10

18:20

18:30

18:40

18:50

19:00

19:10

19:20

18:10

18:20

18:30

18:40

18:50

19:00

19:10

19:20

70

80

80

Ne [cm–2]

10 1 0.2

18:00 28 Frequency [kHz]

25 20 15 10 4 Frequency cut-off picked up: 6.26 kHz

–5

Electric Field Power

10

10

–7

10

–8

10

–9

10

20

30

40

50

60

Fig. 18.4 Cluster 1, 30-July-2002, 18:00–19:20 UT, top panel: PEACE parallel electron temperature (black line), the blue asterisks indicate when an estimate of the electron density was also derived from WHISPER active soundings. Second panel from top: PEACE electron density (black line), the blue asterisks are the electron density values derived from WHISPER active soundings. The third panel from top displays WHISPER active spectrogram of the electric field between 4 and 28 kHz recorded every 52 s. A superimposed thick black line correspond to the PEACE densities converted to Z-mode frequencies using Eqs. 18.1 and 18.2, down sampled to WHISPER time resolution; the electron gyrofrequency is estimated using FGM magnetic field data. All sharp lower cutoff frequencies extracted from individual WHISPER active spectra are overplotted in this third panel as short horizontal white line segments. They correspond to the blue asterisks in the top two panels. The bottom panel displays a single active spectrum recorded at 19:05:31 UT from 4 to 80 kHz. A sharp cutoff is observed at 6.29 kHz and shown with a dotted black line. A blue dashed line is located at 28 kHz to stress the upper frequency limit of the spectrogram displayed in the third panel. The 4–28 kHz part of this spectrum corresponds to the narrow black rectangle pointed out with a black arrow in the third panel

18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

275

soundings. Not every WHISPER active sounding enables the successful derivation of a density value since a sharp lower cutoff, such as the one presented in the bottom panel, is not always observed (see discussion below). A shallow cutoff logically leads to imprecise or impossible density estimation. Active spectra recorded every 52 s by WHISPER between 4 and 28 kHz are presented in the third panel from top. Although the measurements have been performed up to 80 kHz, we focus on this lower range to better compare PEACE and WHISPER density estimates. In the same panel, a superimposed thick black line displays the PEACE densities converted to Z-mode frequencies using Eqs. 18.1 and 18.2, down-sampled to WHISPER time resolution. The electron gyrofrequency is estimated using FGM magnetic field data (not displayed, the modulus of the magnetic field fluctuates around 8 nT). This black line fits the lower cutoff of intense emissions triggered by WHISPER soundings especially after 19:00 UT (Te 500 eV). Lower cutoff frequencies have been extracted from individual WHISPER active spectra when a sharp cutoff was observed. These frequency values are overplotted in this third panel as short horizontal white line segments. They correspond to the blue asterisks in the top two panels. The bottom panel displays a single active spectrum recorded at 19:05:31 UT from 4 to 80 kHz. A sharp cutoff is observed at 6.29 kHz and shown with a dotted black line on the left of the bottom panel. A blue dashed line is located at 28 kHz to stress the upper frequency limit of the spectrogram displayed in the third panel. The 4–28 kHz part of this spectrum corresponds to the narrow black rectangle pointed out with a black arrow in the third panel. As one can notice, the WHISPER response to this hot plasma environment is different from the outer plasmasheet/PSBL. Whereas almost continuous density datasets can be extracted from this latter region (see Section 18.4.2), only a set of clear cutoffs, irregularly spread in time, can be extracted from WHISPER soundings in the central plasmasheet, especially before 18:40 UT when Te > 1 keV. This result is not surprising since, according to antenna theory, the physical length (L) of the antenna has to be higher than the Debye length, which is proportional to the square root of the Te =Ne ratio. Thus, a hot and relatively tenuous population has an associated Debye length much larger than L and is sometimes not seen by the sounder. Unfortunately, a quantitative estimation of the Debye length limit has not yet been estimated for the WHISPER instrument [? ]. In other words, a precise range of electron density and temperature where the WHISPER soundings can be used to estimate the electron density remains to be established. However, a number of numerical simulations, together with statistical analyses of WHISPER/PEACE cross-calibrations in this region, are in progress to estimate it. Nevertheless, the example presented in Fig. 18.4 represents one of the first determinations of densities derived from active soundings in the central plasmasheet. The interpretation of the active resonance proposed in Section 18.3 is used since close to the cutoff the phase speed is very large (>c, diverging to infinity at the cutoff frequency) so the wave-particle interactions should be independent of particle velocities, which are obviously below the speed of light. Of course the cutoff position would be shifted in a relativistic plasma since the electron gyrofrequency would

276

A. Masson et al.

Table 18.1 Instruments available on every Cluster satellite to provide an estimate of the electron density are listed in the left column. The next column details the data coverage in the magnetotail of these instruments. The last three columns indicate by a “X” sign which instruments should be preferably used in the magnetotail to derive the density according to this study PSBL/outer Cluster experiments Magnetotail data coverage Lobes plasmasheet Central plasmasheet EFW 100% X PEACE 100% X X WHISPER 100% X X WBD Few data sets X X

decrease with increasing relativistic mass of the particles. But the energies are still too low for that in the plasmasheet and the cold plasma approximation holds. Analyses of tens of events reveal that PEACE and WHISPER densities agree within a few percent. According to the different findings reported in Sections 18.3 and 18.4, Table 18.1 summarizes which Cluster instruments should be preferably used to derive the electron density in the three main regions of the magnetotail.

18.5 Conclusion Together, particle detectors, DC and AC electric fields instruments onboard magnetospheric satellites enable to accurately estimate the electron density in the Earth’s magnetotail. A multi-instrument approach enhances the calibration of both particle experiment and spacecraft-to-probe potential difference values, thus enabling the scientific community to access a more accurate estimate of the electron density. This paper details the outcome of this approach for such experiments onboard each satellite of the Cluster mission. A new theoretical interpretation in terms of the Z-mode is proposed for the lower cutoff of natural and active soundings observed on spectrograms collected by a wide-band receiver and a relaxation sounder in the far tail (20 RE ). However, since the nominal frequency resolution of the WHISPER relaxation sounder is 162 Hz (but often 324 Hz is used), we cannot fully reject the possibility that the lower cutoff observed on active spectra in the far tail might be the plasma frequency instead. An agreement between the density derivation based on this interpretation and calibrated spacecraft-to-probe potential difference measurements is found to be within a few percent in the lobes. The same level of agreement is found in the PSBL and the outer plasmasheet between densities derived from the electron detectors and from active soundings. One of the first examples of density derivation from active soundings is presented for data collected in the central plasmasheet around 20 RE on the nightside. This observational fact raises the issue of an accurate estimation of the Debye length limit for the WHISPER instrument. In other words, in which plasma conditions is WHISPER able to sense the plasma and accurately derive the density?

18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

277

Overall, a very good agreement was found between estimates of the electron density from these totally independent types of experiments. The multi-instrument approach has helped to fine-tune the calibration of the PEACE and EFW instruments, hence providing a higher accuracy of density estimates in the Earth’s magnetosphere. Acknowledgements This study has been triggered and discussed within the cross-calibration group of the Cluster Active Archive (CAA) sponsored by the European Space Agency. Ondrej Santol´ık acknowledges the grant support from GAAV IAA301120601 and NSF 0307319/ME 842.

References 1. Balogh, A., et al. (2001), The Cluster Magnetic Field Investigation: overview of in-flight performance and initial results, Ann. Geophys., 19, 1207–1217. 2. Benson, R.F., P.A. Webb, J.L. Green, D.L. Carpenter, V.S. Sonwalkar, H.G. James, and B.W. Reinish (2006), Active wave experiments in space plasmas: the Z mode, Lect. Notes Phys., 687, 3–35. 3. Calvert, W. (1981), The auroral plasma cavity, Geophys. Res. Lett., 8, 919–921. 4. Cully, C. M., R. E. Ergun, and A. I. Eriksson (2007), Electrostatic structure around spacecraft in tenuous plasmas, J. Geophys. Res., 112, A09211, doi:10.1029/2007JA012269. 5. D´ecr´eau, P. M. E., Fergeau, P., Krasnosels’kikh, V., Le Guirriec, E., L´evˆeque, M., Martin, Ph., Randriamboarison, O., Rauch, J. L., Sen´e, F. X., S´eran, H. C., Trotignon, J. G., Canu, P., Cornilleau, N., de F´eraudy, H., Alleyne, H., Yearby, K., M¨ogensen, P. B., Gustafsson, G., Andr´e, M., Gurnett, D. A., Darrouzet, F., Lemaire, J., Harvey, C. C., Travnicek, P., and Whisper experimenters (2001), Early results from the Whisper instrument on CLUSTER: an overview, Ann. Geophys., 19, 1241–1258. 6. Donley, J. L., L.H. Brace, J.A. Findlay, J.H. Hoffman, and G.L. Wrenn (1969), Comparison of results of Explorer 31 direct measurement probes, Proc. IEEE, 5710, 1078–1084. 7. Escoubet, C. P., A. Pedersen, R. Schmidt, and P.-A. Lindqvist (1997), Density in the magnetosphere inferred from the ISEE-1 spacecraft potential, J. Geophys. Res., 102, 17,595– 17,609, doi:10.1029/97JA00290. 8. Escoubet, C. P., M. Ferhinger and M. Goldstein (2001), The Cluster mission, Ann. Geophys., 19, 1197–1200. 9. Etcheto, J. and Bloch, J. J., Plasma density measurements from the GEOS-1 relaxation sounder, Space Sci. Rev., 22, 597–610, 1978. 10. Etcheto, J., and A. Saint-Marc (1985), Anomalously High Plasma Densities in the Plasmasheet Boundary Layer, J. Geophys. Res., 90, 5338–5344. 11. Fazakerley, A., S. Schwartz, and G. Paschmann (1998), Measurement of Plasma Velocity Distributions in Analysis Methods for Multi-Spacecraft Data, edited by G. Paschmann and P. W. Daly, ISSI Scientific Report SR-001, ISSI/ESA. 12. Gurnett, D., and J. Green (1978), On the polarization and origin of auroral kilometric radiation, J. Geophys. Res., 83, 689–696. 13. Gurnett, D., S. Shawhan, and R. Shaw (1983), Auroral hiss, Z mode radiation, and auroral kilometric radiation in the polar magnetosphere: DE 1 observations, J. Geophys. Res., 88, 329–340. 14. Gurnett, D. A., R. L. Huff, J. S. Pickett, A. M. Persoon, R. L. Mutel, I.W. Christopher, C. A. Kletzing, U. S. Inan, W. L. Martin, J.-L. Bougeret, H. St. C. Alleyne, and K. H. Yearby, First results from the Cluster wideband plasma wave investigation (2001), Ann. Geophys., 19, 1259–1272.

278

A. Masson et al.

15. Gustafsson, G., M. Andr´e, T. Carozzi, A. I. Eriksson, C.-G. F¨althammar, R. Grard, G. Holmgren, J. A. Holtet, N. Ivchenko, T. Karlsson, Y. Khotyaintsev, S. Klimov, H. Laakso, P.-A. Lindqvist, B. Lybekk, G. Marklund, F. Mozer, K. Mursula, A. Pedersen, B. Popielawska, S. Saving, K. Stasiewicz, P. Tanskanen, A. Vaivads, and J.-E. Wahlund (2001), First results of electric field and density observations by Cluster EFW based on initial months of operation, Ann. Geophys., 19, 1219–1240. 16. Harvey, C. C., J. Etcheto, Y. de Javel, R. Manning, and M. Petit, The ISEE electron density experiment, IEEE Trans. Geosci. Electron., GE-16, 231–238, 1978. 17. Harvey, C. C., J. Etcheto and A. Mangeney (1979), Early results from the ISEE electron density experiment, Space Sci. Rev., 23, 39–58. 18. Johnstone, A. D., et al. (1997), PEACE, a plasma electron and current experiment, Space Sci. Rev., 79, 351–398, doi:10.1023/A:1004938001388. 19. Laakso, H., and A. Pedersen (1998), Ambient electron density derived from differential potential measurements, in Measurement Techniques in Space Plasmas: Fields, Geophys. Monogr. Ser., 102, edited by J. Borovsky, R. Pfaff, and D. Young, pp. 73–78, AGU, Washington, DC. 20. Laakso, H., R. Pfaff, and P. Janhunen (2002), Polar observations of electron density distribution in the Earth’s magnetosphere. Density Profiles, Ann. Geophys., 20, 1725–1735. 21. Labelle, J. and R. A. Treumann, Auroral radio emissions, hisses roars, and bursts, Space Sci. Rev., 101, 364–440, 2002. 22. Lefeuvre, F., M. Parrot, J. L. Rauch, B. Poirier, A. Masson, and M. Mogilevsky (1998), Preliminary results from the MEMO multicomponent measurements of waves on-board INTERBALL 2, Ann. Geophys., 16, 1117–1136. 23. Nagata, D., S. Machida, S. Ohtani, Y. Saito, and T. Mukai, Solar wind control of plasma number density in the near-Earth plasmasheet: three dimensional structure, Ann. Geophys., 26, 4031– 4049, 2008. 24. Nakagawa, T., T. Ishii, K. Tsuruda, H. Hayakawa, and T. Mukai (2000), Net current density of photoelectrons emitted from the surface of the GEOTAIL spacecraft, Earth Planets Space, 52, 283– 292. 25. Nykyri, K. and Otto, A. (2001), Plasma Transport at the Magnetospheric Boundary due to Reconnection in Kelvin-Helmholtz Vortices, Geophys. Res. Lett., 28(18), 3565–3568. 26. Øieroset, M., J. Raeder, T. D. Phan, S. Wing, J. P. McFadden, W. Li, M. Fujimoto, H. R`eme, and A. Balogh (2005), Global cooling and densification of the plasmasheet during an extended period of purely northward IMF on October 22–24, 2003, Geophys. Res. Lett., 32(12), L12S07, doi:10.1029/2004GL021523. 27. Parks, G. K., et al. (1992), Low-Energy particle Layer outside of the plasmasheet boundary, J. Geophys. Res., 97, 2943–2954. 28. Paschmann, G., A. Fazakerley and S. Schwartz (1998), Moments of Plasma Velocity Distributions in Analysis Methods for Multi-Spacecraft Data, edited by G. Paschmann and P. W. Daly, ISSI Scientific Report SR-001, ISSI/ESA. 29. Paschmann, G., Quinn, J. M., Torbert, R. B., Vaith, H., McIlwain, C. E., Haerendel, G., Bauer, O. H., Bauer, T., Baumjohann, W., Fillius, W., F¨orster, M., Frey, S., Georgescu, E., Kerr, S. S., Kletzing, C. A., Matsui, H., Puhl-Quinn, P., and Whipple, E. C.: The Electron Drift Instrument on Cluster: overview of first results, Ann. Geophys., 19, 1273–1288, 2001. 30. Pedersen, A. (1995), Solar wind and magnetosphere plasma diagnostics by spacecraft electrostatic potential measurements, Ann. Geophys., 13, 118–121, doi:10.1007/s00585-995-0118-8. 31. Pedersen, A., F. Mozer, and G. Gustafsson (1998), Electric field measurements in a tenuous plasma with spherical double probes, in Measurement Techniques in Space Plasmas: Fields, Geophys. Monogr. Ser., vol. 103, edited by R. F. Pfaff, J. E. Borovsky, and D. T. Young, pp. 1–12, AGU, Washington, DC. 32. Pedersen, A., P. D´ecr´eau, C. P. Escoubet, G. Gustafsson, H. Laakso, P.-A. Lindqvist, B. Lybekk, A. Masson, F. S. Mozer, and A. Vaivads (2001), Four-point high time resolution information on electron densities by the electric field experiment (EFW) on Cluster, Ann. Geophys., 19, 1491–1504.

18

Electron Density Estimation in the Magnetotail: a Multi-Instrument Approach

279

33. Pedersen, A., B. Lybekk, M. Andr´e, A. Eriksson, A. Masson, F. Mozer, P.-A. Lindqvist, P. M. E. D´ecr´eau, I. Dandouras, J.-A. Sauvaud, A. Fazakerley, M. Taylor, G. Paschmann (2008), Electron density estimations derived from spacecraft potential measurements on Cluster in tenuous plasma regions, J. Geophys. Res., 113, A07S33, doi:10.1029/2007JA012636. 34. Reinisch, B. W., et al. (2001), First results from the radio plasma imager in IMAGE, Geophys. Res. Lett., 28, 1167. 35. Santol´ık, O., A. M. Persoon, D. A. Gurnett, P. M. E. D´ecr´eau, J. S. Pickett, O. Marsalek, M. Maksimovic, and N. Cornilleau-Wehrlin (2005), Drifting field-aligned density structures in the night-side polar cap, Geophys. Res. Lett., 32, L06106, doi:10.1029/2004GL021696. 36. Scudder, J. D., X. Cao, and F. Mozer (2000), Photoemission current spacecraft voltage relation: Key to routine quantitative low energy plasma measurements, J. Geophys. Res., 105, 21,281, doi:10.1029/1999JA900423. 37. Stix, T. H. 1992, Waves in Plasmas, New York, American Institute of Physics. 38. Svenes, K.R., B. Lybekk, A. Pedersen, and S. Haaland (2008), Cluster observations of near-Earth magnetospheric lobe plasma densities – a statistical study, Ann. Geophys., 26, 2845–2852. 39. Terasawa, T., Fujimoto, M., Mukai, T., Shinohara, I., Saito, Y., Yamamoto, T., Machida, S., Kokubun, S., Lazarus, A. J., Steinberg, J. T., and Lepping, R. P. (1997), Solar Wind control of density and temperature in the near-Earth plasmasheet: WIND/GEOTAIL collaboration, Geophys. Res. Lett., 24, 935–938. 40. Thi´ebault, B., A. Hilgers, A. Masson, C. P. Escoubet, and H. Laakso (2006), Simulation of the Cluster-Spacecraft Floating Probe Potential, IEEE Nucl. Plasma Sci. Soc., 34(5), 2078–2083, doi:10.1109/TPS.2006.883407. 41. Torkar, K., Riedler, W., Escoubet, C. P., Fehringer, M., Schmidt, R., Grard, R. J. L., Arends, H., R¨udenauer, F., Steiger, W., Narheim, B. T., Svenes, K., Torbert, R., Andr´e, M., Fazakerley, A., Goldstein, R., Olsen, R. C., Pedersen, A., Whipple, E., and Zhao, H.: Active spacecraft potential control for Cluster – implementation and first results, Ann. Geophys., 19, 1289–1302, 42. Trotignon, J. G., Etcheto, J., and J. P. Thouvenin (1986), Automatic determination of the electron density measured by the relaxation sounder on board ISEE-1, J. Geophys. Res., 91, 4302–4320. 43. Trotignon, J. G., D´ecr´eau, P. M. E., Rauch, J. L., Suraud, X., Grimald, S., El-Lemdani Mazouz, F., Valli`eres, X., Canu, P., Darrouzet, F., Masson, A. (2006), The electron density around the Earth, a high level product of the Cluster/WHISPER relaxation sounder, Cluster and Double Star Symposium, 5th Anniversary of Cluster in Space, held 19–23 September, 2005 in Noordwijk, The Netherlands, edited by K. Fletcher, ESA SP-598, European Space Agency, published on CDROM., p. 70.1.

Chapter 19

Cluster-PEACE In-flight Calibration Status A.N. Fazakerley, A.D. Lahiff, I. Rozum, D. Kataria, H. Bacai, C. Anekallu, ˚ M. West, and A. Asnes

Abstract We briefly summarise key aspects of our on-going in-flight calibration work for the Cluster Plasma Electron And Current Experiment (PEACE) instruments, and demonstrate the quality of moments which may be achieved, by comparisons with measurements from other Cluster instruments. As improved calibrations are generated, data in scientific units which have been produced for the Cluster Active Archive will be systematically updated. This article is not intended as a detailed description of our calibration studies, but rather as a snapshot of the calibration status at the time of writing, which will give the researcher using the CAA an indication of the levels of accuracy that can be achieved at this time.

19.1 Introduction The CAA data for the Plasma Electron And Current Experiment (PEACE) and the experiment itself, have been introduced elsewhere in Chapter 8 of this book. The purpose of this paper is to briefly describe the PEACE calibration parameters, discuss which ones vary over time as the mission progresses and how the variation is accounted for. Examples of the accuracy of moments calculated from PEACE 3D data using the current calibration are provided to illustrate the effectiveness of our in-flight calibration work. The calibrations now available (version 5.1 at time of writing) allow determination of four-point plasma quantities, enabling studies of, for example, the divergence of the electron pressure tensor [2].

A.N. Fazakerley (), A.D. Lahiff, I. Rozum, D. Kataria, H. Bacai, C. Anekallu, and M. West Mullard Space Science Laboratory, University College London, UK e-mail: [email protected] M. West Now at Royal Observatory, Belgium ˚ A. Asnes ESA/ESTEC, Noordwijk, The Netherlands

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 19, c Springer Science+Business Media B.V. 2010

281

282

A.N. Fazakerley et al.

19.2 The Calibration Parameters 19.2.1 Introduction to the Calibration Parameters Every spin, each PEACE sensor makes a series of measurements in different look directions and energies to sample all or part of the full velocity distribution of the plasma electrons (as explained by Fazakerley et al., Chapter 8, this book). An individual measurement is intended to reveal the velocity space density of the electrons in a small region of velocity space, described in terms of a restricted energy (or equivalently speed) interval k and a direction defined by polar angle i and azimuth angle j , which is called fijk . The relationship between measured quantities, calibration factors and fijk is given as follows: fijk D Pijk = tacc vk 4 Gi "ik where: tacc is the data accumulation time, a fixed fraction of the spin period vk is the mean value of the measured electron speed during time tacc Gi is the geometric factor for the i th polar angle sector, which in a sufficiently concentric analyser, reduces to a single value G for all sectors Pijk is the number of counted electrons after dead time correction (related to instrument electronics not the MCP) and "ik D "0 " vk 2 i is the detector efficiency, which is expected to vary with position on the detector and with electron energy Ground calibration work provides values for the energies (speeds) measured during energy sweeps, vk , the geometric factor Gi and the relative sensitivity of the detector " vk 2 i , under conditions of optimum detector performance. These are used as the baseline from which any in-flight calibration corrections are made. It is challenging to establish good values for absolute detector efficiency "0 , during ground calibration, due to the difficulty of measuring the current in an electron beam in a calibration facility with sufficient accuracy (i.e., in generating a beam with very well determined fijk ). Furthermore, although the MCPs are kept purged with dry nitrogen as much as possible before flight, they are exposed to the atmosphere at least once between ground calibration and orbit (at the launch site) which may affect their performance.

19.2.2 Calibration Parameters: HEEA and LEEA The HEEA and LEEA sensors are intended to be identical, except for having different geometric factors (G) due to a different mechanical design for the electrostatic analyser entrance (see [3]). All eight sensors use the same equipment to control electron energy selection and to count detected electrons. Care was taken to make the electrostatic analysers of all the HEEAs as mutually identical as possible, and similarly for the LEEAs. The least controllable aspect of the design is the efficiency

19

Cluster-PEACE In-flight Calibration Status

283

."ik / of the individual micro-channel plate (MCP) detectors in each sensor. The calibration parameters mentioned above were measured in a test chamber to ensure that the individual sensors were well characterized and that their performance was identical to within specified tolerances before the instruments were accepted for flight.

19.2.3 Calibration Correction Factors Three aspects of the instrument calibration were expected to vary during flight operations. The detector efficiency was expected to decline over time, and to require occasional correction by using increased voltage levels across the MCP. The bulk velocity (determined by integration of fijk collected during each spin) was expected to possibly exhibit errors in the spin axis component due to errors in the determination of the relative sensitivity of the anodes; even small errors in inter-anode calibration could cause large errors in the bulk velocity measurement. The detector sensitivity was also expected to respond by under-counting in sufficiently high flux environments, and ground tests of flight detectors suggested this could sometimes affect HEEA (though not usually LEEA) in the magnetosheath. This scenario can be discussed in terms of additional factors in the expression for detector sensitivity, as follows "ik D ˛"0 ˇik " vk 2 i ijk where ˛.t/ is a time dependent correction factor for "0 , used to describe the effect of sub-optimal detector efficiency (equally for all anodes), ˇik represents corrections to " vk 2 i , the relative sensitivity of different parts of the detector, which may in principle vary with energy and ijk represents possible under-counting due to detector saturation in high flux environments (not to be confused with electronic dead time, which sets in at higher count rates). These parameters (˛, ˇ, ) were each equal to 1 in ground test conditions, when the MCP gain was well above 2 106 electrons and beam fluxes were not strong enough to cause detector saturation. In practice, as we will discuss below, we have found that the ˇik term shows dependence on the detector sensitivity, and thus a more elegant formulation of the problem might involve merging the ˛ and ˇ terms.

19.2.4 Relationship Between Calibration Parameters and Moments It can be shown that the plasma velocity (second order moment of f .v/) is independent of geometric factor and instrument sensitivity terms G˛"0 , but is dependent

284

A.N. Fazakerley et al.

on the relative sensitivity of different parts of the detector, i.e, the terms ˇik " vk2 i . Therefore, achieving plasma bulk flow vectors in agreement with independent measurements from other instruments is a good test of successful determination of the ˇik " vk2 i terms. It can also be shown that the plasma density is inversely proportional to the G˛"0 terms, as well as having a more complex dependence on ˇik " vk2 i . Thus accurate densities from PEACE show that all calibration factors are well characterized, including time dependent detector efficiencies. Conversely, to isolate the correct value for "0 (discussed further in Section 19.3.4) requires that ˛ and ˇ and all other calibration factors have been properly determined, that a reliable reference for the electron density is available and that proper moments calculations have been performed (see Section 19.3.1).

19.3 PEACE In-flight Calibration Method In-flight calibration is required to refine knowledge of "0 , and to keep track of (and provide corrections for) changing MCP detector efficiency. Calibration corrections are validated by determining bulk properties of the plasma and comparing them with measurements of the same parameters by instruments which use different techniques. If successful, the process also establishes that the majority of the calibration parameters were satisfactorily determined during ground tests (and that they are still applicable after the instrument has been strongly vibrated during launch).

19.3.1 Note on Determination of Electron Moments In this paper we will discuss electron density and velocity moments data. In all cases, these have been produced by performing moments calculations using transmitted 3-D data (most commonly 3DR), and they are not onboard-calculated moments. These moments are similar to the CAA Moments dataset (i.e. the first release of Moments data, which is the one available at the time of writing), although in some cases they are calculated for a single sensor only, while CAA Moments use data from both sensors. They have been calculated using PEACE Calibration version 5.1. They have been calculated using a correction to the measured energy to account for plasma electron acceleration by the spacecraft potential .Vsc /, which assumes that the true potential is 1 V greater than VEFW , the measured difference between the probe potential and the spacecraft potential, provided by the electric field instrument (EFW). In addition, electrons with measured energies in the energy bin that contains VEFW , and in the energy bin above that, have been assumed to be spacecraft photo-electrons and have not been included in the moments integration. It is worth noting that the conversion from VEFW to Vsc by adding 1 V was recommended by members of the EFW team, and that their recommendation has been

19

Cluster-PEACE In-flight Calibration Status

285

recently revised in the light of a recent study by Cully et al. [1] who showed that for tenuous plasmas such as the magnetotail lobes and plasmasheet, where the Debye length is long compared to the EFW boom length, VEFW underestimates Vsc by 19%. We may reformulate the Cully et al. results to express the spacecraft potential as Vsc D 0:86 1:24 VEFW recalling that VEFW is provided as a potential difference which is negative for a positively charged spacecraft (the expression also accounts for the 0.7 V voltage offset needed to balance the typical bias current used by Cluster EFW). It is interesting to note that the 25% larger value is in accordance with our estimates of spacecraft potential made using PEACE spectra alone (not shown here). Future calibrations will use PEACE moments which have been determined using this expression, where appropriate. It may be that when this formulation is in use, spacecraft photo-electrons will not be seen in the energy bin above the energy bin containing the inferred spacecraft potential. Care has been taken to isolate events for use in calibration analysis for which the relevant electron plasma population is fully sampled (e.g., it does not extend to higher or lower energies than those measured by PEACE). The plasma regions used are usually the magnetosheath and the magnetotail plasmasheet. Background counts have been ignored since their fluxes are considered to be negligible compared to the magnetosheath or plasmasheet electron fluxes in the sensor energy range (and events with high background count rates from, e.g., solar energetic particles, have not been included).

19.3.2 Determination of ˛ Correction Factors As a first step, it was essential to correctly assess the contribution of the ’ term, based on the detector sensitivity time history which we constructed using data from weekly tests of PEACE sensor performance. These tests are performed by setting both sensors to observe the same energy range, and then varying the MCP voltage on one sensor while keeping the other sensor at normal operating voltage. The process is then repeated with the roles of the sensors exchanged. The response of the measured count rate to increasing voltage in the test sensor can separated from count rate changes due to variations in the plasma environment using data from the second sensor. Note that data from other instruments such as WHISPER is not used in this procedure. Figure 19.1 provides an example of the time history of ’ for the Cluster 2 PEACE sensors. The variations in detector efficiency reflected here are due to a competition between the degrading effects of exposures to high electron fluxes (e.g., magnetosheath intervals), and the compensating effects of raised MCP voltage levels, complicated by a change of the PEACE operations plan from 2003, which exposed the sensors to higher fluxes less frequently.

MCP Level

A.N. Fazakerley et al. 30 25 20

C2L alpha

286

0.8

Cluster 2 : LEEA

15 1 0.6 0.4 0.2

MCP Level

30 25 20

C2H alpha

0 2001-01

0.8

2002-01

2003-01

2004-01

2005-01

2006-01

2007-01

2008-01

2009-01

Cluster 2 : HEEA

15 1 0.6 0.4 0.2 0 2001-01

2002-01

2003-01

2004-01

2005-01

2006-01

2007-01

2008-01

2009-01

Fig. 19.1 The plot shows the time history of the ’-factor for the Cluster 2 HEEA (below) and LEEA (above) sensors, together with the time history of the MCP levels used in normal operations (proxy for the voltage across the MCP). The main point to note is that the alpha factor which is ideally 1 has been held near 1 in the later years, but in the early years of the mission it tended to decline in the early part of each year, especially for HEEA, until the MCP level was raised. The sensitivity decline was associated with prolonged exposure to high fluxes in the magnetosheath and solar wind. The exposure to high fluxes has been limited in the later phase of the mission either by deactivating the sensor in such regions (according to a duty cycle) or by operating the sensors with lowered MCP voltages in those regions. The lowered MCP levels are not shown on the MCP level plots (upper panels), but their consequences (two sets of crosses per day) are apparent in the ’-factor plots, especially in early 2005. The curves for other sensors (not shown) exhibit similar features, though in some cases (particularly Cluster 3 and 4) the ’-factor values are not always so close to 1 later in the mission

19.3.3 Determination of ˇ Correction Factors Our approach to estimating inter-anode relative sensitivity corrections is based on the requirement that when measuring gyrotropic pitch angle distributions which are varying only on long timescales, the same pitch angle may be observed by more than one anode for a sensor on a spinning spacecraft, and all such anodes should provide the same measured values of fluxes (in scientific units) at that given pitch angle. The principle may be applied across the full pitch angle range. This general type of approach is also used for the CIS-CODIF sensors (see, for example, [4]). Our study applied this principle in a large scale automated approach, testing each measured energy at which the count rate was considered acceptable, for all spins of

19

Cluster-PEACE In-flight Calibration Status

287

data from data intervals where the plasma was selected to be relatively steady and with suitable temperature anisotropy. Temperature anisotropies in the range 0:95 < Tjj =T? < 5 (but commonly near 1) were used, provided they remained steady to within ˙0:05 during chosen intervals; such intervals ranged in duration from 20 to 360 s. The study focused on 3DR and 3DXP data, from which ground moments are usually calculated (see the companion paper on PEACE data products, Fazakerley et al., Chapter 8 of this book). The resulting correction factors ˇik were organized by energy and anode. In our initial studies, moments data produced using these calibrations were not very satisfactory. It was then recognized that the correction factors ˇ lso showed a dependence on ’. Therefore, by applying the appropriate correction factors for a given anode, energy and ’ (itself a function of time), the corrected calibrations could be used to produce moments which showed much improved accuracy (see Section 19.4.1). Note that while we will validate this method by examining the statistics of v?z values determined using our improved calibrations, the method does not make use of moments and in particular does not seek to minimize the average value of v?z .

19.3.4 Determination of "0 Calibration Parameter The absolute detector efficiency "0 was determined using comparisons of density measurements between PEACE and the plasma wave instrument WHISPER. Only when the ˛ istory is available, together with ˇ.˛/ is it possible to obtain a time independent value for "0 , through cross calibration with WHISPER (see Section 19.2.4). Our density reference was a large set of WHISPER electron density data during all magnetosheath and solar wind data intervals in 2001 and 2002, provided via the CAA. PEACE electron density moments were calculated using data from the LEEA sensors (to minimise the risk of count rate saturation which is greater for HEEA sensors) and applying ’ and “ factor efficiency corrections and a preliminary value of "0 . Figure 19.2 shows a quite narrow and tall histogram of ratios of PEACE/WHISPER densities, peaking near 0.8 for the particular case of Cluster 1 LEEA. The preliminary value of "0 was subsequently reduced so as to shift the peak to a value of 1. The new value of "0 is used in the updated calibration dataset. For events where the HEEA sensor also fully covered the energy range occupied by these solar wind and magnetosheath plasmas, the HEEA ©0 value was similarly adjusted to ensure agreement of HEEA, LEEA and WHISPER densities. There was no systematic attempt to filter out cases where ijk is less than 1, but some cases where a problem was clearly apparent were removed. The process was repeated for all four spacecraft. To illustrate how well the overall calibration (including all parameters mentioned in Sections 19.2.1 and 19.2.3) works, and how similar the results from the HEEA and LEEA sensors are, we present in Fig. 19.3 the ratio nL =nH of densities measured in the same band of energies in the same plasma by LEEA and HEEA. The plot covers 6 years (compared to the 2 year interval used for determining "0 ) and

288

A.N. Fazakerley et al.

Fig. 19.2 Illustration of the approach used to correct the ©0 absolute detector efficiency parameter. PEACE LEEA densities are calculated using efficiency corrections related to detector gain variations (’ and “), with a value for ©0 produced during ground calibration. A comparison is then made with densities determined by WHISPER. The examples here are Cluster 1 data from magnetosheath and solar wind intervals in 2001 and 2002 (using CAA WHISPER data). The most frequently occurring ratio of PEACE LEEA to WHISPER densities, indicated by the histogram peak, shows the discrepancy associated with imperfect knowledge of ©0 and can be used to determine the correct value of ©0

shows that the inter-calibration is usually very good, with the ratio being 1 ˙ 0:02 for much of the time. The main exceptions are intervals in the spring of years 2001, 2002 and 2003 where the ratio is sometimes >1, consistent with undercounting by HEEA in the high fluxes of the magnetosheath, i.e., cases when ijk < 1. After 2003, the HEEA sensors were used less frequently in the solar wind and magnetosheath. At present, we simply flag time intervals in the PEACE CAA files where may depart from 1 and do not attempt a correction.

19.4 Assessment of PEACE Calibrations 19.4.1 Effect of Inter-anode Sensitivity Corrections In order to discuss bulk velocity measurements, we will focus on v? , the velocity vector perpendicular to the local magnetic field, as it can be measured well by other instruments, facilitating comparisons (see below). Comparisons of the full velocity

19

Cluster-PEACE In-flight Calibration Status

289

Fig. 19.3 The effectiveness of the HEEA/LEEA relative calibration is illustrated for Cluster-1 in this plot comparing densities nL =nH . For each case, the density is calculated using 3DR data transmitted to the ground, applying the v5.1 calibrations in use at CAA at the time of writing, and applying corrections for spacecraft potential and removal of spacecraft electrons. The density nL is calculated using LEEA data in the energy overlap region while the density nH uses HEEA data in the overlap region. Thus any difference between the two densities is expected to be due to differences in calibration of the energy overlap region. Generally the ratio is very satisfactory, and lies within 1% or 2% of a perfect result, though there are some exception. The most obvious exceptions are in the early months of 2001, 2002 and 2003, when the HEEA sensor was used in the magnetosheath. Some of the magnetosheath events had sufficiently high fluxes that the HEEA sensor saturated and returned too few counts, hence leading to a density underestimate and a ratio >1. These may be considered as examples of situations where ” < 1 (see Section 19.2.3). For this reason, the CAA Moments data product uses the LEEA data in the energy overlap region

vector between, e.g. ion and electron instruments, may be less successful, for example if there are electron beams but not ion beams travelling along the magnetic field at the spacecraft location. We stated earlier that poor inter-anode calibrations give rise to significant errors in the spin axis component of the velocity vector, v?z . Figure 19.4 shows histograms of values of v?z for the four PEACE-HEEA sensors, using the uncorrected calibration values for " vk 2 i . The datasets are 208 selected quiet magnetotail plasmasheet intervals from 2001–2003. Corresponding data is also provided from CIS-CODIF on spacecraft C1 and C4 (see Section 19.4.3 for further details). The CIS histograms peak around 0 km/s and have a full width at half maximum of roughly 100 km/s. It is clear that the distribution of v?z is not consistent amongst the four PEACE-HEEA sensors, and nor do the distributions resemble those from CIS (with the possible exception of C1).

290

A.N. Fazakerley et al.

Fig. 19.4 Illustration of the quality of spin-axis components of v? from PEACE-HEEA (red) and CIS-CODIF (blue) for a set of 208 magnetotail plasmasheet intervals during 2001–2003 BEFORE improvement of PEACE inter-anode relative calibrations, “. The plots use moments calculated from 3D data collected by the HEEA sensors

Figure 19.5 shows histograms of all three components of PEACE-HEEA v? data for a large collection of data intervals from the quiet magnetotail plasmasheet, for which moments have been calculated using the current calibration (version 5.1). The dataset ranges from 2001–2007 on C1, C2 and C4, and 2001–2005 for C3. There are a greater number of data points for spacecraft C2 due to the greater availability of 3DR data from that spacecraft, as discussed in a companion paper in this volume [? ]. For all spacecraft, and for all three velocity components, the most commonly occurring value is around zero, as expected for plasmasheet data. The results are grouped about the mean value in quite a tight distribution, typically having a range of 80–100 km/s at half maximum, and 200 km=s at 0.1 maximum. These results are similar to the CIS results of Fig. 19.4, and are much better than the PEACE results in the same figure, for which the calibration did not include ˇi k correction factors. The ˇ corrections needed when detector efficiency was similar to that used in the ground calibration are minor (not shown), confirming that the ground calibration was quite effective, but the corrections become increasingly important at smaller detector gains. Nonetheless, even relatively minor corrections were found to significantly improve spin axis velocity moment components.

19

Cluster-PEACE In-flight Calibration Status

291

Fig. 19.5 Illustration of the quality of spin-axis components of PEACE-HEEA v? for a set of plasmasheet intervals from 2001–2007 on spacecraft C1, C2 and C4, and 2001–2005 on C3, AFTER improvement of inter-anode relative calibrations. The histograms indicate that the parameter is as well measured by PEACE as by CIS-CODIF

19.4.2 EFW-FGM-PEACE Comparisons In addition to comparisons with CIS, the v?z parameter determined by PEACE is also expected be consistent with the electric field drift velocity determined using electric and magnetic field data from EFW and FGM. The electric field is only available in the spin plane, so it is only the spin axis component of vE D ExB=B 2 , derived from spin plane components of E and B, that is well determined by this method. These data are less reliable in cases of weak magnetic field, or magnetic field vector near the spin plane. Comparisons of –vxB with individual components of E are also possible, but are not shown here. Figure 19.6 shows a set of electric field derived v?z values from 2001 provided by the CAA (only via the command line interface at the time of writing), which are available for all four spacecraft, together with the corresponding values from PEACE (using dual sensor moments). The dataset includes plasmasheet and magnetosheath intervals. The electric field derived values are also centered on 0 km/s, but show a much narrower spread, and

292

A.N. Fazakerley et al.

Fig. 19.6 Illustration of the difference between spin-axis component of v? as determined by from ExB=B 2 and by PEACE, for magnetosheath and plasmasheet events in 2001. Note that in all cases the most commonly occurring value is 0 km/s, and that the electric field derived values show a very much narrower spread of values. A much larger number of electric field derived results are seen at small speeds, so the histograms were normalized to 1 to allow a clear comparison

when presented as an occurrence frequency histogram, a correspondingly taller peak; hence we show the two histograms together after normalization. Although the electric field derived values are often very good, it is worth recalling that the requirements on the magnetic field mean that the technique is not applicable in such a wide range of situations as those for which CIS and PEACE can measure bulk flow velocities. Also, occasionally, issues such as incorrect EFW bias currents, or cold ion wakes (or, usually less significantly, the electric field offset in the sun direction) may introduce errors which the user should be cautious about. Figure 19.7 compares the two types of v?z measurements in a specific randomly selected interval. During this rather long interval, in which the spacecraft crosses from the magnetosheath to the polar cap via the cusp, the two measurements generally agree to within 30 km/s or better, i.e., to within 20% when significant (100 km=s) flows are present.

19

Cluster-PEACE In-flight Calibration Status

293

2

Diff. Energy Flux (ergs/cm -s-str-eV)

Cluster-1 > PEACE > PEACE PITCH_SPIN data from the HEEA and LEEA sensors

Fig. 19.7 Illustration of the accuracy of PEACE v?z moments for a 14 h inbound pass beginning in the magnetosheath, passing through the southern high altitude cusp and entering the polar cap. The spectrogram confirms that the LEEA sensor measured the complete energy range occupied by the magnetosheath/cusp plasma. The black trace in the spectrogram indicates the EFW probespacecraft potential energy, as a proxy for the slightly larger energy that is gained by an electron accelerated through the spacecraft potential, which must be allowed for in the moments calculation. The next panel shows the v?z values from PEACE and from the ExB method (from CAA files) in red and black. The next panel shows the difference (PEACE – ExB) between them and the lowest panels show the ratio of these two time series (PEACE/ExB). The ratio is naturally rather large where the values are very small (intervals highlighted by red horizontal bars) but where there are significant flows we see ratios in the range 1 ˙ 0:2. In all regions we see differences of order only 20–30 km/s

19.4.3 CIS-PEACE Comparisons As has been shown in Fig. 19.4, a similar test can be carried out using ion velocity measurements from the CIS experiments, although not for all spacecraft. CIS is inoperative on C2 and does not provide reliable v?z on C3 except during the early part of the mission. The v?z parameter is used as a measure of the local convection speed of magnetic flux, which should be the same for ions and electrons that are all “frozen-in” to the local magnetic flux.

294

A.N. Fazakerley et al.

Fig. 19.8 Illustration of the difference Œv?z CODIF Œv?z PEACEHEEA , showing results for current version 5.1 calibrations (green) and earlier version 4 calibrations (red). Note that the difference histogram shows a similar spread to the histograms of CODIF and HEEA data in Figs. 19.4 and 19.5, suggesting that the spread is due to fluctuations which are un-correlated between PEACE and CIS

The CIS measurements have been calibrated by the CIS team, in a similar way to PEACE, prior to this comparison (see [4]). The ion moments used were calculated using the CIS team “cl” software for CODIF on C1 and C4. This dataset was not checked to eliminate cases where perfect agreement is not expected, such as when the CIS or PEACE moments are calculated using incomplete coverage of the range of energies occupied by the plasma energy distribution. Figure 19.8 shows residuals Œv?z CODIF Œv?z PEACE calculated for each of the 208 plasmasheet intervals, using the present (v5.1) and previous (v4) PEACE calibration datasets, for the same intervals as in Fig. 19.4. The improvement in the PEACE results due to the v5.1 calibration is clear. The residuals histogram shows a similar spread to that of the CODIF histogram (Fig. 19.4) and the PEACE histograms (Fig. 19.5). This suggests that the v?z values are not correlated between the sensors, and that both represent a degree of random scatter about the “true” value of v?z . If the measurements had all been “perfect” so that the histogram represented a range of conditions represented in the plasmasheet, the residuals would have been zero, unless differential flows associated with electric current were present. On the other hand, this scenario is inconsistent with the narrow spread of values we see in the electric field based v?z (Fig. 19.6).

19.4.4 PEACE-WHISPER Comparison Figure 19.9 shows a comparison of electron density values determined by PEACE and by WHISPER, from a randomly selected magnetosheath interval. The data files were sourced from the CAA. The difference between the two datasets is typically less than 1 cm3 , and the ratio between them shows agreement to within 5%.

19

Cluster-PEACE In-flight Calibration Status

295

2

Diff. Energy Flux (ergs/cm -s-str-eV)

Cluster-2 > PEACE > PEACE PITCH_SPIN data from the HEEA and LEEA sensors

Fig. 19.9 Illustration of the accuracy of PEACE density moments for a 1 h magnetosheath interval. The spectrogram confirms that the LEEA sensor measured the complete energy range occupied by the magnetosheath plasma. The black trace indicates the EFW probe-spacecraft potential energy, as a proxy for the slightly larger energy that is gained by an electron accelerated through the spacecraft potential, which must be allowed for in the density calculation. The next panel shows the PEACE and WHISPER densities (from CAA files) in red and black. The lower panels show that the ratio of these two time series is generally around 95%, sometimes closer to 100%. The intervals where the ratio departs significantly from 1 are periods of elevated spacecraft potential, and the WHISPER CAA file quality factor is rather low here, indicating less reliable density determination from the WHISPER wave data

There is an ongoing activity to seek useful test intervals in higher energy plasmas for which reliable density is available from WHISPER or WIDEBAND, but far fewer satisfactory intervals are available than for the magnetosheath, as the density tends to be too low for these instruments to give accurate results (e.g. see Masson et al., Chapter 18 of this book). Although we have no reason to think that there is a problem at higher energies, density comparisons for cases with electron energies in the plasmasheet range would complete the in-flight validation of the calibration across the full PEACE instrument energy range.

296

A.N. Fazakerley et al.

19.4.5 PEACE-RAPID Comparisons In order to investigate the inter-calibration of PEACE and RAPID-IES, some preliminary studies have been carried out, using comparison plots such as the example shown in Fig. 19.10. The measurements made by IES were made at the same time as the PEACE measurements. We make the reasonable assumption that the same plasma population is being measured both the top of the PEACE energy range and in the IES range, so that the spectral gradient is expected to be the same according to both instruments. The two parallel green guidelines in the figure show that a common gradient can be seen. The only PEACE points which do not follow the trend well are interpreted as being at energies where the signal falls below instrument background. The general conclusions are that the results based on the existing

10−16

10−18

Parallel Perpendicular Anti-parallel

PEACE “1count’’

PEACE-HEEA data PSD (s3 / m6)

10−20 PEACE “1/8 count’’

10−22

10−24

10−26

RAPID-IES data 10−28

102

103

104

105

Energy (eV)

Fig. 19.10 Representative comparison plot showing electron phase space density vs energy measurements for three pitch angles from PEACE-HEEA and RAPID-IES on Cluster 2 on 10 September 2001, averaged over the interval 08:04–07 UT. There is an energy coverage gap intrinsic to the instrument designs. The trend in the PEACE data above 5 keV (indicated by parallel green lines) is quite well matched by the RAPID data above 50 keV. The plot also shows dashed lines which indicate the phase space density values corresponding to a hypothetical average background count level of 1 count per accumulation in PEACE data, and of a more realistic 1/8 count level background. In this case, the background magnetic field is aligned close to GSE X, so that the anode which measured the parallel .0ı / pitch angle looks anti-sunward. The parallel trace clearly diverts from the data trend to follow the background curve at the highest PEACE energies. The traces for the other two pitch angles follow similar trends, but to show an expected slightly higher background associated with anodes facing sunward

19

Cluster-PEACE In-flight Calibration Status

297

instrument calibrations, without any inter-calibration adjustments, are already quite good (although how good has not yet been quantified). It is clear from the figure that when making such comparisons one needs to be aware of the background count level in PEACE data since the plasma fluxes are often relatively low at the high energy end of the PEACE energy range. The studies have also shown that some care is needed in treating the low energy end of the IES energy range. This work is expected to be discussed in more detail elsewhere.

19.4.6 PEACE Inter-instrument Comparisons Figures 19.11 and 19.12 illustrate good relative accuracy between the 4 Cluster PEACE instruments for randomly selected magnetosheath and plasmasheet burst mode intervals, using the CAA PEACE (version 1) dual sensor Moments data product. As in earlier figures, v? is used to avoid differences that might arise if there are magnetic field-aligned beams at some spacecraft and not others. Exact agreement

Fig. 19.11 Illustration of the relative accuracy of CAA PEACE Ver1 moments for randomly selected 1 h burst mode (BM) telemetry intervals in the magnetosheath. Spacecraft separation is 200 km. Data traces use the usual colour code: C1 black, C2 red, C3 green, C4 blue. Moments data from C1 and C3 is less frequent than spin rate, even though this is BM, as complete 3D distributions from both PEACE sensors are not available for each and every spin. Agreement is quite good in all parameters

298

A.N. Fazakerley et al.

Fig. 19.12 Illustration of the relative accuracy of CAA PEACE Ver1 moments for randomly selected 1 h burst mode (BM) telemetry intervals in the magnetotail plasmasheet. Spacecraft separation is 4;000 km: Data traces use the usual colour code: C1 black, C2 red, C3 green, C4 blue. Moments data from C1 and C3 is less frequent than spin rate, even though this is BM, as complete 3D distributions from both PEACE sensors are not available for each and every spin. Agreement is quite good in all parameters

is not expected as the spacecraft are not perfectly co-located (200 km apart in the magnetosheath case and 4;000 km: apart in the plasmasheet case). Nonetheless, values are generally very similar and the time series show the same trends between spacecraft. The CIS data (not shown) exhibit very similar behaviour, including the unusual variations in the y and z components of the magnetosheath flow in Fig. 19.11 (probably related to solar wind current sheets transiting the magnetosheath) and the enhanced plasmasheet flows near 09:18 UT in Fig. 19.12.

19.5 Summary Our paper has briefly described the PEACE calibration parameters and our approach to improving them through in-flight calibration work. It is outside the scope of this paper to provide a detailed account of the inter-calibration methods. Validation of the calibrations using comparisons with datasets from other Cluster instruments is also demonstrated.

19

Cluster-PEACE In-flight Calibration Status

299

We include examples showing CAA PEACE moments data which show good agreement (to within a few 10’s km/s) between the (difficult to determine) spin axis component of v? and the values derived from electric and magnetic field data, and magnetosheath electron densities that agree within 5% of densities derived from plasma wave instruments. We also provide evidence of effective inter-sensor inter-calibration and effective inter-calibration between PEACE instruments on the 4 Cluster spacecraft. Acknowledgments The authors thank ESA and STFC for funding. We are also grateful to other Cluster instrument teams and the CAA team for their support.

References 1. Cully, C.M., R.E. Ergun and A.I. Eriksson, Electrostatic structure around spacecraft in tenuous plasmas, J. Geophys. Res., 112, A09211, doi:10.1029/2007JA012269, 2007. 2. Henderson, P.D., C.J. Owen, A.D. Lahiff, I.V. Alexeev, A.N. Fazakerley, L. Yin, A.P. Walsh, E.A. Lucek, and H. Reme, The relationship between jxB and div(Pe) in the magnetotail plasma sheet: Cluster observations., J. Geophys. Res., 113, A07S31, doi:10.1029/2007JA012697, 2008. 3. Johnstone A.D., C. Alsop, S. Burge, P.J. Carter, A.J. Coates, A.J. Coker, A.N. Fazakerley, M. Grande, R.A. Gowen, C. Gurgiolo, B.K. Hancock, B. Narheim, A. Preece, P.H. Sheather, J.D. Winningham, and R.D. Woodliffe, Peace: A Plasma Electron and Current Experiment, Space Sci. Rev. 79, pp. 351–398, 1997. 4. McFadden et al., In-flight Instrument Calibration and Performance Verification, in Calibration of Particle Instruments in Space Physics, M. Wuest, D.S. Evans and R. von Steiger (Editors), ISSI Scientific Report SR-007, International Space Science Institute, Bern, 2007.

Chapter 20

Generation and Validation of Ion Energy Spectra Based on Cluster RAPID and CIS Measurements Elena A. Kronberg, Patrick W. Daly, Iannis Dandouras, Stein Haaland, and Edita Georgescu

Abstract We present a method to combine spectral information from the Cluster RAPID (Research with Adaptive Particle Imaging) and CIS (Cluster Ion Spectroscopy) instruments. Preliminary results from the magnetotail show that the kappa distribution can fit the ion distributions in the magnetotail, for example in thin plasma sheets during quiet and disturbed times with kappa factors '4.4 and '10, respectively. With the most recent RAPID and CIS/CODIF calibration files the average ratio of CIS fluxes (approximated using the kappa-fit to the effective energy of the first RAPID energy channel) to the RAPID fluxes at the first energy channel is about 0.74 ˙ 0.24. This result is perfectly acceptable for these two experiments, based on two very different measurement principles, therefore no further spectral shift or recalibrations are needed.

20.1 Introduction Knowledge about the spectral distribution of a particle populations provides important information about e.g. acceleration and diffusion processes. For instance, earlier studies indicate that the kappa distribution (a Maxwellian core with a power law tail) rather than the Maxwellian will lead to enhanced wave growth [10, 12]. Sarris et al. [9], using IMP-7 and eight spacecraft data, have reported on plasma sheet ion distributions over the energy range 0.1 keV to a few MeV in the distant terrestrial magnetotail. They found that during periods of low bulk velocity the ion

E.A. Kronberg (), P.W. Daly, S. Haaland, and E. Georgescu Max-Planck Institute for Solar System Research e-mail: [email protected]; [email protected]; [email protected]; [email protected] I. Dandouras CESR, Universit´e de Toulouse / CNRS, Toulouse, France e-mail: [email protected] S. Haaland University of Bergen, Norway

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 20, c Springer Science+Business Media B.V. 2010

301

302

E.A. Kronberg et al.

distribution is often best characterized by a Maxwellian but with a high energy (50 keV) tail described well by a power law (/E ). A case study of the thermal and suprathermal protons in the Earth’s quasistable plasma sheet over the energy range 13 keV to 1 MeV has been made by Ipavich and Scholer [5] using ISEE-1 data. They reveal that the temporal behaviour of suprathermal ions in the plasma sheet is a complex one, with numerous energy-dependent structures. The typical spectral shape is: a thermal component below 16–19 keV that has the Maxwellian distribution, but the high energy spectrum departs from the Maxwellian extrapolation. The suprathermal distributions fall off steeply beyond 100–200 keV with a high-energy component extending up to 1 MeV. Christon et al. [1, 2] using ISEE 1 data have reported on plasma sheet ion distributions over the energy range 0.1 keV to a few MeV. They find that during periods of low bulk velocity the ion distribution is often best described by a kappa distribution. Further studies by Christon et al. [3] reveal that the spectrum observed during disturbed geomagnetic conditions typically is complex and cannot be represented in general by a single functional form, as during quiet periods when it can be represented by the kappa distribution function. A power law form most able to approximate the shape at higher energies and a Maxwellian form most able to approximate the shape near the knee,1 can be used to characterize the observed distribution. The purpose of this paper is to present a method to fit Cluster ion data to a kappa distribution and to show how it works for different applications. We show the specific of the spectral fitting for the Cluster data, over the combined energy range of the RAPID/IIMS and the CIS/CODIF experiments. The results can be used further for the spectral studies.

20.2 Data and Methodology The measurements of the proton fluxes are obtained by two particle experiments: RAPID/IIMS (Imaging Ion Mass Spectrometer) for the higher energies and CIS/CODIF (Composition and Distribution Function) for the medium and lower energies. The centerpiece of the IIMS is a spectrometric camera for neutral and ion composition. This is a miniature telescope composed of a time-of-flight and an energy detection system which allows us to distinguish different species [11]. The instrument measures 3D distribution of ions and covers the energy range from 27.7 to 1,500 keV for the protons. The CODIF instrument is a high-sensitivity mass-resolving spectrometer which measures full 3D distribution functions of different ion species [8]. The spectrometer combines ion energy per charge selection, by deflection in a rotationally symmetric toroidal electrostatic analyser, with a subsequent time-of-flight analysis. The energy range of CODIF is 0–40 keV/q. 1 The knee is that portion of the spectrum where flux begins to roll over and decrease rapidly with increasing energy [3].

20

Energy Spectra from Cluster RAPID and CIS Measurements

303

These two instruments have an energy overlap which allows us to intercalibrate and verify the data quality. The overlapping channels are the 29–37 keV (CODIF) and 28–64 keV (IIMS). The first six CIS channels are not used as their fluxes are often very scattered (presence of different plasma populations). The RAPID/IIMS instrument has three detector heads which cover the full range of 0–180ı in the polar direction. Unfortunately the central head (60–120ı) has been degraded by sunlight and doughnut shaped data gap appears in the 3D ion distribution. We tried to compensate for this effect by removing measurements from the corresponding part of the CODIF sensor. However, unless the distributions were extremely anisotropic this correction had little effect on the energy spectra above approximately 2 keV. We therefore decided to use the full directional range of CODIF for the fitting. The kappa distribution can be written as follows (taken from Krimigis et al. [6])

. C 1/ n jv Vj2 1C f .v/ D 3=2 3 3=2

. 1=2/ !0 !02

!.C1/ (20.1)

where n is the particle density, v the particle velocity, V is the bulk velocity, is the kappa-distribution power law spectral index, is the Gamma-function and the most probable speed, is given by p

. 1=2/ 1=2 !0 D 2kTH =M . 3=2/

(20.2)

where TH is the temperature of the plasma at E 1 keV, M is the particle mass and k is the Boltzmann constant. When v V the spectrum approaches a power law in energy and the Maxwellian distribution is a special case of expression (20.1) as ! 1:

20.2.1 Fitting Method The fit is obtained by minimization of the difference between the observed fluxes and the values of the kappa fit at the same energies using the following function gD

X .fobs .v/ fexp .v//2 N.N 1/fobs .v/

(20.3)

where fobs .v/ is the observed flux, fexp .v/ is the kappa-fit, and N is the number of fitted points. Therefore we get basically chi-square criteria. The minimization was done using Powell’s direction set method [7]. As free parameters of the kappa-fit we used and kTH : The plasma density and the bulk velocity are defined using CIS/CODIF data and the integration method explained in the CIS-CAA Interface Control Document (ICD) – see Dandouras and Barthe [4].

304

E.A. Kronberg et al.

For lower energies, i.e., most of the CODIF energy range, the bulk motion, V, constitutes a significant part of the total velocity. Prior to fitting, we therefore transform the CIS data into the bulk flow frame. For RAPID, on the other hand, the bulk velocity is negligible compared to the gyro motion, and no such correction has been made.

20.3 Examples Figure 20.1 shows two examples of the kappa fit in the thin plasma sheet during a quiet and a disturbed time. The data are plotted in the flow frame and fitted using energy channels with energies from 1 keV to 4 MeV. One can see significant temperature and spectral index increases during the substorm time. The value of the 4:4 for the quiet time is comparable with the typical values 4–8 derived for the quiet times by Christon et al. [2]. During the disturbed time 10 is in the range of the observed values during disturbed geomagnetic conditions from 3 to 11. We also compared the derived kTH with the temperature obtained from moment calculations. During disturbed time conditions (right panel) the peak of the spectra is shifted, and we obtain a temperature kTH 10:5 keV. For comparison, the CIS moments give 13 keV. During quiet conditions (left panel), we obtain a temperature at 3 keV (CIS moments: 4.7 keV). 10:52:30

23:24:29 106

106

Hydrogen flux, [1/cm2 sr s keV]

Quiet

Disturbed

105

105

104

104

103

103

102 101 100 0.01

102

κ = 4.36 kT = 2.98

0.1

101

1

10

100 1000

100 0.01

κ = 9.71 kT = 10.53

0.1

1

10

100 1000

Energy, [keV]

Fig. 20.1 Example of energy spectra and kappa-fit for the periods: on 20 September 2002, 23:24– 23:25 UT, SC1, during the quiet time in the thin plasma sheet (left) and on 17 September 2003, 10:52–10:53, SC1, during substorm in the thin current sheet. For both examples we fitted the CIS data from 1 keV. The diamonds are the joint CIS/CODIF and RAPID/IIMS data and the thick solid lines are the kappa-fit. kT is in keV

20

Energy Spectra from Cluster RAPID and CIS Measurements

305

20.4 CIS/CODIF and RAPID/IIMS Cross-Calibration The original purpose of this study was to accent the cross-calibration between CIS/CODIF and RAPID/IIMS instruments. The RAPID/IIMS particle fluxes of the lowest energy channel can not be directly compared with CIS/CODIF particle fluxes of the highest energy channel due to their nonsymmetric energy overlapping. The method described above was applied to the CIS/CODIF data in order to be able to compare particle fluxes at the same energies. Events in the plasma sheet region for 2002–2005 are investigated. The results show that the RAPID ion fluxes and CIS ion fluxes are consistent with each other, and do not have a difference in order of magnitude as it was before using direct comparison of the overlapping channels and old calibrations. With new RAPID calibration files and the most recent CIS/CODIF calibration files the average ratio of CIS fluxes (approximated using the kappa-fit to the effective energy of the first RAPID energy channel) to the RAPID fluxes at the first energy channel (27.7–64.4 keV) is about 0.74 ˙ 0.24. Here we applied the kappa-fit only to the higher CIS channels (5.5–38.3 keV) in order to improve the reliability of the approximated overlapping channel. More detailed information on the ratios can be seen in Table 20.1. This result is perfectly acceptable for these two experiments, which are based on two very different measurement principles; (solid state detectors in RAPID versus electrostatic analysers in CODIF). It should be also taken into account that CODIF detects only positive ions whereas RAPID is sensitive to both ions and energetic neutral atoms. We therefore conclude that no further spectral shift or recalibrations are needed.

20.5 Summary and Outlook In this paper we applied the kappa-distribution fit to the joint CIS/CODIF and RAPID/IIMS spectra. Kappa-fit works quite well for the energy range between 1 and 600 keV in the terrestrial magnetotail and gives consistent results for the combined measurements of the two experiments. At the same time there are a number of cases when this fit does not work, for example in thick plasma sheets or during ion beams.

Table 20.1 Ratios of the CIS fluxes (approximated using the kappa-fit to the effective energy of the first RAPID energy channel) to the RAPID fluxes at the first energy channel (27.7–64.4 keV). Data from SC3 were not used for analysis after 2003 because half of anodes on the CODIF instrument were not working satisfactory Year C1 C3 C4 2002 0.69 ˙ 0.09 (15 events) 0.62 ˙ 0.44 (31 events) 0.8 ˙ 0.09 (16 events) 2003 0.8 ˙ 0.17 (20 events) 1.1 ˙ 0.09 (34 events) 2004 0.76 ˙ 0.15 (31 events) 0.95 ˙ 0.19 (27 events) 2005 0.61 ˙ 0.13 (35 events) Average 0.75 ˙ 0.13 (66 events) 0.62 ˙ 0.44 (31 events) 0.86 ˙ 0.15 (112 events)

306

E.A. Kronberg et al.

The problem with the central head at the RAPID does not significantly affect the results of the kappa-fit. Therefore the CIS data are approximated without removing the corresponding data gap. However, cases with strong anisotropy should not be fitted with the kappa-fit and it is recommended to study them separately. The methods described in this paper will be used for the survey of the spectral characteristics and its dependence on solar wind parameters in the terrestrial magnetosphere. Acknowledgements We thank A. Barthe for preparation of the CIS data.

References 1. Christon S.P., Mitchell D.G., Williams D.J., Frank L.A., Huang C.Y., Eastman T.E.: Energy spectra of plasma sheet ions and electrons from about 50 eV/e to about 1 MeV during plamsa temperature transitions. J. Geophys. Res. 93, 2562–2572 (1988) 2. Christon S.P., Williams D.J., Mitchell D.G., Frank L.A., Huang C.Y.: Spectral characteristics of plasma sheet ion and electron populations during undisturbed geomagnetic conditions. J. Geophys. Res. 94, 13,409–13,424 (1989) 3. Christon S.P., Williams D.J., Mitchell D.G., Huang C.Y., Frank L.A.: Spectral characteristics of plasma sheet ion and electron populations during disturbed geomagnetic conditions. J. Geophys. Res. 96, 1–22 (1991) 4. Dandouras I., Barthe A.: Cluster Active Archive: Interface Control Document for CIS. ESA document CAA–CIS–ICD–0001 (2007). Latest version available at http://caa.estec.esa. int/documents/ICD/CIS CAA ICD V2.1.pdf 5. Ipavich F.M., Scholer M.: Thermal and suprathermal protons and alpha particles in the earth’s plasma sheet. J. Geophys. Res. 88, 150–160 (1983) 6. Krimigis S.M., Carbary J.F., Keath E.P., Armstrong T.P., Lanzerotti L.J., Gloeckler G.: General characteristics of hot plasma and energetic particles in the Saturnian magnetosphere - Results from the Voyager spacecraft. J. Geophys. Res. 88, 8871–8892 (1983) 7. Press W.H., Teukolsky S.A., Vetterling W.T., Flannery B.P.: Numerical Recipes in C: The Art of Scientific Computing (Second Edition), chap. 10.5. Cambridge University Press (1992) 8. R`eme H., Aoustin C., Bosqued J.M., et al.: First multispacecraft ion measurements in and near the Earth’s magnetosphere with the identical Cluster ion spectrometry (CIS) experiment. Annales Geophysicae 19, 1303–1354 (2001) 9. Sarris E.T., Krimigis S.M., Lui A.T.Y., Ackerson K.L., Frank L.A., Williams D.J.: Relationship between energetic particles and plasmas in the distant plasma sheet. Geophys. Res. Lett. 8, 349–352 (1981) 10. Thorne R.M., Summers D.: Enhancement of wave growth for warm plasmas with a high-energy tail distribution. J. Geophys. Res. 96, 217–223 (1991) 11. Wilken B., Daly P.W., Mall U., et al.: First results from the RAPID imaging energetic particle spectrometer on board Cluster. Annales Geophysicae 19, 1355–1366 (2001) 12. Xiao F., Zhou Q., He H., Zheng H., Wang S.: Electromagnetic ion cyclotron waves instability threshold condition of suprathermal protons by kappa distribution. Journal of Geophysical Research (Space Physics) 112, 7219–+ (2007). doi10.1029/2006JA012050

Part IV

Magnetospheric Missions

Chapter 21

The Cluster Mission: Space Plasma in Three Dimensions M.G.G.T. Taylor, C.P. Escoubet, H. Laakso, A. Masson, and M.L. Goldstein

Abstract At the time of writing, Cluster is approaching 8 years of successful operation and continues to fulfill, if not exceed its scientific objectives. After a nominal mission lifetime of 2 years Cluster currently in its extended mission phase, up to June 2009, with a further extension request submitted for a further 3.5 years. The primary goals of the Cluster mission include three-dimensional studies of smallscale plasma structures and turbulence in the key plasma regions in the Earth’s environment: solar wind and bow shock, magnetopause, polar cusps, magnetotail, and auroral zone. During the course of the mission, the relative distance between the four spacecraft is being varied to form a nearly perfect tetrahedral configuration at 100, 250, 600, 2,000, 5,000 and 10,000 km inter-spacecraft separation targeted to study scientifically interesting regions at different scales. In the last few years, the constellation strategy has moved towards a multi-scale concept, enabling two scale sizes to be investigated at the same time. In these cases, three spacecraft are separated by 10,000 km with the last spacecraft separated from this plane by varying distances from 16 km up to several 1,000 km. This configuration is targeted at boundaries, with the plane of the large-scale triangle parallel to the plane of the boundary and the final spacecraft separated a small distance from the main triangle in the normal direction. In this paper, we provide a brief overview of the mission concept and implementation and highlight a number of Cluster’s latest science results, which include: the first observation of three dimensional (3-D) surface waves on the bow shock, the first 3-D analysis of turbulence in the magnetosheath, the discovery of magnetosonic waves accelerating electrons to MeV energies in the radiation belts, along with a number of discoveries involving magnetic reconnection.

M.G.G.T. Taylor (), C.P. Escoubet, H. Laakso, and A. Masson ESA/ESTEC, D-SRE, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands e-mail: [email protected] M.L. Goldstein Code 673, NASA/GSFC, Greenbelt, USA 20771

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 21, c Springer Science+Business Media B.V. 2010

309

310

M.G.G.T. Taylor et al.

21.1 Introduction Together with the Solar and Heliospheric Observatory (SOHO), the Cluster mission constitutes the first cornerstone of ESA’s Horizons 2000 Programme, which was presented to and subsequently selected by the ESA Science Programme Committee (SPC) in late 1985 early 1986. Cluster was first proposed in November 1982 in response to an ESA call for proposals for new scientific missions [7] as a mother and three daughter spacecraft. Economic constraints led to the four identical spacecraft being adopted and scientific arguments led to choosing a polar orbital plane to allow for observations of the polar cusp and magnetotail. In March 1988, following a (1987) joint ESA/NASA Announcement of Opportunity, 11 instruments were selected to make up the spacecraft science payload. During the following 7 years instruments and spacecraft were built, integrated and made ready for the infamous failure of the first Ariane 5 launch on 4 June 1996 when the rocket exploded 47 s after liftoff, destroying all four spacecraft. Quickly recovering from the shock of the loss, the instrument and project teams resolved to recover the mission objectives and following a number of science working team and ESA committee meetings, the Cluster team convinced the ESA SPC to rebuild the four spacecraft. In April 1997, Cluster II was born. Taking about half the time and half the cost of the original build, Cluster II was ready for launch in 2000. Even after the successful Soyuz launch of the first two Cluster spacecraft in July 2000 from the Baikonur Cosmodrome, the Cluster teams were not able to relax until after the successful launch and orbit injection of the subsequent pair in August 2000 and the instruments were finally turned on. The aim of this paper is not to review previous magnetospheric missions, but to provide an overview of the Cluster mission itself, in the scope of providing information for these proceedings focusing on orbit, spacecraft separation and instrumentation. We also discuss some recent results leading up to the Cluster/CAA workshop held in Tenerife. Finally we will briefly discuss the proposed future mission extension.

21.2 Mission The four identical spacecraft of the ESA Cluster mission provide a unique capability to study the temporal evolution of space plasma interactions in three dimensions. The near-Earth environment provides an ideal laboratory for investigating plasma interactions, transport and turbulence in a number of environments. These key regions of the magnetosphere are: the solar wind, the bow shock, the magnetopause, the polar cusp and the magnetotail. To sample these regions, the Cluster spacecraft were originally placed in a 4 19:6 RE polar orbit (orbital period 56 h), with the single spacecraft orbits designed to enable a perfect tetrahedron to be formed at particular parts of the orbit, aiming at science targets such as the polar cusps and plasma sheet (Fig. 21.1). Over the years the line of apsides has rotated further below

21

The Cluster Mission: Space Plasma in Three Dimensions

311

Fig. 21.1 Evolution of the Cluster orbit in the x–z GSE plane during the period 2001–2009

the ecliptic, such that now Cluster crosses the ecliptic plane much closer to the Earth (Fig. 21.1). This has allowed the mission to enter regions that were not in the original mission plan (e.g. the low-latitude magnetopause, near-Earth plasma sheet, auroral acceleration regions and radiation belts), thus enhancing the originally anticipated scientific output. The separation strategy during the entire mission to date is detailed in Table 21.1 and Fig. 21.2, where during the first 5 years a perfect tetrahedron was formed twice during each orbit (purple line in Fig. 21.2). During the first 2 years of the mission, the spacecraft separation was altered approximately every 6 months. During the first extension (January 2003–December 2005) manoeuvres were limited to once per year to reduce operational costs and instrument down time and to help ensure uniform spatial coverage of scientifically interesting regions at the same magnetic local time over the calendar year. This reduction was made possible thanks to the expertise of the Flight Dynamics team from the European Space Operations Centre (ESOC), who implemented a strategy that imposed a tetrahedron during the cusp passes and then 6 months later during the magnetotail plasma sheet crossing without large-scale reconfigurations. More details of the preparation and execution of the manoeuvres can be found in Volpp et al. [30]. In 2003/2004 the Cluster mission was augmented with the launch of the Double Star mission. This mission consists of two spacecraft, the equatorial-orbiting TC-1 (550 66;970 km, 28:5ı inclination) launched on 29 December 2003 and the polarorbiting TC-2 (700 39; 000, 90ı inclination) launched in July 2004. The orbits of

312

M.G.G.T. Taylor et al. Table 21.1 Cluster separation strategy split into cusp and tail seasons Year Phase Separation (km) 2001 Cusp 600 2001 Tail 2,000 2002 Cusp 100 2002 Tail 4,000 2003 Cusp 5,000 2003 Tail 200 2004 Cusp 300 2004 Tail 1,000 2005 Cusp 1,300 2005 Tail 10,000 (1,000) 2006 Cusp 10,000 2006 Tail 10,000 2007 Cusp 10,000 (450) 2007 Tail 10,000 (40) 2008 Cusp 10,000 (40) C Tilt Campaign 2008 Tail 10,000 (3,000) 2009 Cusp 10,000 tetrahedron in solar wind C 10,000 (1,000) in auroral region The figures in brackets from 2005 onwards represent the spacecraft 3 and 4 separation during the multi-scale configuration

Fig. 21.2 Detail of the Cluster inter-spacecraft separation strategy

Double Star were specifically designed to maximize the conjunction with Cluster in the plasma sheet and in the polar cusp and magnetopause. Furthermore half of the Double Star payload is made of spare or duplicate of the Cluster instruments, which has allowed full comparisons between all spacecraft to be made. TC-1 deorbited in October 2007 but TC-2 is still operating, although due to loss of attitude control it has periods of loss of contact as well as thermal issues and is expected

21

The Cluster Mission: Space Plasma in Three Dimensions

313

to last only to the end of 2009. More information can be found in Liu et al. [12] and also Dunlop et al. [Chapter 22 of this book] where a number of the science examples are discussed, outlining the benefits of multi-mission conjunction studies and demonstrating the great success of this first ESA-China collaboration. The Cluster spacecraft separation has gradually increased over the years, allowing the same regions of the magnetosphere to be sampled at different scales. This is a one-way strategy, as the spacecraft cannot return to a small tetrahedron configuration due to diminishing fuel reserves that no longer allow substantial separation distance changes, particularly perpendicular to the direction of the orbit. However, capability does exist to move spacecraft ‘along track’ in the direction of the orbit, which has facilitated the implementation of the multi-scale operations in the second extension (since January 2006). In this configuration, three of the spacecraft (Cluster 1, 2 and 3) form a 10,000 km sided triangle, while Cluster 3 and 4, whose orbits are very similar, can be drifted with respect to one another and their separation can be varied from a few 10s of km to 10,000 km. In other words, with such minor manoeuvres only, Cluster will have fuel to continue operations for a number of years. In addition to the multi-scale concept, in May 2008 a special operation was implemented to measure the third component of the electric field along the spin axis. This was achieved by tilting Cluster 3 and Cluster 4 with respect to one another by 45ı , while they were separated by only 40 km. Preliminary data show that the third component of the electric field and in particular the electric field parallel to the magnetic field can, indeed, be measured (Y. Khotyaintsev, private communication). In addition, the possibility of localizing the source of electromagnetic waves is greatly enhanced. During the current second extension period, a mid-term review was carried out in November 2007 and concluded that all four Cluster satellites were fit to enter the second half of the extension. ESA has recently undergone a review of future mission extension assessment, and to bring all mission extension timelines into line, has set the current end of the Cluster mission at 30 June 2009. A proposal has been submitted to extend the Cluster mission from 1 July 2009 to December 2012, with a mid-term review in 2010. The science case is strong, and the spacecraft are technically able to support the extension. Analysis of the spacecraft performance carried out for the autumn 2007 review demonstrated that the platforms are in good health and are able to continue operation until the end 2012. Due to the age of the spacecraft and the finite resources, certain aspects have to be considered (Battery health and fuel limitations for example), but the experience gained by the flight control team in recent years have meant that any aspects of ageing have had limited impact on the science return. In particular, the ESOC team has implemented a number of measures to enable varying levels of spacecraft operations during orbits containing eclipses to optimize science data return. We hope for a positive outcome to the request for further mission extension, as the new science opportunities available to the mission are unique, both in Cluster observations themselves, but in particular in conjunction with other highly complementary missions such as the NASA THEMIS (Time History of Events and Macroscale Interaction during Substorms) mission [1] and the ESA Earth observation multi-spacecraft mission Swarm [6].

314

M.G.G.T. Taylor et al.

21.3 Instrumentation Cluster payload commanding is carried out using the JSOC (Joint Science Operations Centre; URL http://www.jsoc.rl.ac.uk/) system, which creates the overall science plan or Master Science Plan (MSP) that covers a period of several months. This MSP forms the basis of a consolidated payload command schedule, which is iterated via a specialized interface with the principle investigator teams and the final version, which covers a single planning period, equivalent to 3 orbits or 1 week, is finally delivered ESOC for uplink to the spacecraft. The Cluster payload is a complete set of complementary instruments to measure magnetic fields, electric fields, particles (along with a spacecraft potential control device to enable low energy particle measurements in tenuous plasma) and electromagnetic waves (see [5] for more information on these instruments and the instrument archiving papers in this book). Table 21.2 provides an overview of the instruments and their principle investigators. At the time of writing, a vast majority of the instruments are still functioning nominally and provide a wealth of science data to the community. However, the CIS instrument on Cluster 2 is not operational and CIS-HIA is switched off on Cluster 4. We note that CIS-CODIF is still functional on Cluster 4, so thermal ion measurements are still possible on 3 of the 4 spacecraft. The electric field instrument EFW has a malfunctioning probe 1 on spacecraft 1, 2 and 3. This has a minor effect on science data return but instrument can still return 2 orthogonal E components in the spin plane. On all four spacecraft, the energetic ion instruments have no operating head in the look direction of the spin plane (hence only partial pitch angle distributions are possible) and recently all heads on spacecraft 1 have ceased to function. The Electron drift instrument (EDI) does not function on Cluster 4. Most recently, ASPOC, the spacecraft potential control was turned off due to the depletion of all its indium reserves, thus making it more difficult to measure the low energy component of particle distributions.

Table 21.2 Cluster instruments and Principal investigators (December 2008) Instrument Principal investigator ASPOC (spacecraft potential control) CIS (ion composition, 0 < E < 40 keV) EDI (plasma drift velocity) FGM (magnetometer) PEACE (electrons, 0 < E < 30 keV) RAPID (high energy electrons and ions, 20 < Ee < 400 keV, 10 < Ei < 1,500 keV) DWP (wave processor) EFW (electric field and waves) STAFF (magnetic and electric fluctuations) WBD (electric field and wave forms) WHISPER (electric density and waves)

K. Torkar (IWF, A) I. Dandouras (CESR, F) R. Torbert (UNH, USA) E. Lucek (IC, UK) A. Fazakerley (MSSL, UK) P. Daly (MPAe, D) H. Alleyne (Sheffield, UK) M. Andr´e (IRFU, S) N. Cornilleau-Wehrlin (CETP, F) J. Pickett (University of Iowa, USA) J.-G. Trotignon (LPCE, F)

Instruments marked with a are part of the wave experiment consortium (WEC)

21

The Cluster Mission: Space Plasma in Three Dimensions

315

Access to the Cluster data was originally only available via the Cluster Science Data System, which distributed quicklook, summary and prime parameters data to all Cluster Principal and Co-Investigators. The Cluster Co-Is number around 300 plus the 11 Principal Investigators, located in more than 90 laboratories in 24 countries. Now high-resolution Cluster data are open to the worldwide science community via the Cluster Active Archive, which currently has more than 900 registered users (see Laakso et al. [Chapter 1 of this book]).

21.4 Publications Overall, the mission has been very productive and as of the end of December 2008 has generated 927 refereed publications (see Fig. 21.3), which includes publications from the Double Star mission. With 186 papers in 2008 alone, the number of scientific publications is still increasing. This recent increase is partly due to a new and wider user community gaining access to high quality Cluster data through the Cluster Active Archive (CAA) and partly to a special issue of the Journal of Geophysical Research. Of these papers, 30 have been published in high-impact journals not specific to the science of plasma physics: Nature, Nature Physics and Physical Review Letters. This shows that Cluster has not only revolutionized plasma physics, but also physics in general. About 60% of the total number of papers were published in the last 3 years, which shows that the community is still very active. Two Cluster scientists (Prof Balogh and Prof R`eme) are listed by Thomson Scientific as fifth and eighth most cited authors in geosciences over the last decade.

Fig. 21.3 Cluster and Double Star publications as of December 2008

316

M.G.G.T. Taylor et al.

Cluster has proven to be an excellent educational tool for training young scientists in the field of plasma physics and has resulted (as of December 2008) in 49 theses, of which 40 are Ph.D. Highlights of the latest Cluster results are published once a month on the ESA public and technical web sites. In the last 3 years, 37 web stories have been published on the ESA web site.

21.5 Recent Results Since the beginning of the current extension in February 2005, many new science results have been reported using Cluster data. These results were obtained as a result of Cluster’s unique four-point measurement capability, the strategy of varying inter-spacecraft distances and the high quality instrumentation, all of which have facilitated the sampling and derivation of the numerous plasma parameters necessary to characterize and understand fundamental plasma physics phenomena (Fig. 21.4).

Fig. 21.4 Cluster four-point measurements of magnetic field magnitude and density from C1 (black), C2 (red), C3 (green), C4 (blue). Dashed lines show low pass filtered data, solid lines the sum of low pass and band pass filtered data. The band pass covers 0.025–0.083 Hz. Detailed study shows that oscillations are not exactly in phase. In particular, the oscillations phases and amplitudes do not show a nested signature. This is most evident at t D 80 s. An immediate implication is that they cannot be explained by a static one dimensional shock profile oscillating back and forth, but instead must be propagating. This and the profile of the oscillation amplitude indicates that they could be described as a surface wave within the shock structure [17] (Reproduced by permission of American Geophysical Union)

21

The Cluster Mission: Space Plasma in Three Dimensions

317

21.5.1 First Observation of Surface Waves on the Bow Shock The Earth bow shock is the place where the solar wind is decelerated in front of the Earth magnetic field. The shock is usually observed as a sharp discontinuity in magnetic field, plasma density and velocity. New Cluster observations [17] showed coherent oscillations of the plasma density and the magnetic field amplitude around the shock crossing. Phase differences in the oscillations, as seen on the four Cluster spacecraft, indicated that the oscillations were propagating, and could not be explained by a one-dimensional shock profile fluctuating in position. Four-point timing shows that the oscillations correspond to ripples traveling across the surface of the shock, with wavelength 1,000–2,000 km and propagating roughly parallel to the magnetic field. The presence of these oscillations could have major implications for the analysis of crossings observed in the past if it turned out to be a common phenomenon.

21.5.2 Solitons Observed and Successfully Modeled at the Magnetopause Solitary waves or solitons, have application across various fields of physics, including optical fibres (e.g. to enable ultra-fast internet). Fundamental questions on this phenomenon, however, remain open. In an article published by Trines et al. [29] the spontaneous formation of zonal mode solitary waves at the magnetopause is reported and explained. Although different types of solitons have been observed in the magnetosphere, this paper reports, for the first time, a direct comparison of multispacecraft observations and matched numerical modeling of zonal flow solitons, unambiguously showing that these structures penetrate through the magnetospheric border.

21.5.3 First 3-D Analysis of the Turbulence in the Magnetosheath Fundamental 3-D properties of the magnetic turbulence in the magnetosheath have been revealed [25], using the four-point measurements of Cluster and a data processing technique called k-filtering. Characterizing the properties of the magnetic turbulence occurring in this region is of prime importance to understand its role in fundamental processes such as mass transport, energy dissipation and magnetic reconnection (a physical process which enables the incoming solar wind to flow through the magnetopause). The nature of the turbulence could be firmly demonstrated in terms of mirror-like wave structures for the first time. The respective role of the propagation velocity and Doppler effects could be separated because of the 4-point measurement capability of Cluster. A power-law k-spectrum was constructed that had a surprisingly steep index compared to those expected from theory. The other significant result concerned the strong anisotropy of the turbulence in which the spectra were nonlinearly widened in only one direction: parallel to the

318

M.G.G.T. Taylor et al.

magnetopause boundary and perpendicular to the magnetic field. New results, recently published in Physics Review E [26], go beyond the k-spectra and provide findings that had never been obtained before, neither in space nor in laboratory: the coherent nature of the structures composing the turbulence has been quantized and the existence of an inertial range, transferring energy from large to small scales, established.

21.5.4 Discoveries on Magnetic Reconnection Magnetic reconnection is a process by which magnetic energy is converted into particle energy and is accompanied by a topological change in the geometry of the magnetic field. The process allows for the transfer of mass and momentum between adjacent plasma regimes. This process is believed to be important in astrophysics (intergalactic medium, neutron stars), solar and space physics (solar flares, coronal mass ejections, geomagnetic substorms), and laboratory devices (e.g., tokamaks). Other than in the laboratory, planetary magnetospheres and the solar wind are the only places where we can study this process in-situ by measuring electric and magnetic fields and particle distribution functions. The advantage of space plasmas is that the measurement devices, the spacecraft, are much smaller than the plasma scales and therefore do not perturb the medium and the underlying physical process. In addition, a wide range of plasma parameters can be measured and there are no external walls that perturb the plasma as opposed to laboratory plasmas. We describe below a few highlights on reconnection studies.

21.5.5 Discovery of the Largest Reconnection Line (2.5 Millions of Kilometers Length) Using the Cluster spacecraft and the NASA Wind and ACE satellites, the largest jets of particles created between the Earth and the Sun by magnetic reconnection were discovered [20]. This result was published in Nature and the key illustration made it to the cover of the journal. The reported observations in the solar wind provide direct evidence that magnetic reconnection is fundamentally large scale and quasi-steady in nature. The 2.5 million km reconnection region is a world record, two orders of magnitude longer than the previous record.

21.5.6 First Observation of Magnetic Reconnection and Electric Current Disruption Close to Earth Using Cluster and Double Star Colourful aurorae, huge currents induced in power grids, rails and pipelines and perturbations of GPS signals are just some of the effects of a magneto-

21

The Cluster Mission: Space Plasma in Three Dimensions

319

spheric phenomenon known as a magnetic substorm. Up to now, the onset of this phenomenon has been explained by two competing models: current disruption and near-Earth reconnection. A study published in Geophysical Research Letters [28] challenges this binary vision of substorm onset. Thanks to simultaneous measurements performed by the four Cluster satellites and the Double star TC2 spacecraft, a third type of substorm onset is now suggested. Data collected on 26 September 2005 between 08 and 10 UT captured the occurrence of three consecutive substorm onsets. But for the first time, these data indicate that the current disruption process and magnetic reconnection can coincide in space and time showing possibly the two sides of the same process.

21.5.7 Discovery of Magnetic Reconnection in Turbulent Plasma Irregular behavior of particle flows and magnetic fields, known as plasma turbulence, occurs throughout the universe. In turbulent plasma many small-scale boundaries can form, where models predict reconnection to occur. However, magnetic reconnection in turbulent plasma has never been directly observed before. Using measurements of the four Cluster satellites, a study published in Nature Physics [23] shows, for the first time, experimental evidence of magnetic reconnection in turbulent plasma. Reconnection was found to occur within a very thin current sheet embedded in the turbulent plasma with a typical size of about 100 km, a real challenge for the resolution of the instrumentation on Cluster. The observations showed that the turbulent plasma was accelerated and heated at thin, reconnecting current sheets, as had been predicted by Matthaeus and Montgomery [15]. Moreover, this new type of small-scale reconnection seems to be associated with the acceleration of particles to energies much higher than their average energy, something that could explain, in part, the creation of high energy particles by the Sun. This discovery opens new perspectives to better understand the behaviour of turbulent plasmas in the universe.

21.5.8 First Observation of Magnetic Reconnection in Giant Vortices Cluster observations revealed the presence of magnetic reconnection within giant swirls of plasma of 40;000 km size located on the flank of the magnetosphere [19]. These giant swirls were identified to be the result of the Kelvin-Helmholtz instability (also known as ‘wind over water’ instability) and were shown to facilitate the transport of solar wind material into the magnetosphere. Believed to be two distinct physical processes, Kelvin-Helmholtz instabilities and magnetic reconnection have been shown to co-exist under certain conditions. This result improves our knowledge on how, where and under which conditions the solar wind is able to penetrate the Earth’s magnetic shield (Fig. 21.5).

320

M.G.G.T. Taylor et al. time = 91.32 Velocity + Density

Magnetic Field + jz

time = 91.32

20

1.84

2.22

= 0.74

x(–M)

= 1.65

1.06 10

10

1.70

0

0

1.18

–0.50

–10

–10

0.66

–1.29

–20 –9

0.28

–3

T Dens. (cm–3) N Magn.F. (nT) Temp. (MK) T

12 10

0.35 0.30 0.25 0.20 100 0 –100 –200 –300 15 10 5 0 –5 –10 –15 –20

Pressure (nPascal)

–1

(km s ) Vx Vy Magn. Field (nT) Bx By

–10 0 10 20 x Data represents a cut throuth system for const y. Location is in simulation units and velocities are in magnetospheric frame Cut at y = 2.0, phi = 0.0, time = 410

3

9

8

8 6 4 2 0 0.5

6 4 2 0

0.4 0.3 0.2 0.1 100

Ptot Velocity

–y (N)

Cluster 3rd of July 2001 -MFR SC1 N (cm3)

14 12 10 8 6 4 2 0

0.14

–9

T (MK)

9

P (nPa)

–3

v (km/s)

8 6 4 2 0 0.40

–y (N)

B (nT)

14 12 N 10

–20 –3

0 –100 –200 15 10 5 0 –5 –10 –15 05:47:00

05:48:00 05:49:00 Time (UT)

05:50:00

05:51:00

Fig. 21.5 MHD simulation result of the KH wave observed by Cluster on 3 July 2001. The magnetic field and jz are plotted on the top left panel and velocity and density at top right panel. The black lines are magnetic field lines. The wave is moving in the negative x-direction (down the page) and the higher density side indicates the magnetosheath side. Reconnection has generated a magnetic island moving downstream, and the upstream moving flux tube has both ends connected to the Earth’s magnetic field, so that these flux tubes only have magnetospheric plasma in them. The bottom left panel is a cut through the simulation at y D 2. The bottom right panel has Cluster sc1 and sc3 data between 05:46:30–05:51 UT in boundary normal coordinates of sc1. The intervals satisfying reconnection criteria are marked with vertical magenta columns. Although this is a two-dimensional simulation result, many features qualitatively agree with Cluster observations. Both plots show similar type of variation of sheath-like (high density, low temperature) and magnetospheric-like (low density high temperature) regions, and the location of the intervals satisfying the reconnection criteria are also in good agreement [19]

21

The Cluster Mission: Space Plasma in Three Dimensions

321

21.5.9 Discovery of Very Elongated Electron Diffusion Region At the heart of the reconnection process is a small core zone called the electron diffusion region, where extreme topologies of the magnetic field lead to a situation where electron are de-magnetised, which means that their motion is not controlled by the magnetic field anymore. In a paper published in Physics Review Letters [21], Cluster observations in the Earth’s magnetosheath show evidence for the overall size of this diffusion region and find that it is 300 times larger than previously thought. This was supported by particle simulations with large simulation boxes. This means that future missions will have a much better chance of detecting and resolving this region than previously estimated, profoundly impacting mission design and scientific operations.

21.5.10 First Observation of 3D Magnetic Reconnection and Magnetic Nulls By reconnecting with each other, two adjacent regimes of magnetic fields create a magnetic null at the centre of the reconnection region. This is the classical twodimensional picture of reconnection. In three dimensions however, theory predicts not one but two nulls that would be linked to each other. A paper published in Nature Physics reports the first observation of two linked magnetic nulls using data collected by the four Cluster spacecraft [31]. In addition, the rare occurrence of one of the Cluster satellites passing by one null of the pair has uncovered a new phenomenon where the electrons are temporarily trapped around that null [8]. The electron trapping may ultimately lead to the formation of energetic electron beams, a well-known but poorly understood consequence of reconnection. This pioneering discovery will help constrain theoretical models of magnetic reconnection.

21.5.11 Confirmation of Relation Between Magnetic Substorms and Plasma Bursty Bulk Flows High-speed flows of plasma, known as bursty bulk flows (BBF), are propagating in the Earth’s magnetotail at velocities higher than 300 km/s. They carry decisive amounts of mass, energy and magnetic flux towards the Earth but their link to magnetic substorms was never absolutely established. Based on data recorded by three Cluster spacecraft (those with fully functioning ion detectors), a statistical study reveals, for the first time, that more than 95% of magnetic substorms observed in this time period were accompanied by BBF [2]. It also shows that BBF lasts longer than previously estimated and may account for up to 20% of the energy transport of a substorm. Published in the Journal of Geophysical Research, these results enhance previous studies based on single spacecraft measurements (Fig. 21.6).

322

M.G.G.T. Taylor et al.

Fig. 21.6 Magnetic field configuration of a magnetic null pair reconstructed from Cluster measurements when C2 crosses the neutral sheet. (a) Separator reconnection configuration reconstructed from observations by Cluster at 09:48:25.637. A-null and B-null are shown as orange and red balls, respectively, with the separator as a bold pink curve connecting them. Magnetic field lines, colored according to local field strengths, converge along the fan surface (in white) to approach the A-null and then travel out along the spine (marked in black arrows) of the A-null. Magnetic field lines converge along the spine of the B-null and then diverge out along its fan surface (in grey). Cluster C1, C2, C3, C4 are drawn as four cubes in black, red, green, and blue, respectively. The origin of the illustrating coordinates is at the C2 position. (b) Magnetic field strength distributed on the X-line. The total field strength is drawn in bold black. Three components, Bx, By, and Bz, are in red, green and blue, respectively. The x-coordinate is the distance along X-line starting at the beginning with the smallest y in Fig. 21.6a. The orange (red) circle marks the location of the A-null (B-null) [8] (reproduced by permission of American Geophysical Union)

21

The Cluster Mission: Space Plasma in Three Dimensions

323

21.5.12 Strongest Current Measured So Far in the Magnetotail To sustain its elongated shape, the magnetotail has in its centre part a current sheet, also called neutral sheet. Under normal conditions, the neutral sheet is usually wide (a few earth radii) and the current density is around 4 nA m2 . Before the Cluster era the actual values of local current densities in most regions of space have been uncertain due to the difficulty of measuring electric current densities in situ. During magnetic substorms, the neutral sheet becomes very thin. In such cases, the density of electric current increases steeply and subsequently magnetic reconnection can be triggered in the resulting extreme magnetic configurations. Cluster data at smallest separation distances in the magnetotail, around 200 km, were used to quantify for the first time the thickness of the current sheet; it was found to be comparable to or less than one ion inertia length or a few 100s km [18]. In addition using the curlometer technique or applying Ampere’s law to the multipoint magnetic data, the current measured was extremely strong, above 180 nA m2 , a factor 40 more than during quiet conditions. This finding will help to improve theoretical models of neutral sheet dynamics and understand how, when, where and why reconnection is initiated in the magnetotail.

21.5.13 First Measurement of the Size of Ion Diffusion Region Using Cluster Multi-scale Configuration Another thin magnetotail current sheet was observed in August 2005 during a magnetic substorm [24]. During that event, the spacecraft were in a multiscale configuration with C1, C2 and C3 forming a large triangle of 9,000 km parallel to the neutral sheet and C4 above C3 at about 900 km. distance. The analysis of the magnetic, electric and plasma data in the thin current sheet showed plasma flow reversal demonstrating that reconnection was taking place locally and that the spacecraft crossed the ion diffusion region. Using data from the closely spaced C3 and C4 pair of spacecraft, the vertical scale of the ion diffusion region was estimated to be less than 900 km. This is less than previous estimations. This preliminary analysis would however need more analysis and support for simulations since not all observations in this case supported the classical picture of reconnection (Fig. 21.7).

21.5.14 First Measurement of Second Order Terms in the Generalized Ohm’s Law In a plasma the electric field is fundamentally linked to the magnetic field, such that the two fields are perpendicular to each other for most of the time. However, at the heart of the reconnection process, the electric field has a component parallel

324

M.G.G.T. Taylor et al. 1.0

0.0 0.5 1.0 1.0 0.5 0.0 –0.5–1.0 dX’, RE 20

1.0

b

0.5

dZ’, RE

a

–0.5

dZ’, RE

dY’, RE

–1.0

0.0 -0.5 -1.0 1.0 0.5 0.0 –0.5–1.0 dX’ ,RE

c

0.5 0.0 -0.5 -1.0 1.0 0.5 0.0 –0.5–1.0 dY’ ,RE

d CLUSTER 1234

0 Vy,km/s Vx,km/s j ,nA/m2 Bz’, nT By’, nT Bx’, nT

–20 10

e

0

–10 10

f

0

–10 20 10 0 1000 500 0 –500 –1000 500 0 –500

g

h

i

2300:00

2312:00

2324:00

2336:00

2348:00

UT

Fig. 21.7 Thin current sheet measurement by Cluster. Panels (a–c) show the Cluster spacecraft configuration around 23:00 UT on 28 August 2005. Panels d–f show the magnetic field measurements, followed by current density (panel g) and the X and Y components of the ion bulk velocity from C1 and C3 (panels h and i). Between 2,300 and 2,312 UT the magnetic field at Cluster was oscillating with the four traces close together. After 2,312 UT, during the substorm growth phase, Cluster/FGM started to detect an enhancement of the magnetic field gradient: the differences between the C2, C3/C4, and C1 Bx traces (Fig. 21.7d) became significant. The difference between Bx at C3 and C4 is used to estimate the current density in the plasma sheet (Fig. 21.7g). It increased drastically at 2,337 UT and stayed between 20 and 30 nA=m2 until 2,342 UT. During this interval, C3 detected Bx fluctuating at about 15 nT, while Bx at C4 is of 15–20 nT. Thus, Cluster encountered a thin current sheet with a thickness of about 900 km (distance between C3 and C4). A flow reversal from Tailward to Earthward is seen around 2,341 UT, which implies that the spacecraft were close to the reconnection region [24] (Reproduced by permission of American Geophysical Union)

21

The Cluster Mission: Space Plasma in Three Dimensions

325

to the magnetic field, which controls in particular the acceleration of particles. The perpendicular E-field component on the other hand is important, as it causes plasma and magnetic field inflow towards the X-line, necessary for reconnection to occur. This electric field component is known as the “reconnection electric field”. The total electric field is described by the generalised Ohm’s law, were E is the sum of four terms: MHD term, Hall term, pressure tensor term and electron inertia term. The first two terms have already been measured in space, while the last two never before. For the first time the third term, which can be understood as the divergence of the electron pressure tensor, was estimated using the electron detectors on board Cluster [9]. In agreement with numerical simulations, it was shown that the electric field from the pressure tensor was opposite to the Hall electric field and could support the reconnection electric field. Such studies to measure micro-physics quantities were only possible with the four Cluster at small separation distances from each other, around 200 km.

21.5.15 First Observation of “Chorus” Emission Propagation to Low Altitude Natural emissions of whistler-mode chorus consist of electromagnetic waves in the frequency range from a few hundreds of Hz to several kHz. Chorus emissions have been known for several decades but their source mechanism and effects are not yet well understood. Understanding the mechanisms and effects of chorus is important since it is believed that these intense plasma waves may strongly energize other radiation belt electrons to MeV energies. For the first time, Cluster has enabled scientists to determine the location and size of chorus source regions to be determined, together with propagation characteristics from these sources. Ray-tracing studies based on this new knowledge have then revealed that chorus waves can propagate to lower altitudes without being damped and provide a possible embryonic source of plasmaspheric hiss [3]. These results have been further improved in recent studies combining Cluster with the ionospheric DEMETER spacecraft [27].

21.5.16 Discovery of Magnetosonic Waves Accelerating Electrons to MeV Energies in Radiation Belts The relativistic electron flux in Earth’s outer radiation belt varies by over five orders of magnitude and is responsible for certain types of satellite malfunctions. Flux variations are due to a combination of electron acceleration, transport, and loss processes acting within the inner magnetosphere, indicating that the magnetosphere is a gigantic particle accelerator. For many years inward radial diffusion from a source beyond geosynchronous orbit (L D 6:6) has been considered the dominant

326

M.G.G.T. Taylor et al.

Fig. 21.8 Fast magnetosonic waves detected by the STAFF instrument on Cluster 3, as it crossed the magnetic equator on 25 Nov 2002. The event occurred during an unusually long storm recovery phase. The waves can be identified by the intense emissions between 20 and 60 Hz, below the lower hybrid resonance frequency fLHR (solid line) but above the local proton gyrofrequency fcH , (5:1 Hz at the equator). The waves are confined to C= 3ı of the magnetic equator and are far more intense than chorus for this event (the band of emissions above fLHR rising in frequency towards the equator). These waves are generated by ion ring distributions and accelerate electrons, and it is suggested that they contributed t the elevated electron flux observed during this event, and could be an important energy transfer process from the ring current to the Van Allen radiation belts [10] (Reproduced by permission of American Geophysical Union)

mechanism for transport and acceleration but recent observations show that the electron phase space density can peak near L D 4.5. Acceleration by whistler mode chorus waves is an important mechanism for particles in the outer radiation belt during magnetic storms. However, Cluster observations have shown that fast magnetosonic waves can also accelerate electrons between 10 keV and a few MeV and must be considered as an additional local acceleration mechanism [10] (Fig. 21.8).

21.5.17 First Observation of Narrow Angle Emission of the Auroral Kilometric Radiation A recent study reveals how the most powerful emission of terrestrial origin, the Auroral Kilometric Radiation (AKR), is beamed from the auroral zone into space. Seen from deep space, the Earth is a powerful planetary radio source, comparable to

21

The Cluster Mission: Space Plasma in Three Dimensions

327

Jupiter, with maximum output power in the 50–500 kHz range. AKRs are intimately linked with auroras, or more precisely with the acceleration of the energetic electron beams that cause the auroras. The new result clarifies how the radiation is beamed [16], which is in a narrow plane tangent to the magnetic field at the source, rather than a hollow cone as previously suggested. Cluster multi-point measurements could demonstrate that individual AKR bursts do not radiate in the manner described by two models that were proposed some 30 years ago. However, these new data do back up the numerical simulations of Louarn and Le Qu´eau [13] and Pritchett et al. [22] which predicted longitudinal propagation.

21.5.18 First Observation of the Evolution of Electric Circuit Powering the Auroras Aurora are powered by giant electrical circuits that accelerate electrons which then hit the upper layers of the atmosphere and form arcs in high-latitude regions like Scandinavia. It is known that two different types of electrical circuit, each associated with its own type of electric potential structure (“U” or “S” shape), close at low altitude along these arcs, but little was known about the source of this difference. For the first time, two consecutive crossings by two Cluster spacecraft captured the reconfiguration from one type of current system to the other [14]. It was shown that the particular evolution of such current systems may have its origin in the electron population in the magnetotail.

21.6 Conclusions Eight years of Cluster operations have provided a wealth of unique and ground breaking science results, in particular illuminating the three dimensional aspect of space plasma phenomena, some of which were highlighted above. A further extension of Cluster will provide even more exciting new science opportunities. Specifically, the extension proposal core science objectives are: 1. To discover the triggering and time evolution of the auroral acceleration by making the first four spacecraft measurements in the auroral acceleration region 2. To discover the role of current disruption in magnetic substorms, still a controversial problem, by making the first four spacecraft measurements in the near-earth neutral sheet (at 8–10 Earth radii radial distance) and to discover the role of electric fields in tail dynamics by repeating the tilting manoeuvre in the magnetotail 3. To study the relationships between the characteristics of cold plasma and wave activity in the plasmasphere and high-energy particles in the radiation belts using low-perigee data

328

M.G.G.T. Taylor et al.

4. To study the magnetosphere simultaneously at global and ionscales by enhancing common operations with the THEMIS mission 5. To understand the structuring of currents and energy transport in space plasmas between ionosphere and magnetosphere by collaborating with the Earth observation Swarm multi-spacecraft mission in the ionosphere 6. To turn Cluster into an observatory, in order to exploit any yet unidentified new and exciting ideas from a wider user community to utilize this unique set of spacecraft While the first three topics basically are facilitated by the favourable development of the orbit (apogee in the South hemisphere and decreasing altitude at perigee), topics 4 and 5 represent new opportunities arising from multi-spacecraft collaborations in the wider magnetosphere, both inside and outside the Cluster orbit (THEMIS), and down to the ionosphere (Swarm). Topic 6 will open the Cluster science to a wider community. Cluster, by utilizing multi-spacecraft measurements, has provided a quantum leap in space plasma physics investigation. Recently, the five spacecraft THEMIS mission has further demonstrated the utility of such measurements at large separations. Future missions have multiple spacecraft measurements at the heart of their concept. The NASA Magnetospheric Multi-Scale mission, due to launch in 2014, will utilize 4 spacecraft to investigate the phenomena of reconnection at smaller scales than Cluster (down to electron scales), using highly specialized fast plasma detectors to unravel the mysteries of the reconnection diffusion region. Recently ESA made a call to the community for proposals to Cosmic Vision 2015–2025. The Cross-Scale proposal was selected for further study. The mission concept is to quantify the coupling in plasmas between the three different physical scales (electron ion and MHD) in order to address fundamental questions such as: how shocks accelerate and heat particles; how reconnection converts magnetic energy and how turbulence transports energy from source to dissipation. This mission will be made up of up to 12 spacecraft comprising three nested tetrahedra. In collaboration with this activity, JAXA are investigating the multispacecraft mission SCOPE, which consists of a large mother spacecraft and four smaller daughter spacecraft. The future of space plasma physics lies truely in multi-point multi-scale observations, and this development is in no small part due to the Cluster mission. We hope Cluster can continue to remain an important part of space plasma science for a few decades even beyond the anticipated mission extension thanks to the datasets available to the worldwide science community through the Cluster Active Archive. The latest news relating to the Cluster mission can be found at: http://sci.esa.int/ cluster/. Information on the Double Star mission can be found at http://sci.esa.int/ doublestar/. Acknowledgements The authors would like to thank all PIs and their teams for their hard work on the Cluster instruments and data, and the JSOC and ESOC teams for their very efficient operation of the Cluster spacecraft.

21

The Cluster Mission: Space Plasma in Three Dimensions

329

References 1. Angelopoulos, V.: The THEMIS Mission, Space Sci. Rev., doi: 10.1007/s11214–008–9336–1, (2008) 2. Cao, J.B., Ma, Y.D., Parks, G., R`eme, H., Dandouras, I., Nakamura, R., Zhang, T.L., Zong, Q., Lucek, E., Carr, C.M., Liu, Z.X., Zhou, G.C.: Joint observations by Cluster satellites of bursty flows in the magnetotail, J. Geophys. Res., Vol. 111, A04206, 10.1029/2005JA011322, (2006) 3. Chum, J. and Santolik, O.: Propagation of whistler-mode chorus to low altitudes: divergent ray trajectories and ground accessibility, Ann. Geophys., pp. 3727–3738. SRef-ID: 1432– 0576/ag/2005–23–3727, (2005) 4. Dunlop, M.W., Escoubet, C.P., Liu, Z.-X., Shen, C., Laakso, H., Taylor, M.G.G.T., Fazakerley, A.N., and the Double Star PIs: Double Star: Mission, instruments and joint observations (Dayside Boundary phenomena), this issue, (2010) 5. Escoubet, C.P., Fehringer, M., Goldstein, M.: The Cluster mission, Ann. Geophys., 19, 1197, (2001) 6. Friis-Christensen, E., L¨uhr, H., Knudsen, D., Haagmans, R.: Swarm – An Earth Observation Mission investigating Geospace, Adv. Space Res., 41 (1), 210–216, (2008) 7. Haerendel, G., Roux, A., Blanc, M., Paschmann, G., Bryant, D., Korth, A., Hultqvist, B.: Study in three dimensions of plasma turbulence and small-scale structure in the magnetosphere mission proposal submitted to ESA, (1982) 8. He, J.-S., Zong, Q.-G., Deng, X.-H., Tu, C.-Y., Xiao, C.-J., Wang, X.-G., Ma, Z.-W., Pu, Z.-Y., Lucek, E., Pedersen, A., Fazakerley, A., Cornilleau-Wehrlin, N., Dunlop, M.W., Tian, H., Yao, S., Tan, B., Fu, S.-Y., Glassmeier, K.-H., R`eme, H., Dandouras, I., Escoubet, C.P.: Electron trapping around a magnetic null, Geophysical Research Letters, 35, L14104, http://dx.doi.org/10.1029/2008GL034085, (2008) 9. Henderson, P.D., Owen, C.J., Lahiff, A.D., Alexeev, I.V., Fazakerley, A.N., Lucek, E., R`eme, H.: Cluster PEACE observations of electron pressure tensor divergence in the magnetotail, Geophys. Res. Lett., Vol. 33, No. 22, L22106, http://dx.doi.org/10.1029/2006GL027868, (2006) 10. Horne, R.B., Thorne, R.M., Glauert, S.A., Meredith, N.P., Pokhotelov, D., Santol´ık, O.: Electron acceleration in the Van Allen radiation belts by fast magnetosonic waves, Geophys. Res. Lett., 34, L17107, http://dx.doi.org/10.1029/2007GL030267, (2007) 11. Laakso, H.C. Perry, S. McCaffrey, D. Herment, A.J. Allen, C.C. Harvey, C.P. Escoubet, C. Gruenberger, M.G.G.T. Taylor, and R. Turner, Cluster Active Archive: overview, this issue, (2010) 12. Liu, Z.X., Escoubet, C.P., Pu, Z., Laakso, H., Shi, J.K., Shen, C., Hapgood, M.: The Double Star Mission, Ann. Geophys., Page(s) 2707–2712. SRef-ID: 1432–0576/ag/2005–23–2707, (2005) 13. Louarn, P., Le Qu´eau, D.: Generation of the auroral kilometric radiation in plasma cavities-II. The cyclotron maser instability in small size sources, Planet. Space Sci., 44, 211–224, (1996) 14. Marklund, G.; Johansson, T.; Lileo, S.; Karlsson, T.: Cluster observations of an auroral potential and associated field-aligned current reconfiguration during thinning of the plasma sheet boundary layer, J. Geophys. Res., Vol. 112, No. A1, A01208, http://dx.doi.org/10.1029/2006JA011804, (2007) 15. Matthaeus, W.H., Montgomery, D.C.: Selective decay hypothesis at high mechanical and magnetic Reynolds numbers, in Nonlinear Dynamics, ed. Hellman, Hobert H. G., Proc. N. Y. Acad. Sci., Vol.357, pp. 203, (1980) 16. Mutel, R.L., Christopher, I.W., Pickett, J.S.: Cluster multispacecraft determination of AKR angular beaming, Geophys. Res. Lett., 35, L07104, (2008) 17. Moullard, O., Burgess, D., Horbury, T.S., Lucek, E.A.: Ripples observed on the surface of the Earth’s quasi-perpendicular bow shock, J. Geophys. Res., Vol. 111, No. A9, A09113, http://dx.doi.org/10.1029/2005JA011594, (2006) 18. Nakamura, R., Baumjohann, W., Fujimoto, M., Asano, Y., Runov, A., Owen, C.J., Fazakerley, A.N., Klecker, B., R`eme, H., Lucek, E.A., Andr´e, M., Khotyaintsev, Y.: Cluster observations

330

19.

20.

21.

22.

23.

24.

25.

26. 27.

28.

29.

30.

31.

M.G.G.T. Taylor et al. of an ion-scale current sheet in the magnetotail under the presence of a guide field, J. Geophys. Res., 113, A07S16, http://dx.doi.org/10.1029/2007JA012760, (2008) Nykyri, K., Otto, A., Lavraud, B., Mouikis, C., Kistler, L.M., Balogh, A., R`eme, H.: Cluster Observations of reconnection due to the Kelvin-Helmholtz instability at the dawnside magnetospheric flank, Annales Geophysicae, 2619–2643, Vol. 24, No. 10, (2006) Phan, T.D., Gosling, J.T., Davies, M.S., Skoug, R.M., Øieroset, M., Lepping, R.P., McComas, D.J., Smith, C.W., R`eme, H., Balogh, A.: A Magnetic reconnection X-line extending more than 390 Earth Radii in the solar wind, Nature, Vol. 439, (2006) Phan, T.D., Drake, J.F., Shay, M.A., Mozer, F.S., Eastwood, J.P.: Evidence for an Elongated (>60 Ion Skin Depths) Electron Diffusion Region during Fast Magnetic Reconnection, Phys. Rev. Lett., 99, 255002, (2007) Pritchett, P.L., Strangeway, R.J., Ergun, R.E., Carlson, C.W.: Generation and propagation of cyclotron maser emissions in the finite auroral kilometric radiation source cavity, J. Geophys. Res., 107(A12), 1437, doi:10.1029/2002JA009403, (2002) Retin`o, A., Sundkvist, D., Vaivads, A., Mozer, F., Andre, M., Owen, C.J.: In-situ evidence of magnetic reconnection in turbulent plasma, Nature Physics, March 25, doi:10.1038/nphys574, (2007) Runov, A., Baumjohann, W., Nakamura, R., Sergeev, V.A., Amm, O., Frey, H., Alexeev, I., Fazakerley, A.N., Owen, C.J., Lucek, E.A., Andr´e, M., Vaivads, A., Dandouras, I., Klecker, B.: Observations of an active thin current sheet, J. Geophys. Res., 113, A07S27, doi:10.1029/2007JA012685, (2008) Sahraoui, F.; Belmont, G., Rezeau, L., Cornilleau-Wehrlin, N., Pinc¸on, J.-L., Balogh, A.: Anisotropic turbulent spectra in the terrestrial magnetosheath: Cluster observations, Physical Review Letters, 96, 075002, (2006) Sahraoui, F.: Diagnosis of magnetic structures and intermittency in space-plasma turbulence using the technique of surrogate data Phys. Rev. E, 78, 026402, (2008) Santolik, O., Chum, J., Parrot, M., Gurnett, D.A., Pickett, J.S., Cornilleau–Wehrlin, N.: Propagation of whistler mode chorus to low altitudes: Spacecraft observations of structured ELF hiss, J. Geophys. Res., 111, A10208, doi:10.1029/2005JA011462, (2006) Sergeev, V., Semenov, V., Kubyshkina, M., Ivanova, V., Baumjohann, W., Nakamura, R., Penz, T., Runov, A., Zhang, T.L., Glassmeier, K.-H., Angelopoulos, V., Frey, H., Sauvaud, J.-A., Daly, P., Cao, J.B., Singer, H., Lucek, E.: Observation of repeated intense near-Earth reconnection on closed field lines with Cluster, Double Star, and other spacecraft, Geophys. Res. Lett., Vol. 34, No. 2, (2007) Trines, R., Bingham, R., Dunlop, M.W., Vaivads, A., Davies, J.A., Mendonc¸a, J.T., Silva, L.O., Shukla, P.K.: Spontaneous Generation of Self-Organized Solitary Wave Structures at Earth’s Magnetopause, Phys. Rev. Lett., 99, 205006, (2007) Volpp, J., Escoubet, C.P., Foley, S., Godfrey, J., Hapgood, M., Pallaschke, S.: Cluster constellation change manoeuvres – Management and Operations, SpaceOps 2004, Montreal, Canada, (2004) Xiao, C.J., Wang, X.G., Pu, Z.Y., Ma, Z.W., Zhao, H., Zhou, G.P., Wang, J.X., Kivelson, M.G., Fu, S.Y., Liu, Z.X., Zong, Q.G., Dunlop, M.W., Glassmeier, K.-H., Lucek, E., R`eme, H., Dandouras, I., Escoubet, C.P.: Satellite observations of separator line geometry of threedimensional magnetic reconnection, Nature Physics, 3, 609–613, doi:10.1038/nphys650, (2007)

Chapter 22

Double Star: Mission, Instruments and Joint Observations M.W. Dunlop, C.P. Escoubet, Z.-X. Liu, C. Shen, H. Laakso, M.G.G.T. Taylor, A.N. Fazakerley, and the Double star PIs

Abstract The Chinese National Space Agency (CNSA) Double Star (DSP) spacecraft, TC-1 and TC-2 were launched in December 2003 and July 2004 into near equatorial and polar orbits respectively. During more than 3 years of operations they have maintained a close phasing with the ESA four-spacecraft mission to produce the first, well coordinated multi-scale measurements, sampling phenomena with five and six spacecraft. In this short paper we give a brief review of the DSP mission and show its joint capability with Cluster by showing examples of use of some early and more recent analysis techniques and their application to (more than) four spacecraft. We highlight a selection of some co-ordinated events, focussing on dayside phenomena, but also with a brief discussion of a tail event. Other reviews in this special issue will deal more completely with coverage of the other regions of the magnetosphere.

22.1 Introduction The two Double Star spacecraft were launched in order to provide the opportunity to monitor events at distinct locations in the Earth’s magnetosphere, when coordinated with the quartet of Cluster spacecraft. The ESA Cluster mission [4]

M.W. Dunlop () SSTD, Rutherford Appleton Laboratory, Chilton, Didcot, Oxfordshire, OX11 0QX, UK and The Blackett Laboratory, Imperial College London, London, SW7 2AZ, UK e-mail: [email protected] C.P. Escoubet, H. Laakso, and M.G.G.T. Taylor ESA/ESTEC, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands Z.-X. Liu and C. Shen Centre for Space Science and Applied Research, Chinese Academy of Sciences, Beijing 100080, China A.N. Fazakerley Mullard Space Science Laboratory, University College London, Dorking, Surrey, RH5 6NT, UK

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 22, c Springer Science+Business Media B.V. 2010

331

332

M.W. Dunlop et al.

provided extremely detailed multi-point measurements sampling local in situ phenomena, while coordination with the Chinese Double Star mission [6] provided distant, simultaneous monitoring at other locations, often providing multi-scale information. At the magnetopause, for example, such coverage over a wide range of different magnetopause sites, combining up to six spacecraft from the two missions, has provided simultaneous coverage of the high and low latitudes, cusp and low latitude boundary layer. Recent studies have investigated X-line formation (directly and indirectly) and the development of the boundary layer at both high and low latitudes. For example, Pu et al. (2007, [8]) surveyed a number of direct reconnection events under both dominant BY and for low clock angles, finding evidence of predominantly component driven low latitude merging in conjunction with predominantly anti-parallel high latitude sites. The identification of oppositely moving FTEs at the Double Star TC-1 and Cluster spacecraft and detailed tracking of flux tube motion between TC-1 and Cluster along the MP was made by Wild et al. [12], Dunlop et al. [3], Fear et al. [5], and Wang et al. [11], and has been recently compared to MHD simulation results by Berchem et al. [2008]. Comparison of the cusp electron boundary layer and the low latitude boundary layer (LLBL) under northward directed interplanetary field has been investigated recently by Bogdanova et al. [1]. 1. The Double Star Project (DSP) consists of two magnetospheric spacecraft, built and operated by the Chinese National Space Agency (CNSA), following a mission programme which was first accepted by Chinese government on 28 September 2000. An agreement between ESA and CNSA was later signed in July 2001. The DSP spacecraft were launched from Long March-2C/SD vehicles, having an upgrade of the third stage to a more powerful version (responsible for the larger TC-1 orbit). The launches occurred at the Xichang (Equatorial, TC-1) and Taiyuan (polar, TC-2) sites. One equatorial (28ı ) satellite, TC-1, was launched on 29 Dec 2003 into a 1:09 13:4 RE inclined orbit, while the other, polar satellite, TC-2, was launched on 25 July 2004 into a 1:17:1 RE polar orbit. In fact, the apogee achieved by TC-1 was slightly higher than initially planned. Moreover, the spacecraft precesses semi-inertially so that it is initially in phase with the Cluster orbit. TC-2 has an increasing argument of perigee so that its line of apsides fell during the mission lifetime. The TC-1 spacecraft underwent a natural re-entry to Earth on 11 October 2007 (a year and a half beyond its nominal lifetime), while TC-2 is still in operation at the time of writing after an initial loss of contact in August 2007 and recovery in November 2007. The planned scientific objectives for the nominal mission, having a 2 year lifetime, were: to study magnetic reconnection at the magnetopause and in the magnetotail; to understand and locate the trigger mechanism for magnetospheric storms and substorms; to study physical processes like particle acceleration, diffusion, injection and up-flowing ions during storms, and to study temporal variations of field aligned currents and the coupling between tail current and auroral current. Figure 22.1 shows the scientific regions visited by Cluster and Double Star for both the extended spring and summer phases: dayside to nightside. The inclination of TC-1 placed its apogee initially low on the dayside and high on the nightside as shown, and was initially in phase with the Cluster orbital plane. The local time separation between

22

Double Star: Mission, Instruments and Joint Observations

333

Fig. 22.1 Dayside (left panel ) and nightside (right panel ) views of the combined Cluster and DSP orbit configuration

Cluster and TC-1 increased progressively, while TC-2 had an initially high apogee elevation, which progressively fell from that indicated. The views in Fig. 22.1 represent the orbital orientations in February and August of each year, where it is clear that on the nightside all spacecraft sample a range of distances down tail and sample field lines directly or indirectly connecting to both expected current disruption and near Earth reconnection sites. On the dayside, Cluster samples the mid to high latitude magnetopause while TC-1 samples the low to mid latitude magnetopause. Cluster may also be located in the solar wind, bow shock or magnetosheath at low to mid latitudes while TC-1 lies nearer the Earth at the magnetopause or magnetosheath. Cluster and TC-2 may also sample the Cusp simultaneously at different altitudes.

22.2 Operations, Payload and Data Acquisition The European Space Agency gave support to DSP via the bi-lateral agreement with CNSA in 2001 and involved: refurbishment/rebuilding of the Cluster spare instruments; integration of the European instruments in Europe; carrying out the tests for conductivity and magnetic cleanliness; performing the required radiation analysis; provision of the VILSPA-2 ground station (4 h/day) in addition to the 2 Chinese Ground stations, Beijing: ¿ 11 m dish, Shanghai: ¿ 25 m dish, which conducted shared operation and support of science operations for European instruments. The full payload and PI teams are listed in Table 22.1, which summarises the key technical information. Full descriptions of all instruments can be found in the special issue of Annales Geophysicae (2005), on The Double Star Mission, [Ann. Geophys., 23]). Table 22.1 shows the name of each experiment and its mass, power and data telemetry rate and instrument operating ranges. Identical Chinese lead high energy and heavy ion instruments (CSSAR) were accommodated on both TC-1 and TC-2. In addition, similar magnetic field (FGM) and thermal electron (PEACE), UK led instruments were accommodated on both TC-1 and TC-2, together with the French thermal ion (HIA) and wave (STAFF/DWP) experiments on TC-1, developed from flight spare Cluster instruments. The PEACE DSP instrument used a low energy

334

M.W. Dunlop et al.

Table 22.1 Payload distribution for TC-1 and TC-2, with European instruments in red [6] Equatorial double star (TC-1) Polar double star (TC-2) Instruments Active spacecraft Potential control (ASPOC) indium ions. 50 A max current. (2.54 kg, 2.9 W, 108 bps)

PI K. Torkar, IWF, Graz, Austria

Instruments Neutral Atom Imager (NUADU) 45 KeV < Eena < 300 KeV (5.7 kg, 4.6 W, 4,224 bps)

Fluxgate Magnetometer (FGM), 22 vector/s (3.1 kg, 3.6 W, 1,211 bps)

C. Carr IC, UK

Plasma electron and Current exp. (PEACE)# 1eV < Ee < 25 KeV (6.5 kg, 3.8–4.8 W, 4,624 bps) Hot Ion Analyzer(HIA), CIS sensor 2) 5eV < Ee < 32 KeV (3.5 kg, 2.8 W, 4,440 bps)

A. Fazakerley, MSSL, Dorking, UK

Part of Spatio-Temporal Analysis of Field Fluct.(STAFF) 0 < f < 4 kHzC Digital wave processor (4.9 kg, 4.0 W, 3509 bps) High Energy Electron Detector .HEED/ 0:2 < Ee < 10 MeV (2.2 kg, 2.0Watts, 224bps) High Energy Proton Detector .HEPD/ 3 < Ep < 400 MeV (2.2 kg, 2.0 W, 202 bps)

N. Cornilleau/H. Alleyne, CETP, Velizy, France and Sheffield U. UK

Fluxgate Magnetometer (FGM) 22 vector/s (3.1 kg, 3.6 W, 1,211 bps) Plasma Electron and Current Exp. (PEACE) 1 eV < Ee < 25 KeV (6.5 kg, 3.8–4.8 W, 4,624 bps) Low Energy Ion Detector .LEID/ 50 eV < Ee < 25 KeV (4.0 kg, 4.0 W, 2,000 bps) Low Frequency Electromagnetic Wave detector (LFEW) 8 Hz < f < 10 kHz (4.0 kg, 5.0 W, 3,000 bps) High Energy Electron Detector .HEED/ 0:2 < Ee < 10 MeV (2.2 kg, 2.0 W, 224 bps) High Energy Proton Detector .HEPD/ 3 < Ep < 400 MeV (2.2 kg, 2.0 W, 202 bps) Heavy ion detector .HID/10 MeV < Ei < 8 GeV (2.2 kg, 2.0 W, 45 bps)

Heavy ion detector .HID/ 10 MeV < Ei < 8 GeV (2.2 kg, 2.0 W, 45 bps)

H. R`eme, CESR, Toulouse, France

W. Zhang and J.B. Cao, CSSAR, China

J. Liang and J.B. Cao, CSSAR, China

Y. Zhai and J.B. Cao, CSSAR, China

Instrument built by China # PEACE includes only one sensor on each spacecraft

PI S. McKennaLawlor, Ireland U., Ireland L. Lu, CSSAR, China S. Barabash, IRF, Sweden T. Zhang, IWF, Austria

A. Fazakerley, MSSL, Dorking, UK

Q. Ren and J.B. Cao, CSSAR, China

Z. Wang and J.B. Cao, CSSAR, China

W. Zhang and J.B. Cao, CSSAR, China

J. Liang and J.B. Cao, CSSAR, China

Y. Zhai and J.B. Cao, CSSAR, China

22

Double Star: Mission, Instruments and Joint Observations

335

Fig. 22.2 The stowed and flight configuration of the DSP spacecraft

sensor for TC-1 (LEEA) and a high energy sensor for TC-2 (HEEA) and the HIA represented part of the full CIS experiment on Cluster. Thus, many properties of the plasma (distributions of the ions and electrons, and the magnetic field) were measured by almost identical instruments on all six spacecraft. The instrument data set has been re-calibrated on ground, and is available through the PIs and the DSP data centre. The in-flight spacecraft configuration, shown together with the launch configuration in Fig. 22.2, resulted in all instruments operating nominally. The spacecraft had a mass of 280 kg, cylindrical in shape with ¿ 2:1 m diameter and a height of 1.2 m. Each spacecraft was established in orbit with a nominal spin rate of 15 rpm (identical to Cluster) and a spin axis perpendicular to ecliptic. Spacecraft power was achieved via a solar panel array, with 6:33 m2 , 280 W (BOL) and Nickel-Cadmium batteries. The nominal, planned operational lifetime was 18 months (TC-1) and 12 months (TC-2). A number of non critical operational issues ensued after each launch. The solar panels on TC-1 generated an anomalous spacecraft current, introducing a complex, rotating (sun fixed), dipole spacecraft field, which affected the FGM measurements of the ambient magnetic field. Boom deployment, carrying the STAFF instrument, failed resulting in a number of spacecraft interferences in the wave power measurements. The attitude computer on each spacecraft also failed at different times on each spacecraft so that the spin axis orientation could not be controlled, resulting in a drift from nominal perpendicular orientation (after November 2004, for TC-1, and after July 2004, for TC-2). This drift of the spin axis became pronounced during the mission extension: TC-1: 9ı drift up to end of 2006; TC-2: 30ı drift up to July 2006. The spacecraft attitude was also derived from magnetometer data for operations and data analysis, providing an additional tracking of the spin axis orientation. In addition to these particular issues, the spacecraft environment induced a number of anomalous instrument resets, particularly affecting the PEACE instrument, and there were some calibration issues for some of the Chinese instruments. Nevertheless, overall, both DSP spacecraft operated with nominal parameters and all instruments operated continuously. The ground segment provided 99% data recovery via 3.3 h of data dump per day, which was initially disseminated via an adaptation of the Cluster data distribution system, comprising an internet based quick-look, summary data

336

M.W. Dunlop et al.

facility (DSDSweb), holding data up to five days after acquisition, and a data management system (CDMS) holding validated summary (1 min time resolution) and prime (spin averaged) parameters. Both systems contain data viewing capabilities, with the DSDSweb providing a range of standard summary plots, as shown in Fig. 22.3. Data can be downloaded in CDF or ASCII files formats (http://www.rssd.esa.int/index.php?project D DOUBLESTAR&page D about dsds). Science quality or high resolution data, also distributed, on request from the PI teams.

Fig. 22.3 Example summary data plot from the DSDS data system. A number of options for display are possible. This plot shows a set of instrument data for TC1, with magnetic field and plasma moments in the top panels, energy spectrogams for ions and electrons and magnetic wave spectra at the bottom

22

Double Star: Mission, Instruments and Joint Observations

337

22.3 Orbit Strategy and Mission Extension Following the nominal 18 months of operations both the CNSA and ESA accepted a case for a mission extension based on the data quality achieved and the additional science aims: to increase the number of conjunctions between Cluster and Double Star; to measure small, medium and large scales simultaneously; to measure size of large scale structures at the magnetopause/cusp; to observe more rare events like reconnection and storms, and to acquire “stereo” ring current images with IMAGE. Table 22.2 shows the orbits phases for the mission which were executed. The nominal mission is shown in grey shading and summarises the coordination with Cluster, as is plotted in Fig. 22.1. In the plots, the TC-1 and TC-2 orientations shown are nominal. For the initial dayside phases in spring 2004 and 2005, Cluster was set in small and meso-scale configurations, crossing the magnetopause above and below TC-1, at 5–10 RE distance. For the tail phase, summer 2004, the Cluster sampled the whole plasma sheet (15–19 RE ) as TC-1 sampled the near-current sheet at a range of down-tail distances (<12 RE ) and TC-2 sampled the inner magnetosphere. The other phases shown represent the extended DSP mission and correspond to an initial tail phase, summer 2005, when Cluster was in a large planar configuration containing both large and small separations. For this epoch, the orbital plane of TC-1 was raised above the magnetic equator and the apogee of TC-2 had fallen into the tail orientation. The two later phases corresponded to large Cluster separations, with the apogees of TC-1 lying at mid-magnetopause latitudes and TC-2 lying south. During these 3 years the line of apsides on TC-1 separates in local time duskward of Cluster, so that by 2006 onwards, this local time separation is quite pronounced (3–5 h). In addition the Cluster apogee falls during the mission (these orientations can be seen in Fig. 22.4). Finally, the spring phase in 2007, coordinates well with the Themis, 5 spacecraft mission, which were placed in a string of pearls configuration lying on the duskside magnetopause as Cluster lay on the dawnside and TC-1 lay near noon. Table 22.2 The table summarises the orbit phases of DSP through the extended mission Cluster Separation TC1-Cluster TC2- Cluster (km) (Re) (Re) Year Comment 2004 Spring 250 10 Nominal mission 2004 Summer 1000 5 15 Nominal mission 2005 Spring 1300 7 5 Nominal mission 2005 Summer 1000 and 10000 7–8 8–9 -TC-2 apogee tail -TC-1 apogee above equator 2006 Spring 10000 1–2 4–5 -TC-2 South pole -TC-1 mid-latitude magnetopause 2006 Summer 10000 6 15 -TC-1 apogee back to equator separated in azimuth with Cluster -TC-2 South Pole

338

M.W. Dunlop et al.

Fig. 22.4 Summer 2005 orbit configuration, X–Z GSE (left) X–Y GSE (right). TC-1 in black, TC-2 in red and Cluster 3 in magenta

Figure 22.4 shows the status for the summer phase 2005 orbit configuration, illustrating the close coordination possible between all six spacecraft at a range of simultaneous distances down tail (between 5–19 RE ), which allows busty bulk flows originating near Cluster locations to be monitored through the flow breaking region around 9 RE . The separation in local time is also apparent and is more pronounced in the following season, which allows post-midnight processes to be probed such as the azimuthal extent of the current disruption region. On the dayside, this configuration allowed closely separated five spacecraft measurements to be made at mid-high latitudes and in the cusp.

22.4 Selected Scientific Results The key scientific results from the combined mission have been published recently in a special issue of the Journal of Geophysical Research (see Zong et al. [15] for an introduction and for a full list of publications). Here we select a few results to briefly indicate some of the capabilities of the operations.

22.4.1 Magnetotail Phenomena The opportunity provided by DSP and Cluster coordination in the magnetotail is perhaps most simply illustrated by the study by Zhang et al. [14] as summarised in Fig. 22.5. This study dealt with the identification problem of surface waves in the current sheet to confirm their propagation and extent. The comparative spacecraft locations show that the wave source for surface waves moving in the dawn/dusk

22

Double Star: Mission, Instruments and Joint Observations

339

Fig. 22.5 Oscillations in the magnetotail current sheet, sampled along the tail by DSP and Cluster [14]. The right hand figure shows the spacecraft locations and schematic of the expected waves based on the five spacecraft magnetic field profile shown on the left. This trace shows that all Cluster and the TC1 spacecraft see a phased oscillation during the crossings through the current sheet

direction was extended along the length of the magnetotail. The Cluster configuration (at small 200 km separations) and TC-1 are separated by 5 RE along the tail in XGSM . The magnetic field data, shown here as a five spacecraft plot in the left panel, shows matched oscillations during the passage through the current sheet. This implies that all spacecraft sample a train of waves with the same amplitude and frequency as they cross the current sheet. The wave speed in Y is calculated as 30 km=s, which is confirmed by both the local Cluster timing and the positioning of TC-1. Thus, the local propagation properties, seen at Cluster, are confirmed to have a global extent by the coordinated TC1 measurements. Figure 22.6 summarizes the study of Zhang et al. [13], which is selected for illustration to show the favourable location of TC-1. The measurements show that TC-1 is often in the key region for magnetic pile-up. The associated dipolarization front appears to move tailward, passing TC-1. In a number of other events, statistically analysed by Zhang et al. [13], the relative timing of the onset and dipolarisation are confirmed. Some of these events also have good coordination with Cluster, which generally passed north to south through the current sheet at 15–19 RE : To illustrate other studies performed on the combined measurements in the tail, we note Zong et al. [16], Sergeev et al. [9], Takada et al. [10] and Zhang et al. [13], who studied BBFs, flow breaking and dipolarization events in the context of the five and six point measurements. Other combined tail results are fully reviewed in other papers in this volume.

340

M.W. Dunlop et al.

Fig. 22.6 Tailward motion of the dipolarization front, sampled by TC-1, where the magnetic field reconfiguration is seen as a flux pile-up signature 1–3 min before substorm onset (the auroral breakup is co-incident with the onset of Pi2 fluctuations: dotted vertical line), confirmed by a statistical study of 54 events [13]

22.4.2 Selected Scientific Results: Dayside Phenomena Multi-point studies of the dayside magnetopause using Cluster have been extended to five and six spacecraft, multi-scale data, by combining with Double Star to study specific, coordinated events, which include: 1. Investigation of magnetopause morphology and dynamics, such as: orientation, motion and thickness; magnetopause currents; structure, and boundary waves

22

Double Star: Mission, Instruments and Joint Observations

341

(a variety of techniques have allowed quantitative information to be determined such as multi-spacecraft and single spacecraft timing methods). The four spacecraft Cluster methods may be combined with the five and six spacecraft data for analysis of specific, coordinated Double star events. 2. Analysis of reconnection signatures in terms of: their direct X-line sampling; FTE occurrence, motion and structure, and corresponding ground signatures. 3. Boundary layer processes and their structure have been probed in terms of the null field and X-line formation, the formation of thin boundary layers and E-M fields. 4. Cusp morphology and structure have also been studied in coordination with adjacent locations on the magnetopause in terms of: the high altitude and exterior cusp region; its structure and dynamics; energetic particles, and the mid-altitude cusp. Examples of these coordinated events are given below.

22.4.2.1 Cusp-Magnetopause Boundary Layer A recent study by Bogdanova et al. [1] used a reverse conjunction of the spacecraft as shown in Fig. 22.7. The left hand panel shows the view in Z; XGSM of orbit segments of the Double Star TC-1 and Cluster, with the four spacecraft configurations enlarged at times along the orbit. Cluster is outbound and moves across the cusp deep into the dayside plasma sheet as shown. In an overlapping interval, the TC-1 spacecraft moves from the magnetosheath into the low latitude magnetosphere. These crossings occurred during stable northward IMF and it is apparent

Cluster 4 Cluster - Orbit & s/c - configuration 2004 Feb 28

10 8

Z Re

6 4

TC-1

2 0 −2 −4 −2

0

2

4

6

8

10

12

X Re

Fig. 22.7 A close conjunction where Cluster traversed the Cusp while TC-1 entered the low latitude magnetosphere. The locations are shown in the orbit view on the left (projected into Z,X GSM), with the red circles corresponding to the measured electron energy spectrograms shown on the right. TC1 is seen to enter through a complex boundary layer into the magnetosphere, while Cluster crossed through the cusp and equatorward boundary into the magnetosphere. Reproduced by permission of American Geophysical Union

342

M.W. Dunlop et al.

from the panels on the right, which show energy spectrograms of the electron flux, that TC-1 crosses a complex boundary layer at the magnetopause while the Cluster array also cross an extended boundary layer as it traverses from the cusp to the magnetosphere. There is an overlapping interval during which similar changes in the electron distribution are seen at each location and are related to temporal changes. The electron distribution shows a number of characteristic, bi-streaming and trapped and heated populations consistent with a period of lobe reconnection in both hemispheres. These results illustrate how the distributed measurements may cover on the one hand multi-point, local sampling of one region by Cluster, which is magnetically connected to a distant site, sampled simultaneously by TC-1.

22.4.2.2 Flux Tube Motion and Reconnection Flows A similar conjunction is summarised in Fig. 22.8a, reported by Dunlop et al. [3], where the Cluster array and the TC-1 spacecraft lay almost equidistant, but above (green spot) and below (red spot) the sub-solar point as indicated. All spacecraft are outbound and crossed the magnetopause within 20 min of each other. While in the adjacent magnetosheath all spacecraft observed signatures of reconnected flux tubes, moving northward at Cluster (determined by four spacecraft timing) and southward at TC-1 (polarity of the FTEs) in the directions predicted in Fig. 22.8a.

b

Flow position distribution 20

Cooling model

IMF Clock Angle

0

−20 −20 −10 0 10 Y_SSMP (RE)

50

20

50

−20

Motion of Flux Tubes for the IMF [5.0 -5.0 -5.0] at location [8.6 - 2.7 0.0] −40 −40

−20

0

Ygsm in Re

20

40

IMF−90 ± 20°

0

0 Ntotal=59

Nmax=9 θx→y = 100°

10 Ntotal=55 Nmax=12 θx→y = 86.9°

0

−10

−20

50 0 50 100 Num of flow

Flow angle distribution 10

IMF−90 ± 20°

10

−20

θx→y = 153°

100 20 100

Num of flow

TC1

IMF−90 ± 20° Ntotal=322 Nmax=71

0

Flow position distribution

0

Z_SSMP (RE)

Zgsm in Re

10

−10

C4

20

Flow angle distribution 100

IMF−90 ± 20°

Num of flow

40

Z_SSMP (RE)

a

−10 0 10 Y_SSMP (RE)

20

10

10

0 10 Num of flow

Fig. 22.8 (a) Simultaneous sampling of oppositely directed reconnected flux tube motions, arising from a low latitude X-line [3] and (b) Statistical study of fast (reconnection) flows at the magnetopause [8]. In (a) the panel shows expected tracks of reconnected flux as seen from the sun (northward branch, solid; southward branch, dashed), where the diamonds represent the cusps and the green and red spots are the positions of Cluster and TC-1 respectively, while in (b) the panels show the distribution of fast ion flows at low latitude (top) and high latitude (bottom)

22

Double Star: Mission, Instruments and Joint Observations

343

The timing and motion of these signatures confirmed that the reconnected flux arose from a low-latitude X-line, tilted as indicated, and as expected from the IMF orientation. Again the local information determined quantitatively from the Cluster array data, is confirmed via a simultaneous (conjugate) measurement at TC1. Figure 22.8b shows a statistical study performed by Pu et al. [8], which demonstrated that the observed reconnection flows are consistent with a tilted, low-latitude X-line, controlled by the IMF By component, as expected for component driven reconnection. At the high latitudes, seen by Cluster the X-line geometry is consistent with predominantly anti-parallel reconnection and completes a characteristic S-shaped form. Thus, these combined measurements show a statistical correspondence, which leads to a global view of the reconnection geometry.

22.4.2.3 Boundary Layer Extent and Energisation Figure 22.9 shows a study first reported by Wang et al. [11], in which the Cluster array skirted the dusk flank magnetopause at high latitudes, while TC-1 simultaneously sampled the magnetopause at lower latitudes (with a local time correspondence as indicated by the orbit segments shown in the top panel). The five spacecraft are therefore closely located, with TC-1 lying only 3 RE away from the four Cluster spacecraft, so that all show the same time delayed set of magnetopause crossings (see magnetic field traces in the lower two panels, shown in MVA determined boundary normal coordinates). During the period of interest, the spacecraft also sample a number of reconnected flux tubes rolling along the magnetopause surface as indicated in cross-section in the top panel. These flux tubes are shown to be tilted and are sampled at positions along their length. This is the first time that the FTE dimension in both local time and latitude has been simultaneously mapped. Similar FTE signatures result at all five spacecraft, so that the implied flux tube cross-section was shown to be similar at both locations. Energetic electrons, shown in the second and third panels, exhibit characteristic, pitch angle distributions, which correlate with the sampled depth within the boundary layer, so that the five spacecraft measurements allow both the extent in latitude, and the nature, of the magnetopause boundary layer to be investigated. Simultaneous Double Star and Cluster FTE observations on the dawnside flank of the magnetosphere were also analysed by Marchaudon et al. [7].

22.5 Discussion and Conclusions This report has summarised the coordinated science operations of the Double Star mission, which were realised though more than 3 years of extended mission; as well as showing the achievements resulting from such detailed coverage of the Earth’s magnetosphere. Only a few examples from the growing number of published studies, undertaken with the combined data set, have been shown. Other reviews in these

344

M.W. Dunlop et al.

Fig. 22.9 Shows a five spacecraft, close conjunction which achieved multi-scale sampling of a reconnected flux tube, and the magnetopause boundary layer. The top panel shows an X; YGSM view of the Cluster configuration (scaled by 5) and TC-1, for the orbit segment 6–9 UT. The squares are 2 RE each side. The lower panels show from the bottom: five spacecraft magnetic field data, the electron energy spectrograms and the energetic electron pitch angle distribution. The magnetic field traces are shown in standard cluster colours with TC-1 shown in blue. The magnetopause crossing and key FTE are indicated (dashed red line and arrow)

22

Double Star: Mission, Instruments and Joint Observations

345

proceedings cover the main science results. The aim of this paper has also served to demonstrate the generic capabilities of coordinated missions having multi-scale capability. Cluster and DSP have together been able to cover a large number of regions in the magnetosphere. The Cluster array has provided detailed, closely separated, multi-point coverage for the first time ever in space-based measurements, while the DSP spacecraft has provided counterpoint measurements of physical behaviour at nearby or adjacent regions, using a number of almost identical measurements. On the dayside, for example, Cluster sampled the high-latitude magnetopause and cusp, while TC-1 provided comparative observations of the low-latitude magnetopause or adjacent magnetosheath. During each yearly pass of the spacecraft, simultaneous high and low latitude magnetopause and cusp crossings were often measured. Sometimes, combined magnetopause and bow shock or solar wind information was obtained. On the flanks, comparative measurements either side of the low latitude boundary layer have been obtained. This data set expanded into a distributed array of typically six spacecraft measurements during the later, large separation phases of Cluster where spacecraft were located often at different locations in the cusp, magnetopause and magnetosheath simultaneously. The examples shown here provide unambiguous evidence for sub-solar reconnection, where the X-line often lies between the spacecraft, and extensive observations of both FTE signatures in terms of their polarity and inferred motion, and direct sampling of reconnection induced flows. Although only briefly covered here, on the nightside, Cluster often occupied locations which cut the magnetotail at 16–19 Re, while TC-1 spent extended periods in the plasma sheet at around 8–13 RE . This coverage provided multiple tail locations, allowing comparative studies of the near Earth neutral line formation and current disruption at the flow breaking region. Good co-ordination with TC-2 was often achieved. Acknowledgements Qinghe Zhang is thanked for assisting with the figures. This work is partly supported by Chinese Academy of Sciences (CAS) visiting Professorship for senior international scientists grant no. 2009S1-54. This work is also supported by the CNSF Grants 40390150, 40674094 and Chinese Fundamental Research Project G200000784. UK co-authors were supported by the Science and Technology Facilities Council Rolling Grants.

References 1. Bogdanova, Y.V., C.J. Owen, M.W. Dunlop, J.A. Wild, J.A. Davies, A.D. Lahiff, M.G.G. Taylor, A.N. Fazakerley, I. Dandouras, C.M. Carr, E.A. Lucek and H. R`eme (2008), Formation of the Low-Latitude Boundary Layer and Cusp under the northward IMF: conjugated observations by Cluster and Double Star, J. Geophys. Res., 113, A07S07, doi.org/10.1029/2007JA012762. 2. Berchem, J., A. Marchaudon, M. Dunlop, C.P. Escoubet, J.M. Bosqued, H. R`eme, A. Balogh, C. Carr, and Z. Pu (2008), Reconnection at the dayside magnetopause: Comparisons of global MHD simulation results with Cluster and Double Star observations, J. Geophys. Res., 113, A07S12, doi:10.1029/2007JA012743.

346

M.W. Dunlop et al.

3. Dunlop, M.W., M.G.G.T. Taylor, J.A. Davies, C.J. Owen, Z. Pu, H. Laakso, Y.V. Bogdanova, Q.-G. Zong, C. Shen, K. Nykyri, B. Lavraud, S.E. Milan, T.D. Phan, H. R`eme, C.P. Escoubet, C.M. Carr, P. Cargill, M. Lockwood, and B. Sonnerup (2005), Coordinated Cluster/Double Star observations of dayside reconnection signatures, Ann. Geophys., 23, 8, 2867–2875. 4. Escoubet, C. P., M. Fehringer, and M. Goldstein (2001), Introduction: the Cluster mission, Ann. Geophys., 19, 1197–1200. 5. Fear, R. C., Fazakerley, A. N., Owen, C. J., Lahiff, A. D., Lucek, E. A., Balogh, A., Kistler, L. M., Mouikis, C., R`eme, H. (2005), Cluster observations of bounday layer structure and a flux transfer event near the cusp, Annales Geophysicae, 23, Issue 7, 2005, pp. 2605–2620. 6. Liu, Z.X., Escoubet, C.P., Pu, Z., Laakso, H., Shi, J. K., Shen, C., and Hapgood, M. (2005), The Double Star Mission, Ann. Geophys., 23, 2707–2712. SRef-ID: 1432–0576/ag/2005–23–2707. 7. Marchaudon, A., C. J. Owen, J.-M. Bosqued, A. N. Fazakerley, M. W. Dunlop, A. D. Lahiff, C. Carr, A. Balogh and H. R`eme, (2005) Simultaneous Double Star and Cluster FTEs observations on the dawnside flank of the magnetosphere, Ann. Geophys., 23, 8, 2877–2887. 8. Pu, Z.-Y., X. G. Zhang, X. G. Wang, X.-Z. Zhou, L. Xie, M. W. Dunlop, Q. G. Zong,. C. J. Xiao, J. Wang, S. Y. Fu, Z. X. Liu, C. Shen, E. Lucek, C. Carr and H. R`eme (2007), Global view of dayside magnetic reconnection with the dawn-dusk IMF orientation: a statistic study for TC-1 and Cluster data, Geophys. Res. Lett., 34, L20101, doi:10.1029/2007GL030336 9. Sergeev, V., M. Kubyshkina, I. Alexeev, A. Fazakerley, C. Owen, W. Baumjohann, R. Nakamura, A. Runov, Z. V¨or¨os, T.L. Zhang, V. Angelopoulos, J.-A. Sauvaud, P. Daly, J.B. Cao, E. Lucek, Study of near-Earth reconnection events with Cluster and Double Star, J. Geophys. Res., 113, A07S36, doi:10.1029/2007JA012902, 2008. 10. Takada, T., R. Nakamura, L. Juusola, O. Amm, W. Baumjohann, M. Volwerk, A. Matsuoka, B. Klecker, K. Snekvik, C.J. Owen, A.N. Fazakerley, H.U. Frey, H. R`eme, E.A. Lucek, C. Carr (2008), Local field-aligned currents in the magnetotail and ionosphere as observed by a Cluster, Double Star, and MIRACLE conjunction, J. Geophys. Res., 113, A07S20, http://dx.doi.org/10.1029/2007JA012759. 11. Wang, J., M. W. Dunlop, Z. Y. Pu, X. Z. Zhou, X. G. Zhang, Y. Wei, S. Y. Fu, C. J. Xiao, A. Fazakerley, H. Laakso, M. G. G. T. Taylor, Y. Bogdanova, F. Pitout, J. Davies, Q. G. Zong, C. Shen, Z. X. Liu, C. Carr, C. Perry, H. R`eme, I. Dandouras, P. Escoubet, and C. J. Owen (2007), TC1 and Cluster observation of an FTE on 4 January 2005: A close Conjunction, Geophys. Res. Lett., 34, L03106, doi:10.1029/2006GL028241. 12. Wild, J. A., Milan, S. E., Davies, J. A., Dunlop, M. W., Wright, D. M., Carr, C. M., Balogh, A., R`eme, H., Fazakerley, A. N., and Marchaudon, A. (2007), On the location of dayside magnetic reconnection during an interval of duskward oriented IMF, Annales Geophysicae, 25, 219–238. 13. Zhang, H., Z. Y. Pu, X. Cao, S. Y. Fu, Z. X. Liu, Z. W. Ma, M.W. Dunlop, W. Boumjohann, J. B. Cao, C. J. Xiao, Q. G. Zong, X. G. Wang, C. Carr, H. R`eme, I. Dandouras, A. Fazakerley, H. U. Fery and C. P. Escoubet. (2007) TC-1 observations of flux pileup and dipolarization-associated expansion during substorms, Geophys. Res. Lett., 34, L03104. 14. Zhang, T. L., Nakamura, R., Volwerk, M., Runov, A., Baumjohann, W., Eichelberger, H. U., Carr, C., Balogh, A., Sergeev, V., Shi, J. K., Fornacon, K.-H. (2005) Double Star/Cluster observation of neutral sheet oscillations on 5 August 2004, Annales Geophysicae, 23, Issue 8, pp. 2909–2914. 15. Zong, Q.-G., C. P. Escoubet, Z. Y. Pu, and Z. X. Liu (2008), Introduction to special section on Double Star-Cluster Coordinated Studies on Magnetospheric Dynamic Processes, J. Geophys. Res., 113, A07S01, doi:10.1029/2008JA013146. 16. Zong, Q.-G., H. Zhang, S. Y. Fu, Y. F. Wang, Z. Y. Pu, A. Korth, P. W. Daly, and T. A. Fritz (2008), Ionospheric oxygen ions dominant bursty bulk flows: Cluster and Double Star observations, J. Geophys. Res., 113, A07S23, doi:10.1029/2007JA012764.

Part V

Observations of Solar Wind and Magnetosheath

Chapter 23

Effect of Shock Normal Orientation Fluctuations on Field-Aligned Beam Distributions K. Meziane, A.M. Hamza, M. Wilber, M.A. Lee, C. Mazelle, E.A. Lucek, T. Hada, and A. Markowitch

Abstract We address the unsolved question of how foreshock field-aligned beam (FAB) parallel temperatures are produced. Studies including numerical simulations and recent observations have indicated that shocks can be nonstationary and include embedded spatial structures with varied scales. As a first step towards assessing the impact of such variability on backstreaming ions, we examine how a randomly distributed shock normal direction will affect FAB parallel velocity (vk ) distributions. Assuming that the FABs are produced in a quasi-adiabatic reflection process at the shock, we derive a probability distribution function for vk . These derived distributions exhibit second, third and fourth order moments that agree well with the observations for a large range of reflection efficiencies ı, and depend strongly upon the average angle between the magnetic field and the shock normal Bn0 . Best agreement is obtained for fluctuations of the normal orientation of a few degrees about a nominal direction. The derived model predicts a strong correlation between the shock geometry (Bn0 ) and the moments of the parallel velocity distribution, but with stronger tails extending to higher values of Bn0 , a trend opposite to the observations.

K. Meziane () and A.M. Hamza Physics Department, University of New Brunswick, Canada e-mail: [email protected] M. Wilber Space Sciences Laboratory, University of California, Berkeley, USA M.A. Lee Space Science Center, University of New Hampshire, Durham, USA C. Mazelle Centre d’Etudes Spatiales des Rayonnements, Toulouse, France E.A. Lucek Space and Atmospheric Physics, The Blackett Laboratory, Imperial College, London, UK T. Hada ESST, Kyushu University, Fukuora, Japan A. Markowitch Laboratoire de Physique et Astroparticules, Universit´e de Montpellier, France

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 23, c Springer Science+Business Media B.V. 2010

349

350

K. Meziane et al.

23.1 Introduction It is now established that ion beams roughly propagating sunward along the interplanetary field lines are ubiquitous upstream of the Earth’s quasi-perpendicular bow shock [7, 21, and references therein]. These field-aligned beams (FABs) are found on interplanetary magnetic field (IMF) lines connecting to the shock with angles to the shock normal Bn ranging from 40ı to 70ı [2]. The basic properties of FABs including density and speed are very well known, with both density and speed controlled by the shock geometry. Tenuous and high-speed beams are likely to originate at a location on the shock at which the geometry is quasi-perpendicular while denser, slower beams emerge from oblique configurations. Geometric arguments indicate, and observations confirm, that the beam speeds should be 1–2 the shock speed cos V n ; (23.1) Vs D vsw cos Bn where V n is the angle between the local shock normal and the direction of the flow, and vsw is flow (or the solar wind) speed; Vs is the speed at which the shock moves parallel to the magnetic field in the solar wind frame. Beam densities vary considerably, from <0:1% to a few % of the incident solar wind density. An exhaustive investigation of FAB thermal energy has been carried out only recently [14]. In a prior study, [13] showed that the reduced FAB distributions are organized into two categories: those with nearly-Maxwellian profiles, and those exhibiting enhanced high energy tails. In [14], when considering only nearly-Maxwellian distributions, found that FAB parallel thermal energy is dependent upon the solar wind speed and weakly sensitive to the shock geometry. They showed that the parallel thermal speed normalized to the solar wind speed, k , obtained from fitting the core distribution to a Maxwellian form, is in the range 0.27–0.33. The inclusion of the high energy tail, when present, will likely lead to a measured standard deviation † that is higher than k . Despite a precise knowledge of FAB characteristics, a comprehensive model that accounts for the production mechanism is still a topic of active research. Relatively recent investigations based upon Cluster observations found FABs with significant densities for quasi-perpendicular shock geometries [9]. Simple specular reflection should return ions to the shock in these geometries, but measurements in the foot of the shock show phase space densities in the escape cone sufficient for producing the FABs [15], while distributions downstream lack these [9]. One possibility is that these ions have been pitch-angle scattered into the escape cone following the specular reflection [9]; these latter populations are always produced by supercritical shocks. All the suggested mechanisms are geometry dependent; however, it is still not clear, when discrepancies between observations and models exist, whether they are due to uncertainties in determinations of the geometry or are inherent to theoretical models. It is not the purpose of this study to test the different FAB production mechanisms; we only consider a simple, quasi-adiabatic reflection process. A similar approach could be applied for other mechanisms. The reflection

23

Effect of Shock Normal Orientation Fluctuations

351

mechanism assumes that a fraction of the incoming solar wind ions reflects off the shock in a manner that approximately conserves energy and the first adiabatic invariant D v2? =B. Based on this assumption, [20] derived the energy gain in the shock frame. As previously mentioned, the model falls short of accounting for FAB thermalization or reflection efficiency, although bulk speed predictions are satisfactory [17]. As with all simple analytical models, the shock surface is treated as planar over scales of an ion gyroradius. This letter is the first attempt to account for the FAB parallel thermal energy using a geometric particle reflection model (Perpendicular temperatures for FABs may be determined by unrelated considerations, and are not addressed here). Here we employ quasi-adiabatic, instantaneous reflection at a shock having a local normal that exhibits small, normally-distributed fluctuations in orientation from an average. This model has been motivated by recent studies, including observations and simulations, of shock non-stationarity [8, 10, 11, 18] as well as the presence of spatial irregularities on the shock surface [16]. In the next section, we present and briefly discuss typical velocity distribution functions of FABs observed by the Cluster/CIS experiment. The impact of the velocity filtering on the distributions is briefly considered in Section 23.3. In Section 23.4, we describe the model, and in Section 23.5 present the results. Finally, we discuss these results and compare them with observations in Section 23.6.

23.2 Foreshock FAB Reduced Parallel Distribution Functions Figure 23.1 displays a sample of reduced parallel distribution functions associated with FABs observed upstream of Earth’s bow shock. The dashed red lines correspond to the best fit using a shifted Maxwellian with a standard deviation k . The figure illustrates typical FAB distributions present in the terrestrial foreshock. These often exhibit high energy tails with varying degrees of hardness. We note that there appear to be inflection points in some of the tails, such as in the distributions (d), (g) and (j). Preliminary studies suggest that beams in more perpendicular geometries have nearly Maxwellian distribution profiles [13]. The distributions of Fig. 23.1 are observed at various distances from the shock as indicated by the values of D (in RE ) inside each panel. It is important to notice that the distance from the shock – at least within Cluster’s 18:7 3:0 RE orbit – is not an important bias for the distributions’shape. As mentioned in the Introduction, the particle distribution features are essentially controlled by the solar wind speed and the shock geometry. Distribution (c) in Fig. 23.1 has vsw D 660 km s1 , and appears wider than other sampled distributions such as (a) or (b). However, the values of k normalized to the solar wind speed are all in agreement with the 0.27–0.33 range of our previous finding [14]. Despite evident differences in the shape of these distributions, the FAB core distributions are remarkably well-fit by Maxwellians. We can anticipate that when tails are present the full-distribution standard deviation † will be larger than k obtained from a

352

K. Meziane et al. CLUSTER/CIS−HIA P23

a

10

2005−03−06/0304:34−04:47 D = .02

10

10

6

2003−03−28/2027:14−27:26 D = .05

Σ = .36

Σ = .42

γ1= .12

γ1= .92

γ2= .31

γ2= 1.77

0

10

0

b 10

f

6

1500

0

3000

0

g 10

6

1500

3000

6

2005−03−06/0254:52−55:05

2003−03−01/2244:41−44:49

D = 2.5

D = 8.6 Σ = .30

10 10

γ1= .41

γ2= .33

γ2= .41

0

0

c

γ1= .13

1500

3000

6

2005−03−01/0946:02−46:14 D = .01 Σ = .39 γ1= .05 γ2= .23

0

10 −3000

d 10

−1500

0

6

2001−02−23/0215:36−15:48 D = .11 Σ = .41 γ1= .16 γ2= .25

10 10

0

h 10

3000

2003−02−25/0220:09−20:17 D = 12.4 Σ = .42 γ1= .12 γ2= .19

0

10 −3000

i

−1500

0

6

10

2003−01−27/2235:16−35:24 D = 18.2 Σ = .28 γ1= .84 γ2= 1.90

10 −3000

3000

6

2001−02−23/0214:00−14:12

j

−1500

0

6

10

2003−01−27/2235:22−35:30 D = 22.4

Σ = .28

Σ = .38 γ1= .18

γ1= .59

γ2= .52

γ2= .84

0

0

1500

6

0

1500

D = .15

10

0

10

0

0

e

Reduced Parallel Distribution Function [arb. units]

Reduced Parallel Distribution Function [arb. units]

Σ = .40

0

1500

V|| [km/s]

3000

10 −3000

−1500

0

V|| [km/s]

Fig. 23.1 Examples of reduced parallel distribution functions associated with FABs observed upstream of Earth’s bow shock. In each frame, the dashed line represents the best Maxwellian fit. The observations are from the Cluster/CIS experiment

23

Effect of Shock Normal Orientation Fluctuations

353

Maxwellian fit to the core. Moreover, all distributions appear asymmetric and it is common to see a low energy cut-off due to ions that could not overcome the shock speed; nevertheless, the low energy part of the distribution is satisfactorily fit by a Maxwellian. In order to quantify the distributions’ asymmetry and ‘squareness’, we have calculated the skewness 1 and the kurtosis 2 of each distribution of Fig. 23.1; a numerical value of 3 has been subtracted from 2 values to make a null kurtosis for a standard normal distribution. The skewness in our case is a measure of the extent to which particles moving along one direction (in the beam frame) tend to have higher speeds than along the opposite direction. In plasma kinetic theory, the skewness is equivalent to the heat flux. The kurtosis is a measure of how peaked the function is, with values >0 indicating a narrow peak with wide tails (more ‘extreme’ values than a Gaussian distribution). The obtained values of the skewness 1 and the kurtosis 2 are indicated inside each panel of Fig. 23.1. As expected, the highest values of 1 are associated with distributions with strongest apparent tails. These determinations of the first moments of FAB reduced parallel velocity distribution functions lead to values of 1 0:05–0.92 and 2 0:19–1.69.

23.3 Accessible Source Regions and Velocity Filtering With solar wind convection speeds an appreciable fraction of a backstreaming ion’s parallel speed, it is important in a discussion of particle sources to distinguish between field lines and guiding center paths. Below we treat references to field lines as instantaneous mappings along field lines from one contact point to another, while guiding center references consider connections along particle paths that are affected by convection as well as particle parallel motions. Implicit in this is the idea that a particle originating at the shock on a field line threading a spacecraft cannot be observed unless the solar wind flow and the field are aligned, or the ion velocity is infinite. The size of the shock source region accessible to a spacecraft depends upon the direction of the magnetic field as well as the distance from the shock. Simple geometric considerations indicate that the source regions accessible to observers near the shock will be narrowest. The goal of this section is to provide an estimate of the time of flight effect in order to examine its possible impact on the shape of FAB ion distribution functions. For analytical simplicity we consider only the case of stationary upstream solar wind and IMF conditions. Here the bow shock is modeled as a paraboloid with nominal values for the sub-solar distance as and the flaring-parameter bs D 0:25=L, where L is the semi-latus rectum. For zero pitch angle ions, the trajectory lies in the Bvsw -plane, which contains both the solar wind and the magnetic field directions. FAB ions that are detected by the spacecraft all emanate from the region where the Bvsw -plane intercepts the paraboloidal shock. Using simple algebra, the maximum length of the source region is obtained for a given spacecraft location, and solar wind flow and magnetic field directions. This region is bounded by the intercept of the field line threading the spacecraft (‘A’ in Fig. 23.2, described below) and the limiting ion guiding center trajectory, i.e., one that is tangent to the shock (‘C’).

354

K. Meziane et al. Bow shock cross section

Bow shock cross section 0

A

C

η (RE)

η (RE)

30

B

15

−1

A

−20

C SC

0 Event (g) 0 X (RE)

SC

20

Event (d) −2 13

13.25

13.5

X (RE)

Fig. 23.2 Schematic diagram of the shock cross section in the Bvsw -plane showing the shock region source for events (d) and (g) of Fig. 23.1. The dashed blue line with label B indicates the IMF line tangent to the bow shock cross section. Only ions emanating between A and C can reach the spacecraft SC

Figure 23.2 illustrates the geometry for events (d) and (g) of Fig. 23.1. The curves shown do not represent the maximum extents of the paraboloidal bow shock, but instead the parabolic intersections of the Bvsw -plane with it. (The parameterizations of these intersection curves differ from those for the shock itself, and depend upon B vsw and the observer’s location.) The X -axis represents the solar wind direction while the -axis is an orthogonal direction in the Bvsw -plane. The dashed blue line ‘B’ shows the IMF field line in the Bvsw -plane that is tangent to the shock. For the paraboloidal model bow shock [4] we have adopted, distribution (d) is observed D 0:11 RE from the shock, while distribution (g) is observed at D 8:6 RE . Distributions (d) and (g) have Gaussian cores and nearly identical normalized ion bulk speeds, with distribution (d) having a significantly larger standard deviation. We can determine the variations AC in the shock normal direction between points A and C. For event (g), with D D 8:6 RE , we find AC D 25ı . This should lead to significant changes in parallel velocities at these differing source locations, according to a conventional quasi-adiabatic reflection picture, but velocity filtering must also be considered. Particles leaving point C with parallel speeds larger than the shock speed will travel further than 8.6 RE along the magnetic field direction in the time that it takes the field line to convect over to the spacecraft; velocity filtering in fact works to counter the increased beam speeds expected for increases in Bn towards this extreme source location. In contrast, event (d) has D D 0:11 RE and AC D 2:6ı , and a locally-planar shock description should apply. Both distributions (d) and (g) were observed to have similar normalized peak beam speeds (2.5 and 2.6, respectively), suggesting similar nominal shock geometries. However, event (g) has a narrower normalized velocity spread († D 0:30) than event (d) († D 0:41); apparently, a larger AC does not lead to a hotter parallel velocity

23

Effect of Shock Normal Orientation Fluctuations

355

(vk ) distribution. Explanations of the thermal spread in vk seem instead to require a process that is intrinsic to the shock, rather than this form of geometric argument. We note that the particular choice for the shock parameterization has little impact on the results. We found that the above AC determinations, based on the parabolic bow shock model, are very similar to those obtained from an hyperbolic model [19] and an elliptical model [6]. A second consideration is the skewness and kurtosis for these distributions, which indicate the presence of a non-thermal tail. The tail for event (g) (1 D 0:41, 2 D 0:41) is larger than that for event (d) (1 D 0:16, 2 D 0:25), and is consistent with a larger spread in Bn values for the contributing source regions. However, distribution (f), observed at a distance D 0.05 RE from the nomimal shock, has even larger skewness and kurtosis values. This indicates that there is no a direct correspondance between the distance from the source region and the shape of the FAB distributions. However, the thermal broadening intuitively expected for a larger range of Bn values is not borne out for observed distributions considered here. The arguments above suggest that time of flight considerations are not of first importance in determining the shape of the FAB velocity distributions. Straightforward geometry shows that source point Bn values will depend upon guiding center directions, and the span of these will depend upon distance from the shock and IMF direction. Although we should then expect a wider range of vk values, and hence greater parallel temperatures, for a larger span of accessible source locations, the observed FABs in Fig. 23.1 do not support this († is not correlated with AC ). Likewise, if a broad range of source locations plays a role in producing non-thermal tails we should expect the skewness and kurtosis to increase with the observer’s distance from the shock. This too is not observed. A possibility is that the particles arriving at the spacecraft may otherwise be constrained to source points near ‘A’ in Fig. 23.2. In that event, the observed thermal spread in vk might instead be accounted for by localized effects at the shock. In the next section we present a geometric model employing fluctuations in the local shock normal direction to produce upstream velocity distributions with thermal features that are similar to those observed.

23.4 A Geometric Model Two features in the observed upstream velocity distributions (Fig. 23.1) that we wish to explain are the thermal spreads that are greater than those for the incident solar wind beam [2] and the non-thermal tails [13] seen near oblique foreshock geometries. As outlined below, we produce these by assuming quasi-adiabatic reflection of incident solar wind ions in the presence of normally-distributed fluctuations in the local shock normal direction about some mean. A convolution of the resulting range of reflected parallel ion speeds leads to good qualitative agreement, even when making simplifying assumptions.

356

K. Meziane et al.

First, we re-derive expressions due to Paschmann et al. [17] in the plasma frame of reference. Reflection is assumed to conserve energy in the de Hoffman–Teller reference frame [5], with imperfect conservation of . Using primes to denote the de Hoffman–Teller frame, an incoming ion with a parallel velocity v0i k is reflected with its parallel motion retaining a fraction ı of its incident momentum, v0k D ıv0i k , where 0 ı 1. The reflection is adiabatic for ı equal to unity. FAB observations indicate that ı is in the range 0.65–1.0 [3]. After transformation to the solar wind frame of reference (no primes), the pre- and post-encounter parallel velocities are related by vk Vs D ı.vi k Vs /, where Vs is the shock speed introduced previously (Eq. 23.1). Since bulk solar wind ions are at rest, we obtain for the normalized parallel beam speed: vk cos V n D .1 C ı/ (23.2) P D vsw cos Bn In Eq. 23.2, the angles Bn and V n are acute and obtuse, respectively, leading to a positive value for a backstreaming particles’s parallel speed. Equation 23.2 indicates that when the upstream flow vsw and magnetic field B are quasi-steady, the effects of shock geometry on normalized beam speed arise from variations in the normal orientation. Because the solar wind protons gyroradii are much smaller than the bow shock radius of curvature, we can reasonably apply an assumption of a locally planar shock for the interaction of individual thermal solar wind ions. However, the planar approximation is certainly not justified if small scale surface irregularities are present or when there is temporal variability. The development below presupposes that upstream ions emerge from a shock that has fluctuations in the local normal orientation. The model is equally applicable to temporal variability on times scales small compared to an ion distribution sampling time as it is to spatial variability on scales small compared with the (velocity filter) accessible source region. Here we regard n0 as the temporal (or regional) average of the local shock normal, which might be expected to agree well with normals obtained from quasi-empirical model shocks. Finally, we assume that the reflected particles escape upstream with speeds according to Eq. 23.2. Let us now consider an escaping ion which propagates upstream of the shock where the normal direction makes an angle ˆ with the respect to n0 . We assume for this discussion that all variations of ˆ are restricted to the Bvsw -plane. This assumption has no physical basis, but we employ it to render the derivation tractable. In this 2D approximation, the shock normal direction is given by n D .cos ˆ; sin ˆ/. Equation 23.2 can then be rewritten in terms of ˆ: P D .1 C ı/

cos.V n0 ˆ/ cos.Bn0 ˆ/

(23.3)

with V n0 and Bn0 angles characterizing the average local geometry based on n0 D (1, 0). Next we assume that the ions that reach the spacecraft leave the shock where the normal direction n is random while the average direction is n0 . We have no a priori basis for expecting the effective shock normal to fluctuate uniformly about

23

Effect of Shock Normal Orientation Fluctuations

357

its average, but assume a Gaussian distribution in ˆ as an ansatz to be tested by observations. If we permit ˆ to extend to ˙1, a naive application of Eq. 23.3 would include contributions for normal directions pointing into the shock. A rigorous treatment would truncate the range of ˆ such that =2 < Bn0 ˆ < =2, but for analytic simplicity we apply the simpler constraint =2 < ˆ < =2. In ı practice FABs are observed for 45ı < Bn < 75 , and as we will show below, reasonable agreement with observations can be obtained for Gaussian variations in ˆ with e-folding scales of just a few degrees. Consequently, only a tiny contribution in our analytic description will come from ˆ variations that reverse the shock direction. With this in mind, we write the normalized probability density function of ˆ as:

2 2

exp 2ˆ 2 fˆ ./ D p ˆ 2 erfc. ˆ p2 / erfc. ˆ p2 /

(23.4)

where ˆ is the standard deviation. Here we employ the convention used in probability theory, in which represents a particular value of the random variable ˆ. Given the distribution fˆ ./, and using Eq. 23.3, it is possible to derive the distribution fP .p/ of escaping ion parallel velocities (normalized to the solar wind). The probability that the random variable P is equal or less than a value p is given by Z

p

Prob.P p/ D 0

duf P .u/

(23.5)

where fP .u/ is the probability density distribution of P . The procedure for obtaining fP .p/ can be found in any textbook on probability theory (see for example [1]); it is straightforward to show that: fP .p/ D

2 j sin.V n0 Bn0 / j p ˆ 2 erfc. ˆ p2 / erfc. ˆ p2 /

p1

tan1

p cos Bn0 cos V n0 p sin Bn0 sin V n0

2

(23.6)

2ˆ e 2 C p 2 2p cos.V n0 Bn0 /

where D 1 C ı.

23.5 Results For a perfectly adiabatic reflection (ı D 1), the probability density function fP .p/ is plotted for different values of ˆ as shown on Fig. 23.3. For (V n0 , Bn0 ) = (160ı, 48ı ), the maximum normalized beam speed is P0 D 2:81. Both axes of Fig. 23.3 are scaled to allow a qualitative comparison with the observations shown in Fig. 23.1, with the y-axis spanning six orders of magnitude in space–phase

358

K. Meziane et al. 101

σΦ = 1∞

Probability Density Distribution fP (p)

100

θBn = 48∞

σΦ = 2∞

θVn = 160∞

σΦ = 3∞ σΦ = 4∞ σΦ = 5∞

10−1

10−2

10−3 Vs 10−4

0

1

2

3

4

5

6

7

8

P = V|| /VSW

Fig. 23.3 Reduced parallel distribution functions obtained from Eq. 23.6 and a shock geometry corresponding to a normalized beam speed P0 D 2:81. Each curve corresponds to a value of the standard deviation ˆ as indicated in the figure

density and the x-axis reaching 3,000 km s1 , for a typical solar wind speed of 375 km s1 . The dashed vertical goldenrod line indicates the shock speed for (V n0 , Bn0 ) = (160ı, 48ı ). Striking qualitative similarities between the observations and the results of the model illustrated in Fig. 23.3 are obvious. These include the high energy tails, and the Maxwellian form for the distributions just below the peak value. The tails harden for increasing values of ˆ , corresponding to increasing skewness values. By taking moments of Eq. 23.6, we can quantify the shapes of the computed distributions and compare them with the observed distributions of Fig. 23.1. The panels in Fig. 23.4 show moments computed as functions of ˆ for adiabatic reflection (ı D 1), with V n0 fixed at 160ı . Differently-colored symbols show the moments for three values of Bn0 : 45ı (black), 50ı (blue), and 55ı (red). Panel (a) shows that escaping ion bulk speeds are independent of ˆ . However, there is a clear dependence upon ˆ for the 2nd, 3rd and 4th moments of fP .p/. Of these higher moments, the second, †, shown in Panel (b), which can be related to the FAB parallel temperature, has the weakest ˆ -dependence. Most dramatic are the variations in 2 , which indicate the presence of more-extreme values, corresponding to the high-energy tail. Similar trends are found for values of ı smaller than 1 (not shown); for these cases the dependence upon ˆ is weaker.

23

Effect of Shock Normal Orientation Fluctuations

359

Moments of the Distribution Function 4

1 0.8

3 γ1

P

0.6 2

θBn = 45∞

0.4

θBn = 50∞

1

0.2

θBn = 55∞

0

0 0

2 4 σΦ [deg]

6

0

2

4 σΦ [deg]

6

0

2

4 σΦ [deg]

6

2

1 0.8

1.5

Σ

γ2

0.6 1

0.4 0.5

0.2 0

0 0

2

4 σΦ [deg]

6

Fig. 23.4 First (P ), second (†), third (Skewness 1 ) and forth (Kurtosis 2 ) order moments of the distribution fP .p/ versus ˆ for three values of Bn0

Finally we plot the dependence of the moments of Eq. 23.6 as a function of the normalized beam speed P , which we treat as a proxy for the shock geometry (cf. Eq. 23.1). Figure 23.5 shows the behavior of † and 1 for ˆ = 2ı (black open circles), 3ı (blue) and 4ı (red). Both † and 1 show a nearly linear dependence upon P , with the slope steepening as ˆ increases. Reductions in ı to values lower than 1 do not produce significant changes.

23.6 Discussion In this report we have attempted to explain the vk profiles observed for upstream field-aligned beam distributions. The underlying model assumes quasi-adiabatic reflection from a shock having local normal orientations that fluctuate randomly about a mean value n0 . Although the model assumes temporal changes in the shock normal direction at a specific point at the shock, the variations might also be viewed as resulting from spatial homogeneities at scales comparable to escaping ions gyroradii.

360

K. Meziane et al. 1

1

σΦ = 2∞ σΦ = 3∞ σΦ = 4∞

0.8

0.8 0.6

Σ

γ1

0.6 0.4

0.4

0.2

0.2

0

2

3 4 P = V|| / VSW

5

0

2

3 4 P = V|| / VSW

5

Fig. 23.5 Predicted second moment † and third moment 1 versus the beam speed P for three values of ˆ

Some support for both views is provided in the literature. For quasi-perpendicular geometries, moderate Alfvenic Mach number and low ˇ-plasma, numerical simulations have shown that nonstationarity is inherent to the shock. During the shock reformation cycle, the shock structure including the foot, the ramp and the overshoot undergo spatio-temporal changes yielding significant local variations. Also, a relatively recent study from Cluster by [16] reported coherent 1,000–2,000 km wavelength oscillations in the foot and the ramp of the Earth’s perpendicular bow shock. The authors interpreted the observations as ripples propagating across the shock surface along the magnetic field direction. It is obvious that the presence of such ripples would have a significant impact on the local geometry changes. A similar, but much slower, well-resolved variation of the local normal resulting from convection of large-amplitude ultra-low frequency (ULF) waves across the shock, and its consequences on backstreaming ions, was studied by Meziane et al. [12]. Another possibility is that small variations of the upstream plasma and field conditions, resulting from ubiquitous solar wind turbulence, would induce local changes in the shock normal as the shock adjusts to the variations according to well known conservation relations. Much of the ion foreshock is populated by magnetic pulsations, and it might seem that these could be a source of the normal orientation fluctuations. However, the larger pulsations tend to populate the quasi-parallel foreshock where FABs are not present, and would cause rather larger normal fluctuations than the few degrees required of our model. The model that we have developed is 2-dimensional, and unrealistically constrains the fluctuations of the source normal direction to the Bvsw -plane. We have no basis for expecting real fluctuations to have a preferred orientation, and such restrictions on degrees of freedom can be expected to exaggerate the influence of the variations. A complete treatment will necessarily require three dimensions. It is important to emphasize, as demonstrated in Sect. 23.2, that the shape of FABs is not an effect of the particle time of flight, although this might become an issue for beams observed sufficiently far from the source. No less important, the FABs presented in this report are not observed in association with ULF

23

Effect of Shock Normal Orientation Fluctuations

361

waves, although they are considered to be the main free energy source for the latter. Absent such waves, we expect that FABs at moderate distances from the shock will suffer little diffusion, with their properties essentially unchanged. Despite its limitations, our simple analytical model, based upon quasi-adiabatic reflection and normally distributed fluctuations in shock orientation, reproduces interesting features of FAB parallel velocity distributions. We have found that the range of predicted values of † are in remarkable agreement with the observations (Fig. 23.1), and require only modest angular fluctuation levels of ˆ = 2–4ı . Moreover, the qualitative features of this PDF are similar to those observed, and notably include high energy tails. The quantitative values for skewness 1 and kurtosis 2 can agree well with those obtained from the FABs of Fig. 23.1, although the agreement depends upon the value for Bn0 . The tails can be understood to result from the asymmetric influence of 1= cos Bn in Eq. 23.2, which varies slowly for small values of Bn and dramatically as Bn ! 90ı . This effect is folded into the denominator of Eq. 23.6. Notably, the present model suggests that the tail has a local source, rather than originating in populations emanating from other regions of the shock. This result is in agreement with our previous finding regarding the source of the of the distribution tails [13]. Our model does not predict any kinks in the distribution, however. It is important to note that the model predicts significant increases in the beam standard deviation and skewness when the shock geometry is increasingly perpendicular. This disagrees with the preliminary studies of Meziane et al. [13], who found that FAB tails are most apparent in oblique shock geometries, rather than perpendicular foreshocks. That work did not exhaustively isolate the roles of ı and V n0 . Perhaps most significant, however, is the observational fact that foreshock wave activity can increase dramatically in oblique geometries. The sharp ULF wave boundary that is seen apparently separates field-aligned beams from gyrating ion beams and diffuse ions [14]. If, e.g., wave activity falls off exponentially upstream of this ULF boundary, a thus-far neglected fluctuation level might contribute to the non-thermal FAB tails, providing increases that follow the observational trends. This would suggest that the model might be improved by including a variable ˆ , which would increase with smaller values of Bn0 . Quantifying the actual variation would require a comprehensive study of observed upstream plasma and field fluctuations as Bn and other parameters are varied. Acknowledgements Work at U.N.B. is supported NSERC. Work at U.C. Berkeley is supported by NASA Grant NNG05GF99G. The authors are grateful to the International Space Science Institute for supporting topics related to shock acceleration.

References 1. Bertsekas, D.P., Tsitsiklis, J.N.: General Random Variables. Introduction to Probability. Athena Scientific, P. O. Box 805, Nashua, NH 03061-0805, USA (2002) 2. Bonifazi, C., Moreno, G.: Journal of Geophysical Research 86, 4381 (1981)

362

K. Meziane et al.

3. Bonifazi, C., Moreno, G., Russell, C.T.: Journal of Geophysical Research 88, 7853 (1983) 4. Cairns, I.H., Fairfield, D.H., Anderson, R.R., Carlton, V.E.H., Paularena, K.I., Lazarus, A.J.: Journal of Geophysical Research 100, 47 (1995) 5. de Hoffman, F., Teller, E.: Physical Review 80(4), 692 (1950) 6. Farris, M.H., Petrinec, S.M., Russell, C.T.: Geophysical Research Letters 18, 1821 (1991) 7. Fuselier, S.A.: In: Engebretson, M., Takahashi, M., Scholer, M. (eds.) Solar Wind Sources of Magnetospheric Ultra-Low Frequency Waves, p. 91 (1994). American Geophysical Union 8. Hada, T., Onishi, M., Lemb`ege, B., Savoini, P.: Journal of Geophysical Research 108, 1233 (2003) 9. Kucharek, H., M¨obius, E., Scholer, E., Mouikis, C., Kistler, L.M., Horbury, T., Balogh, A., R`eme, H.: Annales Geophysicae 22, 2301 (2004) 10. Lemb`ege, P., Savoini, P.: Physics of Fluids B 4, 3533 (1992) 11. Lobzin, V.V., Krasnoselskikh, V.V., Bosqued, J.M., Pinc¸on, J.L., Schwartz, S.J., Dunlop, M.: Geophysical Research Letters 34, 05107 (2007) 12. Meziane, K., Wilber, M., Mazelle, C., LeQu´eau, D., Kucharek, H., Lucek, E., Sauvaud, J.A., Hamza, A.M., R`eme, H., Bosqued, J.M., Dandouras, I., Parks, G.K., McCarthy, M., Klecker, B., Korth, A., Bavassano-Cattaneo, M.B., Lundin, R.: Journal of Geophysical Research 109, 05107 (2004). doi:10.1029/2003JA010374 13. Meziane, K., Wilber, M., Hamza, A.M., Mazelle, C., Parks, G.K., R`eme, H., Lucek, E.: Journal of Geophysical Research 112(A01101) (2007). doi:10.1029/2006JA011751 14. Meziane, K., Wilber, M., Hamza, A.M., Mazelle, C., Parks, G.K., R`eme, H., Lucek, E.: Journal of Geophysical Research (2008). 15. M¨obius, E., Kucharek, H., Mouikis, C., Georgescu, E., Kislter, L.M., Popecki, M.A., Scholer, J.M. M. Bosqued, R`eme, H., Carlson, C.W., Klecker, A. B. Korth, Parks, G.K., Sauvaud, J.A., Balsiger, M.B. H. Bavassano-Cattaneo, Dandouras, I., DiLellis, L. A. M. Eliasson, Formissano, V., Horbury, W. T. Lennartsson, Lundin, R., McCarthy, M., P., M.J., Paschmann, G.: Annales Geophysicae 19, 1411 (2001) 16. Moullard, O., Burgess, D., Horbury, T.S., Lucek, E.A.: Journal of Geophysical Research 111 (2006) 17. Paschmann, G., Sckopke, N., Asbridge, J.R., Bame, S.J., Gosling, J.T.: Journal of Geophysical Research 85, 4689 (1980) 18. Scholer, M., Matsukiyo, S.: Annales Geophysicae 22, 2345 (2004) 19. Slavin, J.H., Holzer, R.E.: Journal of Geophysical Research 86, 11401 (1981) ¨ Journal of Geophysical Research 74, 1301 (1969) 20. Sonnerup, B. U. O.: 21. Thomsen, M.F.: In: Stone, R.G., Tsurutani, B.T. (eds.) Collisionless Shocks in the Heliosphere: A Tutorial Review, p. 253 (1985). American Geophysical Union

Chapter 24

Wave Number Spectra in the Solar Wind, the Foreshock, and the Magnetosheath Y. Narita, K.-H. Glassmeier, S.P. Gary, M.L. Goldstein, and R.A. Treumann

Abstract The three-component model of magnetic field fluctuations is applied to the analysis of the wave number spectra to study fluctuations in the solar wind, the foreshock, and the magnetosheath. The analysis exhibits a transition of the dominant fluctuation component from the solar wind to the magnetosheath, from the twodimensional to the Alfv´enic in the foreshock, and to the compressible component in the magnetosheath.

24.1 Introduction One of the major tasks of Cluster is to determine spatial properties of space plasma dynamics. Here we focus on the possibility with four-point measurements in order to estimate the energy density of the magnetic field fluctuations in the wave number domain. Identification of fluctuations associated with parallel or perpendicular wave vectors plays an important role in studying turbulence properties. Previous works of the wave number spectrum analyses suggested the existence of turbulence in the foreshock and the magnetosheath [1, 2]. Here we present examples of wave number spectra of magnetic field fluctuations from the solar wind to the magnetosheath. The spectra are determined for the parallel and the perpendicular wave numbers using a three-component model of magnetic fluctuations. Y. Narita () and K.-H. Glassmeier Institute of Geophysics and Extraterrestrial Physics, Mendelssohnstr, 3, D-38106 Braunschweig, Germany e-mail: [email protected] S.P. Gary Los Alamos National Laboratory, Los Alamos, NM 87545, USA M.L. Goldstein Geospace Physics Laboratory, NASA Goddard Space Flight Center, Code 673, Greenbelt, MD 20771, USA R.A. Treumann Department of Geoscience, Geophysics Section, Ludwig-Maximilians-University Munich, Theresienstr, 41, D-80333 Munich, Germany

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 24, c Springer Science+Business Media B.V. 2010

363

364

Y. Narita et al.

Fig. 24.1 Magnetic field fluctuations and wave vectors in the three component model (after Narita et al. [4])

The three component model [3] gives three special kinds of magnetic field fluctuations that are physically relevant. This model is applied to the spectral analysis using Cluster data [4]. The three basis components are: a two-dimensional component, an Alfv´enic component, and a compressible component. The twodimensional component is a magnetic field fluctuation perpendicular to the background field associated with a perpendicular wave vector, where the fluctuation and the wave vector are mutually orthogonal. Physical interpretation of this component is a distortion of magnetic field lines without bending them [5]. The Alfv´enic component exhibits also a perpendicular fluctuation but it is associated with a parallel wave vector. The compressible component is a parallel fluctuation and associated with a perpendicular wave vector. Figure 24.1 displays the geometry of the three components.

24.2 Estimator of Wave Number Spectra The wave number spectra are evaluated using four-point Cluster magnetometer data [6]. We describe the spectral estimator for a one-dimensional, scalar field in this paper for simplicity. We estimate the amplitude of a magnetic field fluctuation characterized by the frequency ! and the wave number k by projecting the measurement state vector in the frequency domain S .!/ D ŒıB1 .!/; ; ıB4 .!/ using a suitable weight vector, b.!; k/ D W .!; k/ S .!/. Here the weight vector W .!; k/ is obtained by solving the problem of minimizing energy (squared amplitude) under a unit gain constraint, and determined by the measurement itself [7–9]. This analysis method assumes that the fluctuations are a set of plane waves, i.e. stationary and homogeneous fluctuations in a statistical sense. The squared amplitude jb.!; k/j2 is determined at various frequencies (up to the Nyquist value in the spacecraft frame) and at various wave numbers (up to the wave number of the sensor separation distance), and then it integrated over the frequency and transformed into the units of energy density in the wave number domain (squared amplitude per wave number) by dividing by the grid size of the wave number domain. The integration over frequency is made in the plasma rest frame, using the Doppler relation. The mean flow velocity from the ion data is used for the Doppler shift correction. The frequency integration is limited by the Nyquist value of the spacecraft frame

24

Measurements of Wave Number Spectra

365

frequency transformed into the plasma frame frequency. The estimator is expressed R as E.k/k D jb.!; k/j2 d!. The concept of the presented spectral estimator is generalized and applied to the three component model [4]. We determine E? .k? /, E? .kk /, and Ek .k? / for the two-dimensional component, the Alfv´enic component, and the compressible component, respectively, where E? and Ek denote the energy of perpendicular and parallel fluctuations.

24.3 Application to Cluster The spectral decomposition is applied to the Cluster magnetic field observations in the solar wind (Feb. 11, 2002, 1730–2030 UT), the foreshock (Feb. 12, 2002, 0630–1230 UT), and the magnetosheath (1615–2100 UT). Cluster crossed the quasiparallel shock during its orbit. Figures 24.2 and 24.3 display the wave number spectra and the relative energy contributions of the three components, respectively. The wave telescope estimator is a projection method and can be computationally applied to fluctuations with arbitrary wavelengths. For the estimate of the largest and the smallest wave numbers we use the spacecraft separation and the distance over which a disturbance propagates at Alfv´en speed within the observed time interval, respectively. The effect of tetrahedron deformation is not considered in the manuscript for simplicity, but it may affect the analysis result. Details of the analysis are presented in Narita et al. [4]. The wave number spectrum from the solar wind interval (Fig. 24.2, left) shows that most of the fluctuation energy is contributed by the two-dimensional component at the wave numbers sampled. The compressible component has the second largest contribution, and the Alfv´enic makes the weakest contribution. The twodimensional component exhibits a power law spectrum with the index is rather close to 5=3, within the error from the actual index 1:9, up to the spectral break at the

Fig. 24.2 Wave number spectra of the three components in the solar wind (left), the foreshock (middle), and the magnetosheath (right). The symbols T, A, and C stand for the two-dimensional, the Alfv´enic, and the compressible component, respectively. A confidence interval with 95% is shown in the plot

366

Y. Narita et al.

Fig. 24.3 Relative energy contribution of the three components

wave numbers about k D 5 103 km1 (marked with an arrow). The compressible and the Alfv´enic component have also similar spectral curves, but the spectrum of the Alfv´enic component is slightly steeper. Figure 24.3, left displays the relative energy contribution of the three components. Up to the spectral break scale (k 5 103 km1 ) the two-dimensional component occupies about 60% by energy, the Alfv´enic component about 15–20%, and the compressible component about 20–25%. Beyond the spectral break scale the contribution of the compressible component increases rather suddenly toward higher wave numbers, whereas that of the two-dimensional component becomes smaller. The wave number spectrum from the foreshock interval (Fig. 24.2, middle) is characterized by enhancement (or amplification) of the spectral power at all wave numbers. The spectrum also exhibits a hump on an intermediate scale (marked with an arrow) in the Alfv´enic component. The hump may be interpreted as the so-called “30 second waves” often observed in the foreshock, as the frequency spectrum (not shown here) exhibits a peak at period about 20–30 s. The enhancement of the spectra is primarily seen in the Alfv´enic and the compressible component, while the spectral power of the two-dimensional component does not change much. The enhancement of the Alfv´enic component is so large that it almost reaches the spectrum of the twodimensional component, especially in the range 103 km1 < k < 102 km1 . The enhancement of the compressible component is also prominent. All the three components exhibit a moderate hump at the wave numbers about k D 1 103 km1 . The relative energy contribution (Fig. 24.3, middle) exhibits the two-dimensional component contribution about 40–45%, the Alfv´enic component about 35–40%, and the compressible component about 20%. The wave number spectrum from the magnetosheath interval (Fig. 24.2, right) shows that all of the three components are enhanced by one to two orders of magnitude by comparison with the foreshock spectrum. The spectral lines are gently curved such that the energy decays more rapidly toward larger wave numbers. The relative energy contribution (Fig. 24.3, right) shows that the compressible component has the dominant contribution, about 45%, while the two-dimensional and the Alfv´enic component contribute about 35% and 20%, respectively.

24

Measurements of Wave Number Spectra

367

24.4 Discussion and Conclusion To summarize, the two-dimensional component is a major contributor to the fluctuation energy from the solar wind to the magnetosheath. The two-dimensional component clearly dominates in the solar wind and exhibits a power law spectrum with an index close to 5=3, which is consistent with earlier conclusions based on single spacecraft measurements [10–12]. The spectral index is close to 5=3 in the Alfv´enic and compressible components, too. There are transitions both in the spectral power (that is proportional to the squared amplitude) and in the relative energy contributions as the solar wind enters the magnetosheath. The three components are amplified first in the foreshock and then further amplified in the magnetosheath. The Alfv´enic component is prominently enhanced in the foreshock and the compressible component is most enhanced in the magnetosheath. The spectra in the foreshock and magnetosheath confirm the suggestions that the Alfv´enic and the mirror mode turbulence may exist in those regions, respectively [1, 2]. Two likely sources for the enhancements are (1) plasma instabilities and (2) amplification of fluctuations at the bow shock. In the foreshock, electron-beam and ion-beam modes are the most likely instabilities, because the bow shock is a source of heated electrons and reflected ions. The most likely mode to grow is the electromagnetic ion/ion right-hand resonant instability [13]. Linear theory [14] shows that this instability has maximum growth parallel to the background magnetic field, so that enhanced fluctuations from this instability have properties of the Alfv´enic component. Magnetosheath plasma shows the consequences of magnetic compression and heating at the shock, so that the temperature anisotropy with the perpendicular component dominant becomes enhanced. This anisotropy leads to the growth of both electromagnetic ion cyclotron and mirror-mode fluctuations, with the mirror instability often dominating the high-ˇ plasmas near and downstream of the shock. The mirror instability has maximum growth at directions strongly oblique to the background field and has a strong parallel component [14], so that enhanced fluctuations from this growing mode have properties of the compressible component. The second possibility is that the fluctuations in the solar wind are amplified at the shock independent of any instabilities. The interaction of the MHD waves with the shock wave can be analytically modeled and solved by perturbing the Rankine– Hugoniot relations to the first order. In general, the interaction results in reflection and transmission of the incident wave across the shock. The analysis suggests that the magnetic field amplitude of an Alfv´en wave incident in the shock-upstream region is enhanced by a factor unity or three, depending on the sense of wave propagation in the upstream and downstream region with respect to the shock normal direction. The analysis also predicts that the amplification of a fast magnetosonic wave is about a factor of four. Therefore we estimate naively the jump of the spectral power by factors 10–20 across the shock, which is interestingly close to our measurement. The analysis of wave number spectra combined with the three component model provide the following conclusions. (1) The two-dimensional component is a major

368

Y. Narita et al.

contributor to the fluctuation energy not only in the solar wind but also in the foreshock and the magnetosheath. (2) The two-dimensional component dominates in the solar wind, and its wave number spectrum is close to the 5=3 spectrum. (3) The transition from the solar wind to the magnetosheath is accompanied by two effects. One is the amplification of the fluctuations in all the three components, and the other effect is reconfiguration of the relative energy contribution of the three components. The foreshock and the magnetosheath exhibit more enhancement in the Alfv´enic and the compressible component, respectively. However, this analysis is a single case study for each region and hence it is a temperate conclusion and needs to be verified with a statistical study. Acknowledgements The work of YN and KHG in Braunschweig was supported by Bundesministerium f¨ur Wirtschaft und Technologie and Deutsches Zentrum f¨ur Luft- und Raumfahrt, Germany, under contract 50OC0103. We thank H. R`eme and I. Dandouras for providing ion data of Cluster. The work of SPG was supported by the Los Alamos National Laboratory LDRD Program and by the NASA Solar and Heliospheric SR&T Program.

References 1. Y. Narita, K.-H. Glassmeier et al.: Phys. Rev. Lett. 97, 191101 (2006) 2. F. Sahraoui, G. Belmont et al.: Phys. Rev. Lett. 96, 075002 (2006) 3. W. H. Matthaeus, S. Ghosh: In Solar Wind Nine CP 471, ed by S. R. Habbal, R. Esser et al. (American Institute of Physics 1999) pp 519–522 4. Y. Narita, K.-H. Glassmeier et al.: J. Geophys. Res., in press (2008) 5. D. Biskamp: Magnetohydrodynamic Turbulence, pp 161 (Cambridge, 2003) 6. A. Balogh, C. M. Carr et al.: Ann. Geophys. 19, 1207 (2001) 7. J. L. Pinc¸on, F. Lefeuvre, J. Geophys. Res. 96, 1789 (1991) 8. U. Motschmann, T. I. Woodward et al.: J. Geophys. Res. 101, 4961 (1996) 9. K.-H. Glassmeier, U. Motschmann et al.: Ann. Geophys. 19, 1439 (2001) 10. W. H. Matthaeus, M. L. Goldstein: J. Geophys. Res. 87, 6011 (1982) 11. W. H. Matthaeus, M. L. Goldstein et al.: J. Geophys. Res. 95, 20,673 (1990) 12. E. Marsch, C.-Y. Tu: J. Geophys. Res. 95, 8211 (1990) 13. Y. Narita, K.-H. Glassmeier et al.: Nonlin. Processes Geophys. 14, 361 (2007) 14. S. P. Gary: Theory of Space Plasma Microinstabilities (Cambridge, 1993)

Chapter 25

Cluster Hot Flow Anomaly Observations During Solar Cycle Minimum G. Facsk´o , M. T´atrallyay, G. Erd˝os, and I. Dandouras

Abstract Hot flow anomalies (HFAs) are studied using observations of the FGM magnetometer and the CIS plasma detector aboard the four Cluster spacecraft. Previously we studied several specific features of tangential discontinuities on the basis of Cluster measurements in February–April 2003 and discovered a new condition for forming HFAs that is the solar wind speed is higher than the average. However during the whole spring season of 2003, the solar wind speed was higher than average. In this study we analyse HFAs detected in 2007, the year of solar cycle minimum. Our earlier result was confirmed: the higher solar wind speed is a real condition for HFA formation; furthermore this constraint is independent of Schwartz et al.’s condition for HFA formation.

25.1 Introduction Hot flow anomalies were discovered in the 1980s [8, 15]. It is commonly agreed that HFAs are generated by the interaction of a tangential discontinuity and the bow shock. They are explosive events, particle acceleration, magnetic depletion take place at their center and the magnitude of the magnetic field increases at the rim. The plasma temperature increases whereas the density drops within the affected region, but the most interesting phenomenon is the directional change of the solar wind flow. The flow turns away from the anti-sunward direction and might even be directed backward [10, 11, 14, 15]. Both the previous global investigation [9] and our earlier studies also indicated that HFAs are not rare phenomena [3]. Cluster multispacecraft measurements help not only to understand the microphysics of HFA formation [5] but also to detect much more events than before because the spacecraft stay close

G. Facsk´o (), M. T´atrallyay, and G. Erd˝os KFKI Research Institute for Particle and Nuclear Physics, H-1525 Budapest, P.O. Box 49, Hungary e-mail: [email protected] I. Dandouras CESR, 9, Avenue du Colonel ROCHE, 31028 TOULOUSE CEDEX 4, France

´ Now at LPCE2, 3A, Avenue de la Recherche Scientifique, 45071 ORLEANS CEDEX 2, France

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 25, c Springer Science+Business Media B.V. 2010

369

370

G. Facsk´o et al.

to the bow shock for a long time; furthermore Cluster satellites have all necessary instruments to detect HFAs. We checked observationally a previous HFA simulation [4] determining the different angles using Cluster FGM and CIS HIA together with ACE MAG and SWEPAM measurements in the solar wind. In doing so, we discovered a new condition of HFA formation which was not predicted by the simulations; namely that the solar wind velocity is 200km/s higher than average when HFA develops [2]. Although a previous investigation in the magnetosheath [13] supports our result, however we want to check this condition in another year closer to the solar cycle minimum because during the whole spring season of 2003, the solar wind speed was higher than average. We chose 2007 because this is close to the solar minimum and the average solar wind velocity was expected to be lower than in 2003 when solar wind velocity was higher than the average value in the whole February–April season. We searched and analysed Cluster measurements in 2007 using the same methods like in [2] and compared them with previous results. The structure of this paper is the following: in Sect. 25.2 we discuss the processed and analysed measurements and in Sect. 25.3 we give the summary of our results.

25.2 Observations We set the same criteria for the selection of HFA events as in our previous papers [2,3]. Using these criteria we found more then 50 HFA events in January–April 2007 and in all cases we could determine the discontinuity normal from Cluster FGM and ACE MAG measurements. We identified the tangential discontinuities which generated HFA events between January 1 and April 30, 2007 using Cluster-1 and -3 FGM 1s average magnetic field and CIS 4s average solar wind speed measurements as well as MAG magnetic (16s) and SWEPAM (64s) plasma data from ACE. We applied both the minimum variance techniques [12] and the cross product method [7] to determine their normal vectors. We checked whether the normal component of the magnetic field disappears at the discontinuity. If its value approaches zero then the discontinuity was identified as tangential and we could use the cross product method to determine its normal vector. We accepted the result of minimum variance method if the angle between the minimum variance and the crossproduct vector was less than 15ı . We plotted all discontinuity normal determined using Cluster FGM and ACE MAG measurements. If a HFA was observed and its TD normal was calculated from more than one Cluster and the ACE spacecraft, the average vector was also plotted and the largest deviation from its direction was considered as the error and it was drawn by dashed–dotted line on Fig. 25.1. Normal belonging to the same event was connected by dashed line. Almost all the cone angles of the TD normals are larger than 45ı as predicted theoretically [4] and confirmed observationally [2,9]. We could not find TD-s neither in ACE nor in Cluster in the second part of March and in April (see: Fig. 25.3). We found many embedded HFAs in SLAMS in this interval and we could not determine the TD of those events.

25

Cluster Hot Flow Anomaly Observations During Solar Cycle Minimum

371

Fig. 25.1 Polar plot of the direction of the normal vectors of TDs. The azimuthal angle is measured between the GSE y direction and the projection of the normal vector onto the GSE yz plane. The distance from the center is the cone angle as determined by the cross-product method. The regions surrounded by dashed dotted lines are the projection of error cones around the average normal vector marked by “X”. Circles and squares symbolize ACE and Cluster data, respectively

We determined the solar wind velocity and fast-magnetosonic Mach-number distributions upstream of HFA formation and compared them to the same distributions in spring 2007 and for about 10 years (Fig. 25.2, Table 25.1). The average velocities seem to be lower than in 2003 but with error it is not so significant. Although Mach number average values seem to be higher in 2007 but its error is too high and makes this difference not so significant. This means that the solar wind velocity is higher when HFA develops but the difference is only 120 km/s in 2007. The difference measured in Mach numbers is greater [2], but the difference is within the error limit. In Fig. 25.2c the typical double peak distribution of solar wind velocity [1] can be seen for spring 2007. It helps us to understand why the solar wind distribution is anomalous during HFA formation on Fig. 25.2a. After comparing the positions of the maximum on Fig. 25.2b and Fig. 25.2d the difference in Mach numbers can be seen. We also plotted the 1h averaged solar wind speed time series from January to April, 2007 (Fig. 25.3). The last HFA event, on April 30, 2007 was not plotted, but it is evident that HFA events are coupled to high solar wind velocity intervals like in spring 2003.

372

G. Facsk´o et al.

a

b

c

d

Fig. 25.2 Upstream solar wind speed distribution measured by Cluster and ACE spacecraft. (a) Solar wind speed distribution measured by Cluster-1 (shifted grey dash dotted line) and Cluster-3 CIS HIA (shifted grey dash dotted line) ACE SWEPAM (solid line) upstream of HFA formation. (b) Fast magnetosonic Mach number distribution calculated using ACE MAG and SWEPAM data during HFA formation. (c) Solar wind speed distribution measured by ACE SWEPAM from January to April, 2007 (grey line) and from 1998 to 2008 (solid line). (d) Same as (c) but measured in Mf units

Table 25.1 Solar wind speed and fast magnetosonic values and their deviation measured by Cluster CIS The last column gives the figure numbers on Fig. 25.2 Solar wind speed (km/s) 2003 During HFA formation by C1 680 ˙ 86 by C3 671 ˙ 92 by ACE 666 ˙ 84 Mf numbers by ACE 8:2 ˙ 1:2 In 3/4 months period by ACE 546 ˙ 97 Between 1998 and 2003/2008 by ACE 492 ˙ 102 Mf numbers by ACE 5:5 ˙ 1:4 Mf 2:7

Mach numbers mean and ACE SWEPAM. 2007 613 ˙ 80 613 ˙ 78 634 ˙ 71 9:9 ˙ 1:1 512 ˙ 102 498 ˙ 101 6:2 ˙ 1:7 3:7

Figure 25.2a

25.2b 25.2c 25.2c 25.2d

Cluster Hot Flow Anomaly Observations During Solar Cycle Minimum

373

400

600

25

070108

070124

070201

070217

070225

600

070209

400

vSW(km/s)

400

600

070116

070313

070321

400

600

070305

070329

070406

070414

Fig. 25.3 The measured SW speed at the L1 point for four solar rotations. ACE 1-hr averages are from ACE Science Center. The times of the observed HFAs are marked by vertical lines

25.3 Discussion and Conclusions Our final conclusions based on measurements in spring 2007 are the following: 1. The solar wind velocity is about 120 km/s or Mf D 3:7 higher than the average when HFA develops. 2. HFA events occur almost always when the solar wind speed is higher than average. 3. The angle between the solar direction and the identified TD normal is usually greater than 45ı like in 2003. 4. Approximately another 40–50 HFA events were embedded in SLAMS, but the TD raising them could not be identified. Based on observations it is hard to say whether the higher solar wind velocity or Mach number is more important. Further independent theoretical considerations [6] suggest that both higher Mach number and higher solar wind speed are important in

374

a

G. Facsk´o et al.

b

Fig. 25.4 The distribution of the value of Schwartz el al.’s formula [9] by Cluster-1 and -3 measurements. The values are mostly less than 1.0

HFA formation. This theoretical work studied the orbit of particles accelerated by the bow-shock and led back to the shock by the convective electric field (v B) on both side of the tangential discontinuity. The particles are accelerated – and form a HFA – if the angle between the solar wind direction and the TD normal is less than the calculated critical angle. This critical angle depends on the ratio of the upstream and downstream magnetic field (which is a function of Mach number) and the solar wind speed. Furthermore this critical angle is 41:8ı calculated for typical Mach numbers and high solar wind speed so the theory confirms our observational results and the prediction of Lin’s computational work [4]. The above discussed analysis of HFA events in the spring 2007 season confirmed and extended our earlier results based on the study of HFA events in spring 2003 that higher solar wind speed is an important condition of HFA formation. This feature restricts Schwartz et al.’s formula [9] but these events also confirm their result: ˇ ˇ ˇ Vtr ˇ cos csWsw ˇ ˇD ; ˇV ˇ 2 cos bsWsw sin Bn sin csWbs g where Vt r is the transit velocity of the current sheet along the bow-shock, Vg is the gyration speed, csWsw , bsWsw and csWbs are the angles between the discontinuity normal, solar wind velocity and the bow-shock; furthermore Bn is the angle between the magnetic field and bow-shock normal. The necessary vectors were calculated and we got that the transition speed is as low as expected by Schwartz et al.’s formula [9] (Fig. 25.4). So, in addition to demonstrating that HFA formation depends on solar wind speed, this study confirms previous work predicting that HFA formation also depends on the geometry of the shock relative to the solar wind. Acknowledgements The authors thank the OTKA grant K75640 of the Hungarian Scientific Research Fund for support and the Cluster FGM, CIS and ACE MAG, SWEPAM teams for providing data for this study.

25

Cluster Hot Flow Anomaly Observations During Solar Cycle Minimum

375

References 1. G. Erd˝os and A. Balogh. In situ observations of magnetic field fluctuations. Advances in Space Research, 35:625–635, 2005. 2. G. Facsk´o, K. Kecskem´ety, G. Erd˝os, M. T´atrallyay, P. W. Daly, and I. Dandouras. A statistical study of hot flow anomalies using Cluster data. Advances in Space Research, 41:1286–1291, 2008. 3. K. Kecskem´ety, G. Erd˝os, G. Facsk´o, M. T´atrallyay, I. Dandouras, P. Daly, and K. Kudela. Distributions of suprathermal ions near hot flow anomalies observed by RAPID aboard Cluster. Advances in Space Research, 38:1587–1594, 2006. 4. Y. Lin. Global hybrid simulation of hot flow anomalies near the bow shock and in the magnetosheath. Planetary and Space Science, 50:577–591, 2002. 5. E. A. Lucek, T. S. Horbury, A. Balogh, I. Dandouras, and H. R`eme. Cluster observations of hot flow anomalies. Journal of Geophysical Research (Space Physics), 109(A6):207, 2004. 6. Z. N´emeth. Particle acceleration at the interaction of shocks and discontinuities. In Proceedings of 30th International Cosmic Ray Conference, 2007. 7. G. Paschmann and P. W. Daly, editors. Analysis Methods for Multi-Spacecraft Data, chapter Shock and Discontinuity Parameters, pages 249–270. ISSI/ESA, 1998. 8. S. J. Schwartz, C. P. Chaloner, D. S. Hall, P. J. Christiansen, and A. D. Johnstones. An active current sheet in the solar wind. Nature, 318:269–271, 1985. 9. S. J. Schwartz, G. Paschmann, N. Sckopke, T. M. Bauer, M. Dunlop, A. N. Fazakerley, and M. F. Thomsen. Conditions for the formation of hot flow anomalies at Earth’s bow shock. Journal of Geophysical Research, 105:12639–12650, 2000. 10. D.G. Sibeck, N.L. Borodkova, S.J. Schwartz, C.J. Owen, R. Kessel, S. Kokubun, R.P. Lepping, R. Lin, K. Liou, H. L¨uhr, R.W. McEntire, C.-I. Meng, T. Mukai, Z. Nemecek, G. Parks, T.D. Phan, S.A. Romanov, J. Safrankova, J.-A. Sauvaud, H.J. Singer, S.I. Solovyev, A. Szabo, K. Takahashi, D.J. Williams, K. Yumoto, and G.N. Zastenker. Comprehensive study of the magnetospheric response to a hot flow anomaly. Journal of Geophysical Research, 104:4577–4594, 1999. 11. D. G. Sibeck, T.-D. Phan, R. Lin, R. P. Lepping, and A. Szabo. Wind observations of foreshock cavities: A case study. Journal of Geophysical Research (Space Physics), 107:4–1, 2002. ¨ Sonnerup and M. Scheible. Minimum and maximum variance analysis. In 12. B. U. O. G. Paschmann and S. J. Schwartz, editors, ISSI Book on Analysis Methods for Multi-Spacecraft Data, pages 185–220. ISSI, 1998. ˇ ankov´a, L. Pˇrech, Z. Nˇemeˇcek, D. G. Sibeck, and T. Mukai. Magnetosheath response 13. J. Safr´ to the interplanetary magnetic field tangential discontinuity. Journal of Geophysical Research, 105:25113–25122, 2000. 14. M. F. Thomsen, , V. A. Thomas, D. Winske, J. T. Gosling, M. H. Farris, and C. T. Russell. Observational test of hot flow abnomaly formation by the interaction of s magnetic discontinuity with the bow shock. Journal of Geophysical Research, 98(A9):15319–15330, 1993. 15. M. F. Thomsen, J. T. Gosling, S. A. Fuselier, S. J. Bame, and C. T. Russell. Hot, diamagnetic cavities upstream from the earth’s bow shock. Journal of Geophysical Research, 91:2961–2973, 1986.

Chapter 26

On the Growth of Mirror Mode Waves in the Magnetosheath Based on Cluster Observations M. T´atrallyay, G. Erd˝os, I. Dandouras, and E. Georgescu

Abstract The evolution of mirror mode waves was investigated in the magnetosheath based on Cluster magnetic field and plasma measurements during four inbound passes in January–February 2006 when the separation between the spacecraft was about 10,000 km. The growth rates estimated from the amplitudes of the field fluctuations (mainly quasi-sinusoidal oscillations and peaks) observed simultaneously at distant locations were usually smaller than the maximum growth rates calculated from the plasma parameters measured in the magnetosheath suggesting that mirror mode waves do not grow beyond a certain degree, they seem to get saturated. Sudden changes in interplanetary conditions influenced the evolution of mirror mode waves, which were observed when solar wind dynamic pressure and the IMF were relatively quiet and the bow shock region connected to the observation point was quasi-perpendicular, while temperature anisotropy was moderate and ion “ was high 1:2 < T? =Tjj < 1:6 and “i > 1 in the local plasma.

26.1 Introduction Mirror mode waves are compressive, non-propagating low frequency magnetic field fluctuations in which the field magnitude is anti-correlated with the plasma pressure. They are linearly polarized with a maximum variance direction which typically lies at about 10ı –20ı to the mean magnetic field direction [3]. Mirror mode waves are generated by the mirror instability [6] developing in high “ plasmas when the ion temperature is anisotropic and the proton-electron plasma contains a small percentage of HeCC ions [4,5] corresponding to the composition of the solar wind. Beyond the terrestrial magnetosheath (cf. Tsurutani et al. [20], Lucek et al. [12]), mirror M. T´atrallyay () and G. Erd˝os KFKI Research Institute for Particle and Nuclear Physics, 1525 Budapest, P.O. Box 49, Hungary e-mail: [email protected] I. Dandouras CESR, BP 4346, 31028 Toulouse, C´edex 4, France E. Georgescu MPS, Max-Planck-Str. 2. 37191 Katlenburg-Lindau, Germany H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 26, c Springer Science+Business Media B.V. 2010

377

378

M. T´atrallyay et al.

mode fluctuations were most frequently observed in the magnetosheath of the outer planets ([3, 15, 20, 22] and a number of later studies), but recently they were also identified in the induced magnetosheath of Venus [23]. Mirror mode waves were encountered in the coma of comets [14, 21], in the interplanetary field [24], and also in the heliosheath [2]. These observations confirm that the actual parameters of different space plasmas often favour the development of the mirror instability. T´atrallyay and Erd˝os [18] investigated the evolution of mirror type fluctuations in the terrestrial magnetosheath based on magnetic field data measured aboard the ISEE-1 spacecraft. Fluctuations were observed in different regions of the sheath with increasing amplitudes from the bow shock towards the magnetopause. The growth rate of the instability was determined assuming that the fluctuations were originating at the bow shock and the field strength perturbations were growing exponentially while traveling frozen in the plasma flow. The compared amplitudes were observed at different locations and at different times during the magnetosheath pass of one satellite. T´atrallyay et al. [19] revisited the evolution of mirror mode waves in the magnetosheath comparing the amplitudes of linearly polarized magnetic field fluctuations observed simultaneously aboard Cluster-2 and Cluster-4 at distant locations (10;000 km) during two inbound passes. The growth rate values obtained by both studies were about an order of magnitude smaller than those calculated from a numerical evaluation of the full kinetic dispersion relation by Gary et al. [5]. Plausible explanations were given for the significant difference between the observed and calculated growth rate values: (1) the growth of mirror type fluctuations cannot be described by linear approximations in the magnetosheath; (2) the waves are not always originating at the bow shock, there may be other sources inside the magnetosheath. In this paper, the evolution of mirror type magnetic fluctuations is further investigated in the magnetosheath involving also plasma data. During the four selected inbound passes, the magnetic field variations were mainly quasi-sinusoidal oscillations or peaks, except for the regions close to the magnetopause. T´atrallyay et al. [19] used only magnetic field data simultaneously measured by the Cluster spacecraft. In the present study, local plasma parameters measured by the Cluster Ion Spectrometer are also analyzed and the influence of changes in the interplanetary magnetic field and solar wind is also taken into account.

26.2 Data Analysis During the winter/spring season in 2006, the separation of the Cluster spacecraft was about 10,000 km when they passed through the magnetosheath, so that fluctuations measured simultaneously at significantly different distances from the magnetopause could be compared. Supposing that the field variations were originating from the same source, their evolution between two distant observation points could be investigated. Mirror mode fluctuations were identified from 1 s resolution magnetic field time series measured by the Fluxgate Magnetometer [1]. Applying variance analysis,

26

On the Growth of Mirror Mode Waves in the Magnetosheath

379

several orbits were selected when magnetic field fluctuations were linearly polarized (characteristic of mirror mode waves) for several hours in different regions of the magnetosheath. Two criteria were used for the selection: (1) the angle between the maximum variance direction and the average field vector is smaller than 20ı ; (2) the ratio of the maximum eigenvalue to the intermediate eigenvalue of the variance matrix is relatively large, i.e. œmax =œint > 5. The length of the sliding window applied for the variance analysis was 1 min, while the typical period of the mirror type fluctuations was 10–25 s. In order to have the possible largest separation between two spacecraft and also reliable local plasma data, mirror type fluctuations observed by Cluster-3 and Cluster-4 are compared in this study. Four seconds resolution plasma parameters measured by the Hot Ion Analyzer of the Cluster Ion Spectrometer [13] aboard Cluster-3 were used. CIS-HIA did not provide data for Cluster-2 and Cluster-4. The following parameters were determined from ion densities and temperatures parallel Tjj and perpendicular T? to the magnetic field: the correlation between plasma and magnetic field pressure; plasma “i , i.e. the ratio of ion pressure to magnetic field pressure; and the mirror instability criterion CM D “? T? =Tjj 1 as given by Hasegawa [6]. In mirror mode waves, plasma pressure is anti-correlated with magnetic pressure and the plasma beta is high .“i > 1/. Earlier Hasegawa [6] argued that mirror mode waves grow when the plasma is in unstable condition i.e. CM > 1. However, recent investigations suggested that the value of CM controls the shape of the mirror mode variations. Magnetic peaks develop when the plasma is unstable while dips survive in linearly stable plasma when CM < 1 (cf. Soucek et al. [17]). The influence of interplanetary effects on the occurrence of mirror mode fluctuations was investigated using the 1 min resolution OMNI data set providing magnetic field and plasma parameters upstream of the nose of the terrestrial bow shock (http://nssdc.gsfc.nasa.gov/omniweb/html/omsc min.html). Variations in solar wind dynamic pressure and in the magnitude and direction of the Interplanetary Magnetic Field were analyzed while the flow-time from the bow shock to the observation point of the mirror wave in the magnetosheath was taken into account. The angle between the IMF and the bow shock normal BN was determined in the bow shock region which was connected to the location of Cluster-3 by the plasma flow-line. Solar wind plasma flow-lines in the magnetosheath and flow-times along the flowlines were determined using the model of Kobel and Fl¨uckiger [9] as described by T´atrallyay et al. [19].

26.3 The Influence of Plasma Parameters on the Occurrence of Mirror Mode Waves T´atrallyay et al. [19] discussed the growth of mirror type magnetic field fluctuations during two magnetosheath passes based only on magnetic field measurements observed by Cluster-2 and Cluster-4. On 7–8 February 2006, solar wind dynamic pressure and the interplanetary magnetic field were undisturbed, the IMF stayed

380

M. T´atrallyay et al.

quasi-perpendicular to the bow shock normal .BN > 60ı / during the whole pass. Mirror mode magnetic fluctuations (mainly peaks) were observed by all satellites throughout the middle and inner magnetosheath where “i > 10 and CM Š 2 was measured. The amplitude of the waves was almost continuously growing while traveling inbound. During the inbound pass on 3 January 2006, however, there was a sudden change in the growth of the mirror type magnetic fluctuations at about 5:00 UT when the direction of the IMF changed significantly as shown in Fig. 26.1. Mirror waves did not occur downstream of the quasi-parallel bow shock .BN < 45ı / after 05:00 UT. Later (closer to the magnetopause where CM Š 1 and “i Š 1, i.e. both parameters decreased compared to the earlier measured values), Cluster-3 observed mirror mode fluctuations less frequently than Cluster-4 which was farther away from

Fig. 26.1 Inbound magnetosheath pass on 3 January 2006. Upper three panels: 1 min averaged plasma parameters calculated from data measured by CIS-HIA aboard Cluster-3: plasma “i ; correlation between plasma and magnetic pressure; mirror criterion CM D “? .T? =Tjj 1/, CM D 1 critical value is marked; panel 4 and bottom: 4 s resolution Cluster-3 and Cluster-4 magnetic field, vertical bars above the time axis mark the minutes when linearly polarized variations occur; panels 5–7: interplanetary parameters from the 1 min resolution OMNI data set: BN at the bow shock region connected by plasma flow-line, BN D 45ı is marked; IMF total value; solar wind dynamic pressure

26

On the Growth of Mirror Mode Waves in the Magnetosheath

381

the magnetopause. Earlier, between 03:15 and 04:00 UT when there were changes in the magnitude of the IMF, mirror mode waves were observed rarely by both satellites. During the time interval presented by Fig. 26.1, there was no significant change in solar wind dynamic pressure. Two additional inbound passes were also investigated in details: 22 January and 10 February 2006. In both cases, there were sudden changes in solar wind dynamic pressure (usually together with changes in the IMF) moving the magnetopause inwards or outwards thus influencing plasma and magnetic field parameters in the magnetosheath. When the magnetopause moved closer to Cluster-3 in the inner sheath, this satellite did not observe mirror mode fluctuations any more since CM and “i parameters decreased below their critical values in the vicinity of the magnetopause. Cluster-4 which was farther out in the magnetosheath observed mirror mode fluctuations later when the variations in the interplanetary parameters became smoother. Also, the direction and/or magnitude of the IMF changed several times during both magnetosheath passes influencing the growth of the waves. No mirror mode waves were observed when the IMF was quasi-parallel to the bow shock normal in the region connected to the observation point.

26.4 Growth Rate of Fluctuations The growth rate of the fluctuations can be estimated by assuming that mirror waves are growing as •B exp. t/ while flowing frozen in the plasma along the streamlines where > 0 is the growth rate [18,19]. The value B presented in Fig. 26.2 was calculated from relation •B3 =•B4 D exp .B Œt3 –t4 / where •B3 and •B4 were the amplitudes of the fluctuations simultaneously measured by Cluster-3 and Cluster-4 at distant locations. The flow-times (from the source along the flow-line to the observation point) t3 and t4 were calculated from the model of Kobel and Fl¨uckiger [9] for the middle of the selected 20 min time intervals of the above discussed four inbound passes. Most of the time, the shape of the magnetic variations was quasi-sinusoidal

Fig. 26.2 The relation between growth rates B determined from simultaneously measured magnetic field variations aboard Cluster-3 and Cluster-4 and maximum growth rates C calculated from plasma parameters measured by CIS-HIA aboard Cluster-3 in 20 min intervals on 3 Jan, 22 Jan, 7/8 Feb, and 10 Feb 2006. Averaged amplitudes •B3 and •B4 were determined from individually selected mirror type variations (gray symbols) and from the maximum eigenvalues of the variance matrix (black symbols)

382

M. T´atrallyay et al.

or peak type since the value of the mirror instability criterion was well above 1, often CM > 2 was measured. Magnetic dips were observed only in the inner regions of the magnetosheath where CM was below the critical value or close to it corresponding to the results of Soucek et al. [17]. The averaged amplitudes •B3 and •B4 were determined in two ways: (1) from the difference between the magnetic field maximum and minimum values of individually selected mirror type variations when the change in the field direction was smaller than 20ı (similar conditions were used by Tatrallyay et al. [19]) – shown by gray symbols in Fig. 26.2; (2) from the maximum eigenvalues of the variance matrix for 1 min intervals applying the conditions discussed in Section 26.2 – shown by black symbols in Fig. 26.2. The B values determined from individual events are in good agreement with those calculated using the variance method. Based on the dispersion relation of Hasegawa [6], Liu et al. [11] analytically derived an expression for the maximum growth rate of mirror mode waves: C D 3=2 .3= /1=2 .6“? /1 T? =Tjj .CM 1/2 p for the conditions kjj k? in the long wavelength limit, where CM D “? T? =Tjj 1 > 1 is the mirror instability criterion, k is the wave number, and p is the proton gyro-frequency. This formula provides significant growth rates only in high “jj regions when the thermal anisotropy is moderate in agreement with observations (cf. Lacombe and Belmont [10], Samsonov et al. [16]) and with the results of hybrid simulation [7]: proton cyclotron waves dominate when “jj is lower and T? =T jj is well above 1, while mirror mode waves grow when the plasma beta is higher “jj > 1 and T? =Tjj is slightly over 1. For the calculation of C values presented in Fig. 26.2, ion densities and temperatures measured by CIS HIA aboard Cluster-3 were used. The parameters were averaged for 20 min intervals. Figure 26.2 illustrates that B growth rates estimated from the amplitudes of magnetic fluctuations are comparable with C maximum growth rates calculated from the local plasma parameters when the latter values are by “? when the anisotropy is moderate small. The value of C is mainly influenced as measured 1:2 < T? =Tjj < 1:6 during the selected events. The value of “? was 1–3 on 3 Jan providing smaller C values, while it was 10–20 on 7/8 Feb resulting in larger C . On 22 Jan and 10 Feb, the value of “? was changing in a wider range (mainly caused by significant changes in the IMF), therefore the dispersion in the C values was relatively large. According to Gary et al. [5], the numerical evaluation of the linear Vlasov dispersion equations provided the value of max 0:02 p for the maximum growth rate of the mirror instability when T? =Tjj D 1:5, “ D 4, and the plasma contained 4–10% HeCC ions. Hubert et al. [8] obtained very similar max values for plasmas of slightly different parameters using a linear model. Since p > 1:5 s1 during the selected magnetosheath passes, it is evident that both B and C growth rate values obtained in this study are smaller than the maximum growth rates predicted by these linear models. Samsonov et al. [16] compared the proton temperature anisotropy measured by Cluster in the magnetosheath with T? =Tjj values predicted by a 3D double adiabatic MHD model. According to their results, the maximum growth rate of the proton cyclotron and mirror instabilities could be equal in the high “

26

On the Growth of Mirror Mode Waves in the Magnetosheath

383

regions varying in the range 104 < max =p < 2:5 102 . The values B and C determined in the present study fit well in the range of possible growth rates provided by Samsonov et al. [16]. It is obvious that linear models can provide a good approximation for the growth of mirror mode waves only around the onset of the instability. Later in the process, growing waves scatter ions, reduce the temperature anisotropy, and the growth rate becomes lower [5]. Therefore the amplitudes of the waves cannot grow beyond a certain degree, they get saturated as they are continuously responding to the slow changes in plasma conditions. Beside saturation, there may be other reasons for the significant difference between the growth rates determined here from the amplitudes of magnetic field variations and the growth rates provided by other models. The assumptions used here have some limitations, namely mirror mode waves simultaneously observed by the two distant satellites may not always originate from the same source as supposed in our study. Different local sources may be active at different times in the magnetosheath which may be located between the two observation points. However, when interplanetary conditions are quiet, local disturbances are less likely and changes in the magnetosheath plasma are expected to be smooth.

26.5 Summary In this study, four orbits were selected when the separation between the Cluster spacecraft was large and mirror mode fluctuations (mainly quasi-sinusoidal or peak type oscillations) were observed by all spacecraft in different regions of the terrestrial magnetosheath. The growth rates B D 0:001–0:008 s1 as estimated from the amplitudes of the magnetic field fluctuations observed simultaneously at distant locations were usually smaller than the maximum growth rates C D 0:001 0:03 s1 provided by the analytical formula of Liu et al. [11] using the locally measured plasma parameters. Other linear models [5, 8], which are valid when the amplitudes are small, obviously also predicted larger growth rates max Š 0:03 0:06 s1 for plasma and field parameters similar to those observed during the selected passes. The wide range for maximum growth rates 104 < max < 102 provided by Samsonov et al. [16] covers the B growth rate values determined in this study. According to the presented results, mirror mode waves do not grow beyond a certain degree, they seem to get saturated since they are continuously responding to any change in local plasma conditions. The effect of interplanetary parameters on the evolution of mirror mode waves was also investigated. Abrupt changes in solar wind dynamic pressure (usually occurring together with a sudden change in the IMF) move the magnetopause in or out influencing mainly theinnermost region of the magnetosheath. When the ion “ drops below a critical value “jj < 1 and the anisotropy increases above T? =Tjj > 1:7 after a sudden change in interplanetary conditions, mirror waves cannot be observed in the magnetosheath since the relation between “ and the temperature anisotropy does not favour the development of mirror mode waves [7, 10, 16]. Later, when the

384

M. T´atrallyay et al.

interplanetary activity gets quiet again, mirror mode waves can be observed mainly in the middle or outer regions of the magnetosheath, except for a few dips close to the magnetopause in the mirror stable plasma regions. An essential change in the direction of the IMF (with no significant variation in any other interplanetary parameter) may also influence the occurrence of mirror mode waves. When the interplanetary field gets quasi-parallel in the bow shock region connected by plasma flow-lines to the location of the satellite, no mirror mode waves are observed there. This result confirms the earlier suggestion of T´atrallyay et al. [19] that magnetosheath plasma connected to the compression region of quasiperpendicular bow shocks favour the growth of mirror mode waves.

References 1. Balogh, A., Carr, C.M., Acuna, M.H., et al.: The Cluster Magnetic Field Investigation: overview of in-flight performance and initial results. Ann Geophys. 19, 1207–1217 (2001) 2. Burlaga, L.F., Ness, N.F., Acuna, M.H.: Trains of magnetic holes and magnetic humps in the heliosheath. Geophys. Res. Lett. 33, L21106 (2006) doi:10.1029/2006GL027276 3. Erd˝os, G., Balogh, A.: Statistical properties of mirror mode structures observed by Ulysses in the magnetosheath of Jupiter. J. Geophys. Res. 101, 1–12 (1996) 4. Gary, S.P.: The mirror and ion cyclotron anisotropy instabilities. J. Geophys. Res. 97, 8519–8529 (1992) 5. Gary, S.P., Fuselier, S.A., Anderson, B.J.: Ion anisotropy instabilities in the magnetosheath. J. Geophys. Res. 98, 1481–1488 (1993) 6. Hasegawa, A.: Drift mirror instability in the magnetosphere. Phys. Fluids 12, 2642–2650 (1969) 7. Hellinger, P., Travnicek, P., Mangeney, A., Grappin, R.: Hybrid simulations of the magnetosheath compression: Marginal stability path. Geophys. Res. Lett. 30(18) (2003), doi:10.1029/2003GL017855 8. Hubert, D., Lacombe C., Harvey, C.C, et al.: Nature, properties, and origin of low-frequency waves from an oblique shock to the inner magnetosheath. J. Geophys. Res. 103, 26783–26798 (1998) 9. Kobel, E., Fl¨uckiger, E.O.: A model of the steady state field in the magnetosheath. J. Geophys. Res. 99, 23617–23622 (1993) 10. Lacombe, C. and Belmont, G.: Waves in the Earth’s magnetosheath: observations and interpretations. Adv. Space Res. 15(8/9), 329–340 (1995) 11. Liu, Y., Richardson, J.D., Belcher, J.W., Kasper, J.C., Skoug, R.M.: Plasma depletion and mirror waves ahead of interplanetary coronal mass ejections. J. Geophys. Res. 111, A09108 (2006), doi:10.1029/2006JA011723 12. Lucek, E.A., Dunlop, M.W., Balogh, A., et al.: Mirror mode structures observed in the dawnside magnetosheath by Equator-S. Geophys. Res. Lett. 26, 2159–2162 (1999) 13. Reme, H., Aoustin, C., Bosqued, J.M., et al.: First multispacecraft ion measurements in and near the Earth’s magnetosphere with the identical Cluster ion spectrometry (CIS) experiment. Annales Geophys. 19, 1303–1354 (2001) 14. Russell, C.T., Riedler, W., Schwingenschuh, K., Yeroshenko, Ye.: Mirror instability in the magnetosphere of comet Halley. Geophys. Res. Lett. 14, 644–647 (1987) 15. Russell, C.T., Song, P., Lepping, R.P.: The Uranian magnetopause: Lessons from Earth. Geophys. Res. Lett. 16, 1485–1488 (1989) 16. Samsonov, A.A., Alexandrova, O., Lacombe, C., et al.: Proton temperature anisotropy in the magnetosheath: comparison of 3-D MHD modelling with Cluster data. Ann Geophys. 25, 1157–1173 (2007)

26

On the Growth of Mirror Mode Waves in the Magnetosheath

385

17. Soucek, J., Lucek, E., Dandouras, I.: Properties of magnetosheath mirror modes observed by Cluster and their response to changes in plasma parameters. J. Geophys. Res. 113, A04203 (2008), doi: 10.1029/2007JA012649 18. T´atrallyay, M. and Erd˝os, G.: The evolution of mirror mode fluctuations in the terrestrial magnetosheath. Planet. Space Sci. 50, 593–599 (2002) 19. T´atrallyay, M., Erd˝os, G., Balogh, A., Dandouras, I.: The evolution of mirror type magnetic fluctuations in the magnetosheath based on multipoint observations. Adv. Space Res. 41, 1537–1544 (2008) 20. Tsurutani, B.T., Smith, E.J., Anderson, R.R., et al.: Lion roars and nonoscillatory drift mirror waves in the magnetosheath. J. Geophys. Res. 87, 6060–6072 (1982) 21. Tsurutani, B.T., Lakhina, G.S., Smith, E.J., et al.: Mirror mode structures and ELF plasma waves in the Giacobini-Zinner magnetosheath. Nonlin. Proc. in Geophys. 6, 229–234 (1999) 22. Violante, L., Bavassano Cattaneo, M.B., Moreno, G., Richardson, J.D.: Observations of mirror waves and plasma depletion layer upstream of Saturn’s magnetopause. J. Geopys. Res. 100, 12047–12055 (1995) 23. Volwerk, M., Zhang, T.L., Delva, M., et al.: First identification of mirror mode waves in Venus’ magnetosheath? Geophys. Res. Lett. 35, L12204 (2008) doi: 10.1029/2008GL033621 24. Winterhalter, D., Neugebauer, M., Goldstein, B.E., et al.: Ulysses field and plasma observations of magnetic holes in the solar wind and their relation to mirror mode structures, J. Geophys. Res. 99, 23371–23382 (1994)

Part VI

Observations of Magnetopause and Cusp

Chapter 27

Mixed Azimuthal Scales of Flux Transfer Events R.C. Fear, S.E. Milan, E.A. Lucek, S.W.H. Cowley, and A.N. Fazakerley

Abstract Previous observations have allowed the scale size of flux transfer events (FTEs) to be determined both normal to the magnetopause and in the direction of motion of the FTE, but a key difference between some different models of FTE structure is their azimuthal scale size. Previous ground-based observations of the ionospheric signatures of FTEs indicated that magnetic reconnection can occur coherently over large extents of the magnetopause, but in situ determination of the azimuthal scale size of FTEs has not been possible until recent Cluster magnetopause crossing seasons when the separation of the spacecraft was 10;000 km. In this paper, we present Cluster observations of flux transfer events from the 27th March 2007, along with observations of the conjugate ionospheric signatures. We highlight two magnetospheric FTEs which were consistent with long X-line FTE models, but note also several FTEs with considerably smaller azimuthal scale.

27.1 Introduction Flux transfer events [15] are the time-varying embodiment of magnetopause reconnection [3]. Various attempts have been made in the past to quantify how much flux an FTE transfers into the magnetospheric system [11, 14–16], but these estimates are dependant on an assumption about the structure that an FTE takes. The original structure proposed was of an azimuthally narrow, elbow-shaped flux tube that crossed the magnetopause through an approximately circular hole [15], but subsequently other reconnection-based models were proposed which could extend significantly further azimuthally [10, 17, 20], and therefore could each reconnect more flux. R.C. Fear (), S.E. Milan, and S.W.H. Cowley University of Leicester, UK e-mail: [email protected] E.A. Lucek Imperial College London, UK A.N. Fazakerley Mullard Space Science Laboratory, University College London, UK

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 27, c Springer Science+Business Media B.V. 2010

389

390

R.C. Fear et al.

Therefore, a key difference between some of these models is the azimuthal extent they can allow in comparison with their poleward extent. Since the scale of an FTE both normal to the magnetopause and in its direction of motion (poleward under strongly southward IMF) is typically of order 1–2 RE [4, 12, 14–16], observations from the Cluster spacecraft in their 10;000 km separation magnetopause crossing seasons (2006 onwards) can provide valuable information about whether the azimuthal scale of FTEs is comparable to or larger than their poleward extent. The first such FTE observations have recently been presented [5], but the magnetic shear across the magnetopause was comparatively low .110ı /. Whilst it was shown that the flux transfer events observed had a larger azimuthal extent than their poleward extent, the elbow-shaped flux rope model [15] could not be excluded completely. In this paper, we present a separate case study of flux transfer events observed by Cluster on the 27th March 2007 and their ionospheric signatures. At this time, the magnetic shear across the magnetopause was much closer to 180ı, which allows a clearer distinction between some of the models. Magnetic field data are provided at 5 Hz resolution from the Cluster FGM instrument [1], electron data from the Cluster PEACE instrument [7], and ionospheric observations from the SuperDARN radars [2].

27.2 Cluster Orientation and Solar Wind Conditions Figure 27.1 illustrates the location and orientation of the Cluster spacecraft (the tetrahedron has been expanded by a factor of two relative to the location of Cluster 1). In this and all subsequent figures, the Cluster spacecraft are denoted by a consistent colour scale which is shown in the legend. The model magnetopause (dashed line) and bow shock (dash-dot) are derived from the models by Shue et al. [18] and Peredo et al. [13], using upstream conditions as model inputs. The spacecraft were at

20

–20

–10 YGSE (RE)

ZGSE (RE)

10

0

–10

–20 20

C1 C2 C3 C4

0

10

10

0 XGSE (RE)

–10

–20

20 20

10

0 XGSE (RE)

–10

–20

Fig. 27.1 The location and orientation of the Cluster spacecraft tetrahedron at 04:50 UT on the 27th March 2007. The tetrahedron has been expanded by a factor of two relative to Cluster 1

27

Mixed Azimuthal Scales of Flux Transfer Events

391

Fig. 27.2 OMNI lagged solar wind conditions. From top: the IMF GSM components and magnitude, the magnitude of the IMF clock angle (expressed between 0ı and 180ı ) and the solar wind dynamic pressure

low latitudes in the northern hemisphere, near local noon. Typically for early 2007, the spacecraft were in a flat triangular formation, rather than a good quality tetrahedron, with the surface of the triangle being contained approximately in a plane tangential to the magnetopause surface. The solar wind conditions are shown in Fig. 27.2, which contains propagated observations from the OMNI database. These data were taken by the Wind spacecraft, which was situated 198 RE upstream of the Earth, and they have been propagated to the Earth’s bow shock using a variable lag time [9]. At 0440 UT, the interplanetary magnetic field (IMF) rotated southward, and the clock angle remained close to 180ı until 0550 UT, except for a brief downward rotation between 0510 and 0520 UT. We shall present the Cluster data in a boundary normal coordinate system [15], which was derived as follows. The unit vector normal to the magnetopause was derived by carrying out minimum variance analysis [19] on the Cluster magnetic field data between 0500 and 0512 UT (the Cluster spacecraft all crossed the magnetopause at around 0506 UT). The largest ratio between the intermediate and minimum eigenvalues was observed using Cluster 4 data .œint =œmin D 3:74/, so we adopted the minimum eigenvector from the magnetopause crossing made by Cluster 4 as the normal vector Œn D .0:938; 0:044; 0:342/GSE . The azimuthal vector was derived by normalising the cross product of n with the Earth’s dipole vector

392

R.C. Fear et al.

Fig. 27.3 The orientation of the Cluster tetrahedron, relative to Cluster 3, in boundary normal coordinates at 05:15 UT on the 27th March 2007

Œm D .0:174; 0:917; 0:359/GSE , and the poleward vector completes the righthanded set Œl D m n D .0:298; 0:397; 0:868/GSE . The coordinates n, m and l therefore correspond closely to the GSE x, y and z vectors, as is expected for a low-latitude, near-noon magnetopause crossing. The Cluster spacecraft separation is shown in these boundary normal coordinates in Fig. 27.3. The spacecraft formed a roughly equilateral triangle in the l-m plane, with each side of order 8,000 km. The maximum separation of the spacecraft normal to the nominal magnetopause was 1,200 km.

27.3 Extent of Dayside Reconnection: Ionospheric Signatures We first investigate the ground-based radar observations of the ionospheric signatures of FTEs to show the azimuthal extent over which magnetopause reconnection was taking place. Ionospheric observations from the Hankasalmi and TIGER coherent scatter radars (part of the SuperDARN network) are shown in Fig. 27.4. The Hankasalmi radar is situated in Finland, and has a field of view that is directed towards magnetic north. TIGER is situated in Tasmania, and its field of view is directed towards magnetic south. The top two panels show a snapshot of the ionospheric velocities from the entire fields of view of these two radars at 0508 UT; each panel is a polar grid of magnetic local time against magnetic latitude with 12 h MLT directed towards the top, and 06 h MLT directed towards the right. Both radars observed strong flows away from the radars (shaded red) in the beams that are directed poleward. TIGER also observed more azimuthal flows toward the radar (blue). Both are consistent with a twin-cell convection pattern. Time series of these observations are shown in the bottom half of Fig. 27.4. The top two panels show the ionospheric velocities against magnetic latitude (and time). Each panel shows data from one radar beam (beam 15 from Hankasalmi and beam 0 from TIGER). The bottom two panels show the corresponding backscatter power data. Three patches of enhanced backscatter power were observed by each radar, which propagated poleward, and which corresponded to enhanced ionospheric flows. These Poleward Moving Radar Auroral Forms (PMRAFs) are the ionospheric signature of time-varying magnetopause reconnection (Milan et al. [11]

27

Mixed Azimuthal Scales of Flux Transfer Events

0508 00s (086)

393

0508 00s (086)

11.235 MHz

10.210 MHz 800 600

200 0 –200

Velocity (ms–1)

400

–400 –600 –800 Ground Scatter

Magnetic Latitude Magnetic Latitude

80 79 78 77 76 75

Hankasalmi

Beam 15

0450 (086) to 0530 (086)

800 600 400

TIGER

Beam 0

200

0450 (086) to 0530 (086)

0

–200 –400 –600 –800

Ground Scatter

Hankasalmi

Beam 15

0450 (086) to 0530 (086)

27 24 21

–80 –79 –78 –77 –76 –75 –74

TIGER

Beam 0

0450

0450 (086) to 0530 (086)

18 15 12 9

Power (dB)

–80 –79 –78 –77 –76 –75

Magnetic Latitude

80 79 78 77 76 75

TIGER

Velocity (ms–1)

Magnetic Latitude

Hankasalmi

6 3 0

0500

0510

0520

0530

UT Fig. 27.4 Ionospheric signatures of flux transfer events on the 27th March 2007. Top: 2D snapshots of the fast flows observed by the Hankasalmi radar (northern hemisphere, left) and the TIGER radar (southern hemisphere, right). The green asterisk in each panel denotes the footprint of Cluster 3. Bottom: Time series of data from individual beams of these two radars. The top two panels show the ionospheric flow velocities deduced from the radar backscatter, and the bottom two panels show the backscatter power. Poleward moving features are observed in both hemispheres (PMRAFs)

and references therein). The second and third PMRAFs observed by Hankasalmi correspond well with the first and second PMRAFs observed by TIGER. The fact that the Hankasalmi and TIGER radars observed PMRAFs in the post- and pre-noon sectors respectively indicates that pulsed reconnection was occurring over a large

394

R.C. Fear et al.

part of the magnetopause during this interval. As there is a gap in radar coverage around 12 h MLT, it cannot be shown that pulsed reconnection occurred along this entire sector of local time, but we note that the magnetic footprint of the Cluster spacecraft is in this region. As will be shown in the next section, signatures of pulsed reconnection were also observed by Cluster.

27.4 In Situ Observations: FTEs and Their Azimuthal Extent Cluster observations of the flux transfer events before and after the spacecraft crossed the magnetopause are shown in Figs. 27.5 and 27.6. In each figure, the top two panels show the components of the magnetic field tangential to the magnetopause, followed by the BN components observed by the four spacecraft, which are ordered by the positions of the spacecraft (equatorward to poleward: C3, C4, C1 and C2). The next panel shows a spectrogram of the electrons observed moving away from the magnetopause by Cluster 2 (0ı and 180ı pitch angles in Figs. 27.5 and 27.6 respectively). The angle between the magnetic field observed by Cluster 3 and the l vector .’LM /, and the IMF clock angle .™CA / are shown in the penultimate panel by green and purple traces respectively, and the magnetic field magnitude is shown in the bottom panel.

Fig. 27.5 Magnetospheric FTEs observed by Cluster. From top: The magnetic field components observed by the four Cluster spacecraft, a spectrogram of 0ı pitch angle electrons, the angle between the magnetic field observed by Cluster 3 and the vector l (green trace) and IMF clock angle (purple), and the magnetic field strength

27

Mixed Azimuthal Scales of Flux Transfer Events

395

Fig. 27.6 Magnetosheath FTEs observed by Cluster. The spectrogram shows 180ı pitch angle electrons; the format is otherwise the same as Fig. 27.5

Before 0504 UT, all four spacecraft were in the magnetosphere, and observed a magnetic field which was almost entirely in the l component. At 0504 UT Clusters 2 and 3 crossed the magnetopause, followed by Clusters 1 and 4 at 0508 UT. The magnetosheath magnetic field was largely in the negative l direction .’LM D 180ı/, except between 0515 and 0524 UT when ’LM dropped to 150ı (observed in ™CA between 0509 and 0516 UT, and indicating that the IMF lag is too small). Between 0450 and 0524 UT, all four Cluster spacecraft observed bipolar variations of the BN component which are characteristic signatures of flux transfer events. We identify five FTEs which were clearly observed by all four spacecraft, which are indicated by vertical purple lines in Figs. 27.5 and 27.6. Two of these FTEs (at 04:56 and 05:22 UT) exhibited signatures at C3 and C4 that were not clearly bipolar, but they are similar enough to the other signatures observed for us to regard them as ‘irregular’ polarity FTEs [14]. In each case, the FTE was observed first by C3 and C4 (simultaneously), then by C1 and C2 (at approximately the same time as each other, but 30–60 s after C3). This order indicates a northward motion which is consistent with the ‘standard’ polarity of the BN signatures that was observed whenever the polarity was clear [14]. Since the Cluster tetrahedron had little extent in the n dimension, we cannot obtain a reliable three-dimensional velocity for these two FTEs by conventional multi-spacecraft timing analysis [6]. Therefore, we use the time delays between the FTEs being observed at Cluster 4 (which we take as a reference spacecraft) and Clusters 1 and 2 to obtain the FTE velocities in the l-m plane (which does contain large spacecraft separations as shown in Fig. 27.3).

396

R.C. Fear et al. Table 27.1 2D FTE velocities determined from three-spacecraft Time (UT) dt41 .s/ dt42 .s/ jvj.km s 1 / vO LM 04:54:20 15.8 32.4 257 (0.932, 0.363) 04:56:20 37.8 32.6 197 (0.959, 0:285) 05:14:40 18.2 18.8 393 (0.990, 0:140) 05:17:52 27.4 61.4 168 (0.950, 0.312) 05:22:31 59.0 61.6 122 (0.991, 0:132)

timing Angle to l 21ı 16ı 8ı 18ı 8ı

The results are shown in Table 27.1. The two magnetospheric FTEs were observed to move at speeds of 257 and 197 km s1 , in a largely poleward direction. Such a direction of motion is expected from the fact that there are no significant azimuthal forces exerted on the reconnected flux tubes by either the magnetic tension in the reconnected field lines or the magnetosheath flow. Unfortunately, neither of these FTEs exhibited plasma signatures at any of the spacecraft (i.e. the draping region alone was observed). We therefore take the “characteristic time” (the time difference between the positive and negative peaks of BN , as defined by Kawano et al. [8]) as a proxy for the duration of the FTE. These two FTEs had characteristic times of 26 and 29 s respectively; these correspond to poleward extents of 6,700 and 5,700 km respectively, which is smaller than the 8,500 km azimuthal separation of Clusters 1 and 2. Consequently, it appears that these two FTEs have a larger azimuthal than poleward extent; in such conditions of high magnetic shear this is more consistent with extended X-line models of FTEs (e.g. [10, 17, 20]). However, there are some interesting differences between spacecraft in some other FTEs which were observed, and which are indicated by green arrows in Fig. 27.5. At 0455 UT an FTE was observed at Clusters 1, 3 and 4 (most clearly at Cluster 1). However, no BN signature is evident at Cluster 2. Considering that Cluster 2 was nearer the magnetopause than Cluster 1 and crossed the magnetopause before Cluster 1 (see Figs. 27.3 and 27.6), this implies that the azimuthal extent of this FTE is more limited. The dearth of FTEs at Cluster 1 between 0458 and 0502 UT could be explained by the fact that Cluster 1 is the furthest spacecraft from the magnetopause, but considering the size of the FTEs observed by Cluster 2 at 0459 and 0502 UT, and the fact that the separation between Clusters 1 and 2 normal to the magnetopause is only 1,200 km, it is perhaps more likely that these are also more azimuthally-limited features. Finally, between 0502 and 0504 UT, Cluster 2 observed three FTEs whilst Cluster 1 only observed a single event. The magnetosheath observations are shown in Fig. 27.6. The three FTEs identified at all four spacecraft are again indicated by purple lines. All three occurred during the interval where the magnetosheath magnetic field rotated dawnward (’LM , penultimate panel), caused by the rotation in the lagged IMF clock angle observed a few minutes before. The electron signature observed at 0514 UT was not clearly distinct from that of the previous FTE, but distinct electron signatures were observed for the other two FTEs, most clearly in the antiparallel pitch angle bin at Cluster 2. Three distinct electron energy enhancements were observed during the course of the

27

Mixed Azimuthal Scales of Flux Transfer Events

397

0518 FTE; the time from the beginning of the first enhancement to the end of the last was 60 s. A single electron signature was observed at 0523 UT, which lasted 67 s. Multiplying these durations by the speeds reported in Table 27.1 gives poleward scale sizes of 8,200 and 10,000 km respectively (these two FTEs had a lower speed than those observed before the magnetopause crossing). Prior to 0515 UT, the rate at which FTEs were observed was higher, due at least in part to the proximity of the spacecraft to the magnetopause. It is therefore harder to link specific FTEs at different spacecraft with a high degree of certainty. However it is again clear that there are significant differences between Clusters 1 and 2, separated predominantly in the azimuthal direction. We highlight two FTEs in Fig. 27.6 which were observed by a subset of the spacecraft (both indicated by a green arrow). Clusters 3, 4 and 2 observed a clear FTE at 0513 UT, which was not observed by Cluster 1, despite the fact that this spacecraft was now the closest to the magnetopause. Cluster 1 alone observed a clear FTE at 0516 UT; recalling again that the maximum separation of the spacecraft normal to the magnetopause was only 1,200 km, this also suggests a spatially-limited structure.

27.5 Summary We have presented ground- and space-based observations of flux transfer events during an interval when the IMF clock angle was near 180ı and Cluster was near magnetic noon. The ionospheric observations indicate that pulsed magnetic reconnection occurred over a large part of the dayside magnetopause. Some FTEs were observed at all four spacecraft indicating that they extended azimuthally by at least 8,500 km. In some cases, this is larger than their observed poleward scale size, consistent with long X-line models of FTE formation. However, there are also significant differences in FTE observation on these azimuthal scales, implying that FTEs can also occur more patchily. These FTEs could be explained by any of the models. Acknowledgments OMNI data obtained from the GSFC/SPDF OMNIWeb interface at http://omniweb.gsfc.nasa.gov, and based upon Wind MFI and SWE observations courtesy of R. Lepping and K. Ogilvie. Work at Leicester was supported by Science and Technology Facilities Council (STFC) grant PP/E000983/1. Cluster data analysis was done with the QSAS science analysis system provided by the United Kingdom Cluster Science Centre, supported by STFC.

References 1. Balogh, A., Carr, C. M., Acu˜na, M. H., Dunlop, M. W., Beek, T. J., Brown, P., Fornac¸on, K.-H., Georgescu, E., Glassmeier, K.-H., Harris, J., Musmann, G., Oddy, T., and Schwingenschuh, K.: The Cluster Magnetic Field Investigation: Overview of in-flight performance and initial results. Ann. Geophys., 19, 1207–1217 (2001). http://www.ann-geophys.net/19/1207/2001/

398

R.C. Fear et al.

2. Chisham, G., Lester, M., Milan, S. E., Freeman, M. P., Bristow, W. A., Grocott, A., McWilliams, K. A., Ruohoniemi, J. M., Yeoman, T. K., Dyson, P. L., Greenwald, R. A., Kikuchi, T., Pinnock, M., Rash, J. P. S., Sato, N., Sofko, G. J., Villain, J.-P., Walker, A. D. M.: A decade of the Super Dual Auroral Radar Network (SuperDARN): scientific achievements, new techniques and future directions. Surv. Geophys., 28, 33–109 (2007). doi:10.1007/s10712007-9017-8 3. Dungey, J. W.: Interplanetary magnetic field and the auroral zones. Phys. Rev. Lett., 6, 47–48 (1961). doi:10.1103/PhysRevLett.6.47 4. Fear, R. C., Milan, S. E., Fazakerley, A. N., Owen, C. J., Asikainen, T., Taylor, M. G. G. T., Lucek, E. A., R`eme, H., Dandouras, I., and Daly, P. W.: Motion of flux transfer events: A test of the Cooling model. Ann. Geophys., 25, 1669–1690 (2007). http://www.anngeophys.net/25/1669/2007/ 5. Fear, R. C., Milan, S. E., Fazakerley, A. N., Lucek, E. A., Cowley, S. W. H., and Dandouras, I.: The azimuthal extent of three flux transfer events. Ann. Geophys., 26, 2353–2369 (2008). http://www.ann-geophys.net/26/2353/2008/ 6. Harvey, C. C.: Spatial gradients and the volumetric tensor. In: Paschmann, G., Daly, P.W. (eds.) Analysis Methods for Multi-Spacecraft Data, pp. 307–348, ISSI, Berne (1998) 7. Johnstone, A. D., Alsop, C., Burdge, S., Carter, P. J., Coates, A. J., Coker, A. J., Fazakerley, A. N., Grande, M., Gowen, R. A., Gurgiolo, C., Hancock, B. K., Narheim, B., Preece, A., Sheather, P. H., Winningham, J. D., and Woodliffe, R. D.: PEACE: A Plasma Electron and Current Experiment. Space Sci. Rev., 79, 351–398 (1997). doi:10.1023/A:1004938001388 8. Kawano, H., Kokubun, S., and Takahashi, K.: Survey of transient magnetic field events in the dayside magnetosphere. J. Geophys. Res., 97, 10 677–10 692 (1992) 9. King, J. H. and Papitashvili, N. E.: Solar wind spatial scales in and comparisons of hourly Wind and ACE plasma and magnetic field data. J. Geophys. Res., 110, A02104 (2005). doi:10.1029/2004JA010649 10. Lee, L. C. and Fu, Z. F.: A theory of magnetic flux transfer at the Earth’s magnetopause. Geophys. Res. Lett., 12, 105–108 (1985) 11. Milan, S. E., Lester, M., Cowley, S. W. H., and Brittnacher, M.: Convection and auroral response to a southward turning of the IMF: Polar UVI, CUTLASS, and IMAGE signatures of transient magnetic flux transfer at the magnetopause. J. Geophys. Res., 105, 15741–15756 (2000). doi:10.1029/2000JA900022 12. Owen, C. J., Fazakerley, A. N., Carter, P. J., Coates, A. J., Krauklis, I. C., Szita, S., Taylor, M. G. G. T., Travnicek, P., Watson, G., Wilson, R. J., Balogh, A., and Dunlop, M. W.: Cluster PEACE observations of electrons during magnetospheric flux transfer events. Ann. Geophys., 19, 1509–1522 (2001). http://www.ann-geophys.net/19/1509/2001/ 13. Peredo, M., Slavin, J. A., Mazur, E., and Curtis, S. A.: Three-dimensional position and shape of the bow shock and their variation with Alfvenic, sonic and magnetosheath Mach numbers and interplanetary magnetic field orientation. J. Geophys. Res., 100, 7907–7916 (1995) 14. Rijnbeek, R. P., Cowley, S. W. H., Southwood, D. J., and Russell, C. T.: A survey of dayside flux transfer events observed by ISEE-1 and ISEE-2 magnetometers. J. Geophys. Res., 89, 786–800 (1984) 15. Russell, C. T. and Elphic, R. C.: Initial ISEE magnetometer results: Magnetopause observations. Space Sci. Rev., 22, 681–715 (1978). doi:10.1007/BF00212619 16. Saunders, M. A., Russell, C. T., and Sckopke, N.: Flux transfer events: Scale size and interior structure. Geophys. Res. Lett., 11, 131–134 (1984) 17. Scholer, M.: Magnetic flux transfer at the magnetopause based on single X-line bursty reconnection. Geophys. Res. Lett., 15, 291–294 (1988) 18. Shue, J.-H., Chao, J. K., Fu, H. C., Russell, C. T., Song, P., Khurana, K. K., and Singer, H. J.: A new functional form to study the solar wind control of the magnetopause size and shape. J. Geophys. Res., 102, 9497–9511 (1997) ¨ and Cahill, Jr., L. J.: Magnetopause structure and attitude from Explorer 12 19. Sonnerup, B. U. O. observations. J. Geophys. Res., 72, 171–183 (1967) 20. Southwood, D. J., Farrugia, C. J., and Saunders, M. A.: What are flux transfer events? Planet. Space Sci., 36, 503–508 (1988). doi:10.1016/0032-0633(88)90109-2

Chapter 28

Perspectives Gained from a Combination of Polar, Cluster and ISEE Energetic Particle Measurements in the Dayside Cusp T.A. Fritz

Abstract Energetic particles are a consistent and common feature of the high-altitude dayside cusp. Observing these particles in a region where they cannot be stably trapped is one of the most striking findings of the Polar and Cluster satellites. The source of these cusp energetic particles (CEP) has centered on the possible role of the bow shock, leakage from the magnetosphere, and local acceleration within the cusp itself. The Polar satellite has documented that the shocked solar wind plasma enters the weak geomagnetic field of the polar region and produces cusp diamagnetic cavities (CDC) of apparent tremendous size .6 RE / well within the traditional magnetosphere. Within these cavities the local magnetic field is depressed and very turbulent. The intensities of the energetic ions are observed to increase by many orders of the magnitudes during the CDC encounters. The four Cluster spacecraft have typically not observed such a large cusp and extended diamagnetic cavity. The search for evidence to resolve this inconsistency led to a revisiting of ISEE 1 and 2 satellite measurements during two encounters with the high altitude cusp. In each of these cases the very good energy and angular resolution of the ISEE energetic particle experiment revealed that energetic ions within the cusp located at GSM Z ranging from 4 RE to 5 RE appeared from closer to the Earth and streamed outward in very close association with the diamagnetic cavities in the measured magnetic field. The electrons demonstrated a peaked at 90ı pitch angle distribution indicative of being confined within a cusp minimum field trap. In one of these cases the electron fluxes peak near the cusp boundary and in the other case they sharply define the CDC boundaries varying in a strictly anticorrelated manner during a large geomagnetic storm. The charge state distribution of these cusp cavity ions is indicative of their seed populations being a mixture of ionospheric and solar wind particles in many cases. Taken together these facts argue for a local acceleration of plasma within the cusp to many 10s and 100s of keV. By their geometry cusp magnetic field lines are connected to all of the magnetopause boundary layers and these cusp charged particles will form an energetic particle layer on the magnetopause. A source of energetic particles in the dayside T.A. Fritz () Center for Space Physics, Boston University, Boston, MA 02215, USA e-mail: [email protected]

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 28, c Springer Science+Business Media B.V. 2010

399

400

T.A. Fritz

high-altitude CDC will be effective in transferring the solar wind energy, mass, and momentum into the Earth’s magnetosphere and could provide the source of the ring current ions.

28.1 Introduction The concept of the magnetosphere cusp prior to the Polar and Cluster missions was that of a narrow funnel-shaped volume in the sub-solar region that contains open geomagnetic field lines that topologically form the magnetopause. The region of the cusp has been studied and characterized at low altitudes where the narrow geometry seems to be consistent with observations [13, 14]. At higher altitude controversy exists as to what the topology might be and there are a plethora of names that have been created to cover this confusion. The most striking and significant finding of the Polar satellite is that energetic particles are observed to be consistently and continuously present in this high latitude, high altitude dayside cusp region where they cannot be stably trapped yet occupy the same region as the shocked solar wind. “Energetic” refers to ions with energies greater than 25 keV with upper limits being in the 100s of keV to MeV energies. The source of these energetic particles remains a topic of debate and scientific discussion. The initial observations of such energetic particles in this region were described by Shabansky [19] who interpreted them to be the result of the azimuthal drift of radiation belt particles caused by gradient and curvature effects in the geomagnetic field. These particles over a narrow range of radial distances will move to high latitudes as they follow the magnetic field minimum into this region. More recently Chen et al. [3, 4] and Fritz et al. [9] have reported observations of cusp energetic particles (CEP) and argued that they are accelerated locally within the cusp.

28.2 Polar Observations In April 1999 the Polar satellite observed regions of shocked solar wind plasma for many hours at a time at locations well inside the nominal location of the magnetopause. Polar was launched into a 1.8 by 9 RE orbit on February 24, 1996 with an inclination of 86ı . On board Polar, the Magnetospheric Ion Composition Spectrometer (MICS), a part of the Charge and Mass Magnetospheric Ion Composition Experiment (CAMMICE), is a one-dimensional time-of-flight electrostatic analyzer with post acceleration measuring ions with an energy/charge of 1–220 keV/e with very good angular resolution. The MICS, mounted perpendicular to the spin axis, is able to obtain a two-dimensional distribution at one energy per charge during each 6 s spin period. A complete energy spectrum is obtained in 32 spin periods. Polar was located on the morning side of local noon and was near to apogee of 9 RE when it became immersed in these extended regions of shocked solar wind plasma.

28

Perspectives Gained from a Combination of Polar, Cluster and ISEE

401

Fig. 28.1 Polar pass through the dayside cusp on 20 April 1999. The five panels on the left side of this figure are energy spectrograms of the major ion species. The top panel labeled OC6 displays the intensity as function of energy for the high charge state oxygen ions characteristic of the solar wind. In the second panel the intensity of oxygen ions with charge states associated with an ionospheric source are displayed. The threshold energy of 50 keV for the detection of these OC , OCC ions is much higher than that of OC6 shown in the first panel and is due to an instrumental threshold for the unique identification of these ions. In the third and fourth panels the two charge states of Helium ions are displayed. In the fifth panel the energy spectrogram for hydrogen ions (protons) is displayed. The vertical scale on the far left indicates the energy of the ions in keV. The color bars indicate the intensity of the various ion species in particles/s cm2 sr keV. On the upper right side the figure is the orbit of Polar in SM coordinates (provided by NASA/GSFC SPOF). The red trace along the orbit marks the location of the region corresponding to the same red bar in the top left panel. The bottom right panel displays the measurement of the magnetic field [17]

In Fig. 28.1 a three part figure of data obtained by Polar on April 20, 1999 is displayed. Polar was northward bound through the radiation belts from 07 UT to about 12 UT. The CAMMICE/MICS sensor with it selection by energy/charge followed by a post acceleration of 20 kV/charge is able to measure the intensity of these ion species to energies of less than 1 keV. Note the abrupt appearance of intense fluxes of these shocked solar wind ions at 12 UT which lasted until 18W20 UT in Fig. 28.1. Although lacking the lower energies (<50 keV) due to the threshold response for these ions, very energetic (>50 keV) OC ions of presumed ionospheric origin were observed to be present at the same location as the oxygen ions of solar wind origin. A similar picture holds for the helium ions as well. HeCC (or alpha particles) are abundantly present in the solar wind but HeC is the dominant charge state in

402

T.A. Fritz

the ionosphere. At 12 UT there was an abrupt increase in the fluxes of the low energy HeCC that lasted until about 18:20 UT and showed the same variations in time as the high charge state oxygen ions in the top panel. HeC ions were present as well at higher energies indicating a mixture of ions from two separate origins. The hydrogen ions presented in the fifth panel also indicate that an abrupt increase in the intensity of protons most noteable for energies <10 keV began at 12 UT and lasted until 18:20 UT. In each of the spectrograms there are very energetic >25 keV ions present during the 12–18 UT interval reaching the upper energy cutoff of the MICS sensor (200 keV) in the case of the protons. In the other two sections of Fig. 28.1 note that as Polar approached apogee the magnetic field decreased gradually to nominal values of 60 to 70 nT but from 12 UT to 18 UT the field was further depressed with large fluctuations of high frequency some of which appear to approach extremely small values (e.g., 0 nT). This behavior and the occurrence of a near zero magnetic field in the cusp was noted by Russell et al. [18]. These observations by the Polar satellite indicate that shocked solar wind ions that are associated with the magnetospheric cusp were continuously present for a period of 6 h. These ions appear to create a diamagnetic cavity of large dimensions (but this could be a series of many smaller cavities) with the magnetic field demonstrating large fluctuations and turbulence, a region described as a Cusp Diamagentic Cavity or CDC. The presence of very energetic 10s and 100s of keV ions co-located with the intense solar wind ions is surprising as these ions should have a short lifetime against loss of seconds in an open field line topology associated with the magnetic cusp. The energetic particles have been first described by Chen et al. [3,4] as Cusp Energetic Particles or CEP. Upstream at the L1 liberation point the ACE satellite was measuring a steady solar wind (data not shown) with a measured pressure of 4 nPa. At 15W30 UT the pressure ramped up to a peak of 8nPa. The IMF was also fairly steady with positive By and negative Bx for most of the 6 h interval. Bz was initially negative, switched positive at 14W20 UT, negative again at 15:30 UT, then back positive at 16:15 UT where it remained for the rest of the interval. In Fig. 28.2 in the same format as Fig. 28.1 the Polar data are displayed for April 24, 1999. Polar encountered a region of solar wind plasma from 08:00 UT to 12:10 UT, a period of more than 5 h, on this day. In this case the features of a particle spectrum extending to very energetic particles (CEP) are coincident with a depressed and turbulent measured magnetic field (CDC) that averaged about 20 nT over the bulk of the diamagnetic cavity with many measured values close to 0 nT. This same behavior is basically observed by Polar pass after pass through the dayside cusp region. Upstream at the ACE spacecraft the solar wind was again fairly steady with a pressure of 3 nPa. The orientation of the IMF is such that all three GSM components are positive with Bz very small. Niehof et al. [15] have shown that not all cusp crossings by Polar consist of CDC and CEP events but rather that these are features within a possibly broader cusp. Finkemeyer et al. [6] and Fritz et al. [8] reported that the CDC/CEP features are found in 90% of the Polar crossing of the cusp during the first 2 years of Polar operations. Niehof et al. [15] further found that the variations of the magnetic field and particle fluxes were strongly anti-correlated inside the

28

Perspectives Gained from a Combination of Polar, Cluster and ISEE

403

Fig. 28.2 Polar pass through the dayside cusp on 24 April 1999 in the format of Fig. 28.1

CDC cavities and that this anti-correlation existed at the maximum time resolution available for the particle fluxes which for the Polar Hydra experiment was 13.8 s. This particular cusp crossing by Polar has been discussed by Rae et al. [16] who argued that the variations of particle intensity and magnetic field in the cusp were caused by Alfven waves upstream of the magnetopause. These authors based their discussion on variations in the magnetic field and particle fluxes being out of phase, that is, anti-correlated. While it may be possible for such upstream waves to modulate the entry of the shocked solar wind into the cusp, such waves or any other feature in the upstream magnetosheath appear unlikely to produce the CEP particles as a Polar pass on April 21, 1999 demonstrated. During this pass presented in Fig. 28.3 CEP particles were present as Polar first entered the cusp (denoted as region A), were absent as Polar moved into the magnetosheath (region B), and were again present as Polar returned to the cusp (region C). As Polar encountered the cusp with a CDC/CEP at 0643 UT the particles present were of solar wind origin HeCC with an energetic (25 to >200 keV) component present of the same composition. A pressure pulse in the solar wind shown in the middle panel of Fig. 28.3 struck the magnetopause at 0730 UT pushing the magnetopause in over Polar. Polar remained in the magnetosheath from 0730 UT until returning to the cusp at 0920 UT. During this period the characteristics of the magnetic field variations changed and the solar wind velocity along GSE X jumped to approximately 200 km=s

404

T.A. Fritz

Fig. 28.3 Polar and Wind observations on 21 April 1999. The Wind measurements of the solar wind pressure have been propagated to the subsolar magnetopause by the measured solar wind velocity. The color-coded intensity scale for the Polar proton and alpha particle fluxes are the same as those shown in Figs. 28.1 and 28.2. The region between the two red lines labeled B is the magnetosheath while the regions A and C between the red and black lines have the properties described in the text for a CDC and CEP region

indicative of being in the magnetosheath. No energetic particles were observed in the magnetosheath. Upon reentering the cusp, CEP particles were again observed until Polar moved into the polar cap around 12:00 UT. In this event there were no energetic particles in the magnetosheath but they were present during the two CDC intervals that sandwiched this excursion into the magnetosheath. This argues that processes external to the cusp were not responsible for producing the CEP particles in the magnetosheath or bow shock that then find their way into the cusp.

28.3 Cluster Observations Cluster on the other hand has rarely observed a cusp similar to those that are very frequent in the Polar data. On some occasions the satellites did detect a CDC in combination with simultaneously enhanced energetic ion fluxes (e.g., [1, 20]). Usually Cluster sampled what has be-come known as the Stagnant Exterior Cusp or SEC [12]. Zhang et al. [24] found that energetic particles were present on 80% of the Cluster cusp crossings during the first 2 years of Cluster operations but the magnetic field was rarely depressed for an extended period and those energetic particles

28

Perspectives Gained from a Combination of Polar, Cluster and ISEE

405

Fig. 28.4 An overview of Cluster Ion Spectrometer (CIS), Fluxgate Magnetometer (FGM), and Research with Adaptive Particle Imaging Detectors (RAPID) data from 10 to 13 UT on 23 March 2002. (a) Plasma density, (b) plasma velocity Vx component, and (c–e) magnetic field components in the GSM coordinate system (in nT). (f–i) Energetic electron, proton, helium, and heavy ion flux. Note that the energetic particles are confined to the cusp boundary. The energy ranges shown are: Electrons (>30 keV), Protons and He (>30 keV), Heavy Ions (>210 keV) (From [24])

present were usually associated with a cusp boundary. An example occurring on March 23, 2002 that she has published showing the association of these particles with the cusp boundary is presented in Fig. 28.4. Examples of Cluster magnetic field observations in the outer cusp are shown in Fig. 28.5 [5] indicating a cusp diamagnetic cavity of only some 10 min duration.

406

T.A. Fritz

Cluster – Orbit & s/c – configuration

Cluster – Orbit configuration 8

2001 Feb 13

10

2001 Feb 20/21

6 5

4 2

0 Z Re

0

–5

–2 –4 00:0 UT 21 Feb

–6 –10

–8

–15 –5

0

5 X Re 10

15

20

100 90

B

Cusp @ 23.20 UT

–10

Cusp @ 20.00 UT

–12 70

mag

80

60

70

50

60

–5

0

80

B

X Re

5

10

mag

40

50 40

30

30

20

20

10

10 0 19:0 19:10 19:20 19:30 19:40 19:50 20:0 20:10 20:20 20:30 20:40 20:50 21:0

0 22:0 22:10 22:20 22:30 22:40 22:50 23:00 23:10 23:20 23:30 23:40 23:50 24:00

UT

UT

Fig. 28.5 Cluster FGM observations of short duration CDCs (From [5])

28.4 ISEE 1 and 2 Observations This dichotomy in the impression of the energetic particles within the cusp gained from the Polar and Cluster missions suggested that an examination of data from the older International Sun Earth Explorer (ISEE 1 and 2) satellites may indicate encounters with the cusp. This would allow the very good energy and angular resolution of the energetic particle experiment on these satellites to be used. Two events, where ISEE 1 and 2 were in the cusp, have been reported by Whitaker et al. [22, 23] and Walsh et al. [21]. On September 29, 1978, the two ISEE satellites were outbound in near identical orbits close to local noon and each encountered a CDC. The event happened during a major magnetic storm where the upstream Bz was about 30 nT, Dst went to 224 nT, the hourly AE index reached 775 nT and the dynamic pressure increased by a factor of 3. Surprisingly ISEE-2 slightly ahead of ISEE-1 entered the CDC almost 30 min before ISEE-1 and stayed in the cavity for almost 1 h as seen in Fig. 28.6. Both satellites exited the CDC within 6 min of one another. This implies that the boundary of the CDC maintained a location between the two satellites for 27 min and must therefore have been in motion. Figure 28.7 shows ISEE-1 data as that satellite entered and exited the CDC. Although the time resolution of the composition data is poorer than the particle and magnetic field data, it is clear that the CDC was filled with a combination of solar HeCC

28

Perspectives Gained from a Combination of Polar, Cluster and ISEE

407

Fig. 28.6 Trajectory in GSM of ISEE-1 and 2 on September 29, 1978 (figure 1 from [21]).The X’s indicate the initial entrance and final exit of the CDC cavity

BT (nT)

Log Flux (particles cm−2 sec−1 ster −1 keV−1)

ISEE-1 1054 10 103 102 101 100 4

10

103 102

September 29, 1978

H+ He++ He+

O+

P1 24-44.5 keV P2 44.5-65.3 keV P3 65.3-95.5 keV P4 95.5-143 keV

101 104

1

2

3

4

103 102

E1 22.5-39 keV E2 39-75 keV E3 75-120 keV E4 120-189 keV

101 160 140 120 100 80 60 12:00

12:30

13:00 Universal Time

13:30

14:00

Fig. 28.7 ISEE-1 data on September 29, 1978. The top panel shows data from the Plasma Experiment (40 eV–17.36 keV), where triangles, X’s, open squares, and diamonds correspond to HC , HeCC , HeC , and OC ions respectively. The time resolution for the plasma data is 18 min. The second and third panels contain the proton and electron fluxes respectively, integrated over all pitch angles for four energy channels measured by ISEE-1. The proton and electron flux are presented with a time resolution of 36 s. Panel four shows the total local magnetic field strength. The local magnetic field strength is displayed with 4 s resolution. The four zones of increased flux are marked with vertical dotted lines and labeled sequentially in panels three and four (figure 2 from [21])

408

T.A. Fritz Isee1 - Ions 4

3

KeV

Intensity

3

2

2 1 0

−1 12:00:00 1978 272

13:00:00 Time

14:00:00 1978 272

Fig. 28.8 ISEE-1 energetic ion spectrum for the event period of Fig. 28.7. Note the lack of any energy dispersion with time in the intensity of the ions during this encounter with a CDC (From the MIDL Application at http://sd-www.jhuapl.edu/MIDL/)

and ionospheric HeC ; OC ions. There appear to have been multiple cavities that ISEE-1 encountered with the energetic electrons defining each cavity exactly. This is true of the fluxes at all energies shown from 20 to 190 keV. The variation of the ions is similar to the electrons with the ions intensities remaining high and extending beyond the sharp cavity boundary locations shown in the magnetic field and electron data. There is an absence of any velocity dispersion in the case of both the electrons and the ions indicating there was essentially no magnetic gradient or curvature drift from the source responsible for their energization. An energy spectrum of the ions is shown in Fig. 28.8 indicating that there is little, if any, energy dispersion in the appearance of the energetic ions and the process for energization of these ions must be able to produce energies in excess of 1 MeV. Furthermore the pitch angle distributions of these particles shown below in Figs. 28.10–28.12 provide strong evidence of a local mechanism for the acceleration of these particles. In order to interpret the format of these figures the energetic ions (24 to 45 keV) are displayed in a latitude/longitude mapping of the unit sphere and shown in Fig. 28.9 for a normal peaked-at-90ı pitch angle distribution. In Fig. 28.10 the three dimensional distribution these ions are presented in the format described in Fig. 28.9 and its caption. These ions have a consistent distribution showing the hemisphere centered on a pitch angle of 180ı fully filled while the opposite hemisphere is devoid of particles indicative of these fluxes moving away from Earth. Figures 28.8 and 28.11 illustrate that the source of these energetic ions is active into the 100s of keV of energy. Figure 28.12 is the display for same period but for energetic electrons with energy from 22.5 to 39.0 keV. The electron 3-D distributions are different from the ions in that they show a sharply peaked at 90ı distribution indicative of a trapped distribution. If the CDC is the energization source/location of these electrons they can become trapped in the magnetic field minimum of the cavity producing the distribution observed. Note the three panels on the left side of the bottom row which show a region where electrons were flowing from a region closer to the Earth just before ISEE-1 left the CDC cavity. The hemisphere centered on the anti-field line

Perspectives Gained from a Combination of Polar, Cluster and ISEE

a

409

ISEE-1 Energetic particle WAPS sensor b spin

B

spin

axis

0

15

0

12 90

60

30

scanning platform

spacecraft

scanning platform:12 steps

28

270

0

Φ

90

180

spacecraft spin : 8 sectors The locus of α = 90 degree pitch angle∗

∗ Will plot as a sinusoid on a latitude/longitude projection of the unit sphere

Fig. 28.9 This figure demonstrates how one interprets Figs. 28.10–28.12. The detector scans through the unit sphere in 36 s (left) and the fluxes are linearly translated from elevation and azimuth onto an orthogonal Cartesian coordinate system (right) that is essentially the GSE coordinate system. The horizontal axis has GSE X at 0ı and GSE Y at 90ı . The vertical axis has the GSE X, Y plane at the midplane of the figure, the GSE – Z axis is at the bottom and the GSE C Z axis at the top of the figure. The red line corresponds to the locus of all 90ı pitch angles. The pitch angle contours are determined by calculating the angle between the instantaneous magnetic field vector [17] and the sensor viewing direction and then overlaying the contoured array onto the plot on the right. The blue arrow is the magnetic field vector, where it points from the center of the 30ı contour to the center of the 150ı contour in the display on the right

direction remains filled for over 100 s. The velocity of these electrons is about 0.3 c at these energies so they will travel over 15 earth radii in 1 s. The most likely explanation for their being observed for 100 s is that they are created by a local source. The cusp is therefore most likely the location for the energization of ions and electrons to many 10s and 100 of keV. A possible engerization mechanism for such a source in the cusp has recently been described by Chen [2].

28.5 Conclusions The observations presented here demonstrate that the cusp is most likely a location capable of producing electrons and ions with energies of many 10s and 100s of keV from a source population that consists of a combination of ionospheric and solar wind ions. These ions and electrons will have easy access to the equatorial magnetosphere via their gradient and curvature drifts following a trajectory described initially by Shabansky [19]. In addition they will form a layer of energetic particles on the magnetopause and as shown by Fritz [7] and Zhou et al. [25]

410

T.A. Fritz

Fig. 28.10 ISEE-1 3-D PAD for ions with energy from 24 to 45 keV. The figure spans the time period 12:54:40–13:03:10 UT

Fig. 28.11 ISEE-1 3-D PAD for ions with energy from 142 to 210 keV. The figure spans the time period 12:54:40–13:03:10 UT

28

Perspectives Gained from a Combination of Polar, Cluster and ISEE

411

Fig. 28.12 ISEE-1 3-D pitch angle distributions of electron flux (22.5–39.0 keV) as the spacecraft exits the CDC for the final time. The figure spans the time period 12:54:40–13:03:10 UT. Throughout the CDC and prior to the satellite exiting the CDC, the flux shows a clear peak at a PA of 90ı (figure 7 from [21])

they could be the source of the ring current. Due to the size of the CDC cavities and their common occurrence, the role of these energetic particles must be factored into any model that purports to describe the topology and dynamics of the magnetosphere. Acknowledgements I would thank the reviewers of this paper, Drs. Jiasheng Chen, George Siscoe, Hui Zhang and Q.-G. Zong and graduate students Brian Walsh, Kate Whitaker, and Jon Niehof for many useful discussions and to acknowledge the contributions of the various instrument teams for CAMMICE, CEPPAD, and Hydra on Polar and RAPID on Cluster. The Polar effort has been supported at Boston University under a series of NASA grants: NAG5–2578, NAG5–7677, NAG5–11397, and NNG05GD23G. The Cluster effort at Boston University has been supported under another series of NASA grants: NAG5–10108 and NNG05GE90G.

References 1. Asikainen, T. and K. Mursula, Energetic particle fluxes in the exterior cusp and the high latitude dayside magnetosphere: statistical results form the Cluster/RAPID instrument, Ann. Geophys., 23, 2217–2230 (2005). 2. Chen, J., Evidence for particle acceleration in the magnetospheric cusp, Ann. Geophys., 26, 1993–1997 (2008).

412

T.A. Fritz

3. Chen, Jiasheng, T. A. Fritz, R. B. Sheldon, H. E. Spence, W. N. Spjeldvik, J. F. Fennell, S. Livi, A new, temporarily confined population in the polar cap during the August 27, 1996 geomagnetic field distortion period, Geophys. Res. Lett., 24 (12), 1447–1450 (1997). 4. Chen, Jiasheng, Theodore A. Fritz, Robert B. Sheldon, Harlan E. Spence, Walther N. Spjeldvik, Joseph F. Fennell, Stefano Livi, Christopher T. Russell, Jolene S. Pickett, and Donald A. Gurnett, Cusp energetic particle events: Implications for a major acceleration region of the magnetosphere, J. Geophys. Res., 103 (No. A1), 69–78 (1998). 5. Dunlop, M. W., B. Lavraud, P. Cargill, M. G. G. T. Taylor, A. Balogh, H. Reme, P. Decreau, K.-H. Glassmeier, R. C. Elphic, J.-M. Bosqued, A. N. Fazakerley, I. Dandouras, C. P. Escoubet, H. Laakso and A. Marchaudon, Cluster Observations of the Cusp: Magnetic Structure and Dynamics, Surveys in Geophysics 26: 1–51, DOI: 10.1007/s10712–005–1871–7 (2005). 6. Finkemeyer, B U, Harris, J J, Fritz, T A, Chen, J, Matthews, D L, A Survey of the Frequency of Occurrence of CEP Events During the First Two Years of POLAR Operations, SM-42A-02 at the 1999 AGU Spring Meeting (1999). 7. Fritz, Theodore A., The Role of the Cusp as a Source for Magnetospheric Particles: A New Paradigm, ESA Special Publication of the Proceedings of the Cluster II Workshop on Multiscale/Multipoint Plasma Measurements held at Imperial College, London 22–24 September 1999, ESA SP-499 (2000). 8. Fritz, T A, Karra, M, Finkemeyer, B, Chen, J, Statistical Studies with Polar of Energetic Particles near the dayside Cusp, SM22B-04 at the 1999 Fall AGU meeting (1999a). 9. Fritz, T.A., J. Chen, R.B. Sheldon, H.B. Spence, J.F. Fennell, S. Livi, C.T. Russell, and J.S. Pickett, Cusp energetic particle events measured by POLAR spacecraft, Phys. Chem. Earth (C), 24, 135–140 (1999b). 10. Fritz, T.A., J. Chen, R.B. Sheldon, The Role of the Cusp as a Source for Magnetospheric Particles: A New Paradigm? Adv. in Space Res., 25, No 7–8, 1445–1457 (2000). 11. Fritz, T. A., J. F. Fennell, S. Livi, J. L. Roeder, A. Daglis, H. Sommer, B. Wilken, B. Kellert, M. Henderson, M. Grande, J. B. Blake, R. Koga, J. Chen, “The Polar CAMMICE Investigation”, available on the following web site: http://www.bu.edu/buspace/papers/ cammice instrument.html 12. Lavraud, B., A. Fedorov, E. Budnik, A. Grigoriev, P. J. Cargill, M. Dunlop, H. Reme, I. Dandouras, and A. Balogh, Cluster survey of the high-altitude cusp properties: A three-year statistical study, Ann. Geophys., 22(8), 3009–3019 (2004). 13. Newell, P. T., C.-I. Meng, D. G. Sibeck, and R. Lepping, Some low-altitude cusp dependencies on the interplanetary magnetic field, J. Geophys. Res., 94, 8921–8927 (1989). 14. Newell, Patrick T., and Ching-I. Meng, Magnetopause dynamics as inferred from plasma observations on low-altitude satellites, Physics of the Magnetopause, AGU Geophysical Monograph 90, 407–416 (1995). 15. Niehof, J. T., T. A. Fritz, R.H.W. Friedel, and J. Chen, Interdependence of magnetic field and plasma pressures in cusp diamagnetic cavities, Geophys. Res. Lett., 35, L11101, doi:10.1029/2008GL033589 (2008). 16. Rae, I. J., F. R. Fenrich, M. Lester, K. A. McWilliams, J. D. Scudder, Solar wind modulation of cusp particle signatures and their associated ionospheric flows, J. Geophys. Res., 109, A03223, doi:10.1029/2003JA010188 (2004). 17. Russell, C.T., Snare, R.C., Means, J.D., Pierce, D., Dearborn, D., Larson, M., Barr, G. and Le, G., The GGS/Polar Magnetic Fields Investigation, Space Sci. Rev., 71 (1/4), also published as The Global Geospace Mission, in C.T. Russell (ed.), 877 pp, Kluwer, Dordrecht, The Netherlands, pp. 563–582 (1995). 18. Russell, C. T., J. A. Fedder, S. P. Slinker, X-W. Zhou, G. Le, J. G. Luhmann, F. R. Fenrich, M. O. Chandler, T. E. Moore and S. A. Fuselier, Entry of the POLAR spacecraft into the polar cusp under northward IMF conditions, Geophys. Res. Lett., 25, 3015–3018 (1998). 19. Shabansky, V. P., Some processes in the magnetosphere, Sp. Sci. Rev., 12, 299–418 (1970). 20. Vogiatzis, I. I., T. E. Sarris, E. T. Sarris, O. Santol´ık, I. Dandouras, P. Robert, T. A. Fritz, Q.-G. Zong, and H. Zhang, Cluster observations of particle acceleration up to supra-thermal energies in the cusp region related to low-frequency wave activity–possible implications for the substorm initiation process, Ann. Geophys., 26, 653–669 (2008).

28

Perspectives Gained from a Combination of Polar, Cluster and ISEE

413

21. Walsh, B. M.,T. A. Fritz, N. M. Lender, J. Chen, and K. E. Whitaker, Energetic Particles Observed by ISEE-1 and ISEE-2 in a Cusp Diamagnetic Cavity on September 29, 1978, Ann. Geophys., 25, 2633–2640 (2007). 22. Whitaker, K. E., J. Chen, and T. A. Fritz, CEP populations observed by ISEE 1, Geophys. Res. Lett., 33, L23105, doi:10.1029/2006GL027731 (2006). 23. Whitaker, K.E., T.A. Fritz, J. Chen, and M.M. Klida, Energetic particle sounding of the magnetospheric cusp with ISEE-1, Ann. Geophys., 25, No. 5, p1175–1182 (2007). 24. Zhang, H., Q.-G. Zong, T. A. Fritz, S. Y. Fu, S. Schaefer, K. H. Glassmeier, P. W. Daly, H. R`eme, and A. Balogh, Cluster observations of collisionless Hall reconnection at high-latitude magnetopause, J. Geophys. Res., 113, A03204, doi:10.1029/2007JA012769 (2008). 25. Zhou, X.-Z., T. A. Fritz, Q.-G. Zong, Z. Y. Pu, Y.-Q. Hao, and J.-B. Cao, The cusp: a window for particle exchange between the radiation belt and the solar wind, Annales Geophysicae, 24, 3131–3137 (2006).

Chapter 29

Energetic Particles in the Cusp: A Cluster/RAPID View T. Asikainen

Abstract Energetic particles have been persistently observed in the exterior cusp by different satellite missions such as POLAR, Cluster-II, Viking, ISEE etc. Yet the source and the acceleration mechanism of these particles have remained unclear. In this paper I review our studies of energetic particles in the cusp and the nearby high-latitude region of closed magnetospheric field lines (HLPS, high-latitude dayside plasma sheet) using the data obtained by the RAPID instrument onboard the Cluster-II satellites. We conducted a large scale statistical study to examine the dependence of the energetic particle fluxes in the cusp and HLPS on solar wind/IMF conditions as well as on geomagnetic activity. The study showed that energetic ion fluxes in the HLPS correlate strongly with substorm activity and electron fluxes with solar wind speed and geomagnetic activity. In the exterior cusp a clear correlation between lower energy ions (E < 75 keV) and IMF jBy j was found while the more energetic particles in the cusp (E > 75 keV) correlated with substorm activity. Our case studies have shown that when IMF By dominates reconnection can take place near the cusp and release energetic particles from closed field lines to the cusp. Coupled with these detailed observations the statistical results imply that the energetic particles in the HLPS and the cusp originate in the near-Earth magnetotail from where they can drift to the HLPS region. From the HLPS the higher energy particles diffuse more or less directly into the cusp while the lower energy particles are released into the cusp by reconnection. These observations provide a consistent explanation for the cusp energetic particles without a need for significant local acceleration of shocked solar wind plasma to MeV energies. While some energy transfer from the electromagnetic waves to plasma particles is known to occur in the cusp it cannot explain the observations discussed here.

T. Asikainen () Department of Physics Centre of Excellence in Research, P.O. Box 3000, FIN-90014, University of Oulu, Finland e-mail: [email protected]

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 29, c Springer Science+Business Media B.V. 2010

415

416

T. Asikainen

29.1 Introduction As the solar wind blows past the Earth and confines the geomagnetic field into the magnetospheric cavity two null points of magnetic field are formed at high latitudes on the dayside magnetopause. The field lines around these points form a funnel, a cusp, that extends all the way from the magnetopause into the ionosphere. The cusps have a central role in the magnetospheric dynamics as most of the solar wind plasma that enters the magnetosphere during dayside magnetospheric reconnection travels through them (e.g., Reiff et al. [14], Wing et al. [22], Onsager and Lockwood [12]). When reconnection occurs on the magnetopause shocked solar wind plasma from the magnetosheath gains access to the cusp and forms a turbulent diamagnetic cavity. Accordingly, the high altitude cusp is typically seen as a region of enhanced plasma density with a fluctuating and depressed magnetic field relative to the surroundings. In the classical model of the magnetosphere, the high-latitude dayside magnetosphere and the adjacent cusp regions cannot trap particles stably [16] because they are swept to the dayside magnetopause by the convection before they can execute a complete drift orbit. Still, a number of studies concerning the high altitude cusp [1, 2, 5, 6, 9, 10, 23, 24] have reported persistent and significant fluxes of energetic particles in this region. Numerical studies based on realistic magnetic field models have also revealed that the high-latitude dayside regions around the cusp can quasistably trap energetic particles whose motion is primarily governed by the magnetic gradient-curvature drift rather than the convection [19]. It has also been shown that the regions around the cusp are connected to the night side magnetotail by so called Shabansky drift orbits [7, 18]. Despite the vast observational evidence of energetic particles in the cusp and related numerical modeling, the origin of these particles has remained controversial. Three different scenarios have been suggested to account for these particles: (1) local acceleration in the cusp [5], (2) inner magnetospheric source and transport from there to the cusp [1], (3) a bow shock source [4]. Chen et al. [5] and Fritz et al. [8] have reported a correlation between the energetic ion fluxes in the cusp and the power of the magnetic fluctuations in the cusp diamagnetic cavity. Based on this and other suggestive observations they have argued that the energetic particles are accelerated locally in the cusp by electromagnetic waves. Chang et al. [4] and Trattner et al. [20] have argued that the lower energy particles in the cusp may be accelerated at the quasi-parallel bow shock, entering from there into the cusp along field lines connecting the two regions. They also speculate that higher energy particles observed in the cusp may diffuse there from within the magnetosphere. Kremser et al. [10] studied the low-altitude cusp and showed that ions with energy above about 50 keV in the cusp are of magnetospheric origin while a separate lower energy ion population in the cusp (E < 15 keV) is of magnetosheath origin. Nearly all the work done on the energetic particles in the cusp has been concentrated purely on the particles observed within the cusp itself while much less attention have been given to the surrounding closed field line regions of the

29

Energetic Particles in the Cusp: A Cluster/RAPID View

417

magnetosphere which can efficiently trap energetic particles. In our work we have studied the energetic particle fluxes in the cusp and also on the adjacent closed field line region equatorward of the cusp (which we call the high-latitude dayside plasma sheet or HLPS) as well as possible particle transport between these regions. In this paper I review our work on the subject and present statistical and detailed evidence for magnetospheric origin of energetic particles in the cusp.

29.2 Instrumentation Much of the work on cusp energetic particles dates to the era of POLAR satellite (launched in February 1996) whose orbit was ideal for such studies. To provide an alternate view of the same region we have used the Cluster-II satellites whose orbital configuration allows the study of high-latitude dayside regions like the cusp, magnetopause and the high-latitude dayside plasma sheet during the northern hemisphere vernal equinox (January to May season). Each Cluster spacecraft includes a versatile set of instruments for measuring electromagnetic fields at different frequencies and charged particles at different energies. In this work we have utilized magnetic field data from the FGM instrument [3] and particle data from the CIS plasma instrument which measures low energy ion species from spacecraft potential up to 40 keV/q [15]. However, our most important source of energetic particle data is the RAPID instrument which measures electrons from 20 to 400 keV and ions (protons, helium ions and CNO group ions) at the energy range 28–1,500 keV for protons, 138–1,500 keV for He ions and 90–1,500 keV for CNO-group ions. The RAPID is capable of measuring the full 3D distribution of electrons at spacecraft spin resolution (4 s) and the 3D distribution of ions at 32 spin resolution (128 s). However, the corruption of the central ion detector heads has rendered the 3D ion distributions almost useless. A detailed description of the RAPID instrument is given by Wilken et al. [21] and an updated description of RAPID instrument and all the other Cluster instruments is available at the Cluster Active Archive website (http://caa.estec.esa.int/caa/instr doc.xml).

29.3 The High-Latitude Dayside Regions During vernal equinoxes the apogee of the Cluster orbit is at about 19.6 RE at the dayside, i.e., well outside the magnetosphere in the solar wind. On such an orbit the satellites traverse from the nightside ring current over the auroral regions and the polar cap into the high latitude dayside regions. Depending on the solar wind conditions (dynamic pressure, IMF etc.) and the geodipole tilt angle, which affect the orientation of the whole dayside magnetosphere relative to the Cluster orbit, the satellites can enter different regions on the dayside magnetosphere.

418

T. Asikainen

Fig. 29.1 Cluster-1 data from 17 February 2003. Panels from top to bottom are magnetic field intensity (FGM), plasma flow velocity (CIS), low energy ion spectrum (CIS/CODIF), energetic proton spectrum (RAPID) and energetic electron spectrum (RAPID)

Figure 29.1 shows a collection of data from Cluster 1 during a typical Cluster dayside pass on 17 February 2003. The panels from top to bottom depict magnetic field intensity, plasma flow velocity, low energy proton spectrum (CIS), high energy proton spectrum (RAPID) and electron spectrum (RAPID). At the beginning of the depicted time interval the satellite was flying in the northern tail lobe over the polar cap towards the dayside. At about 03:00 UT the plasma density and energetic particle fluxes were greatly enhanced indicating a mixture of magnetospheric and magnetosheath plasma. By 04:00 UT the satellite had moved into a region where only magnetospheric energetic particles were present, the plasma density was low and the plasma flow velocity was small. At the same time the magnetic field intensity behaved rather smoothly without any high frequency fluctuations or turbulence. This region, which is equatorward of the cusp funnel and inside the magnetopause with high fluxes of energetic particles, is the high-latitude dayside plasma sheet, HLPS, or the high latitude trapping region. In the HLPS significant energetic particle fluxes are persistently present as will be discussed later on. At about 07:10 UT the satellite crossed the dayside magnetopause into the magnetosheath, which is seen as an abrupt change in all the depicted parameters. Most notably the magnetic field intensity dropped and became very turbulent, the flow velocity increased to a steady level, plasma density increased and energetic particles all but disappeared (except for the few enhancements mostly seen in energetic protons). The satellites do not observe the HLPS region on every orbit. If the orientation of the dayside magnetosphere relative to the Cluster orbit is suitable the satellites

29

Energetic Particles in the Cusp: A Cluster/RAPID View

419

may enter the exterior cusp. A collection of data from such an orbit is shown in Fig. 29.2. Again the panels from top to bottom depict magnetic field intensity, plasma flow velocity, low energy proton spectrum (CIS), high energy proton spectrum (RAPID) and electron spectrum (RAPID). At the beginning of the depicted time interval the satellite flied over the polar cap in the northern tail lobe and began to see plasma of increasing density as it approached the high-latitude magnetopause. After about 22:00 UT the satellite observed a region of decreased and highly fluctuating magnetic field with increased plasma density and small and turbulent plasma flow velocity. This is the cusp diamagnetic cavity. High fluxes of energetic particles (both electrons and ions) were present in the cusp as can be seen from the RAPID electron and ion spectra in Fig. 29.2. Such elevated fluxes of energetic particles are seen nearly every time the satellite observes the cusp although the flux level depends on a variety of things as will be discussed later. The satellite left the cusp around 23:00 UT skimming the dayside magnetopause and entering the magnetosheath. As the satellite exited the cusp it observed a number of sharp fluctuations in the magnetic field intensity and associated strong peaks in energetic particle fluxes from about 22:50 to 23:30 UT. Such signatures are not a feature particular to this orbit alone but are actually observed often as the satellite exits the cusp and the IMF By component dominates (like in the event under discussion). The two types of orbits discussed above are the most typical ones during vernal equinoxes. As mentioned above, whether the satellites observe the cusp or the HLPS mainly depends on the orientation of the whole dayside magnetosphere relative to the Cluster orbit. It is well known that the dipole tilt angle affects the cusp position greatly. The greater the dipole tilt angle the more equatorward the cusp is. In addition it has been shown that the cusp position is affected by IMF direction [11, 13, 17] being generally more equatorward during negative IMF Bz . It is thus expected that Cluster sees the HLPS region more preferentially when the dipole is tilted away from the Sun and the IMF is northward, while the cusp would preferentially be observed when the dipole is tilted towards the Sun and the IMF being southward. However, also the inclination of the Cluster orbit has an effect. In seven years since the launch of Cluster-II late in 2000 the inclination of the orbit has decayed from the initial 0ı to more than 30ı below the ecliptic. The orbital decay has led to the fact that the northern cusp is now rarely seen by the Clusters while in the beginning of the mission the northern cusp was seen more often. The southern cusp in contrast is now seen more often than at the beginning of the mission.

29.4 Reconnection Near the Cusp Asikainen and Mursula [2] studied the event shown in Fig. 29.2 in detail and showed that the sharp fluctuations in the magnetic field and associated energetic particle peaks seen at the cusp edge were associated with reconnection near the cusp. Figure 29.3 shows a close-up of the time interval from 22:50 to 23:30 UT from Cluster 4 satellite. The panels from top to bottom depict the magnetic field components

420

T. Asikainen

Fig. 29.2 Cluster-1 data from 2–3 February 2003. Panels from top to bottom are magnetic field intensity (FGM), plasma flow velocity (CIS), low energy ion spectrum (CIS/CODIF), energetic proton spectrum (RAPID) and energetic electron spectrum (RAPID)

in a boundary normal (LMN) coordinate system, energetic proton spectrum and energetic electron spectrum. One can see that each energetic particle enhancement occurs during a bi-polar variation in the magnetic field normal component which is a well known signature of transient reconnection at the magnetopause or a flux transfer event (FTE). In an FTE a reconnected flux tube moves along the magnetopause across the spacecraft giving rise to the bi-polar signature in the magnetic field normal component. During the time interval in Fig. 29.3 the IMF was rather steady with By D 10 nT and Bz D 3 nT. During such an IMF one expects the antiparallel reconnection to occur near the dusk edge of the cusp funnel. Asikainen and Mursula [2] showed indeed that the observations are consistent with such a reconnection and that the energetic particles observed during the FTEs were particles released from the magnetospheric closed field lines. The particles were streaming antiparallel to the reconnected flux tube towards the cusp. This work showed that reconnection near the cusp can release particles from the closed field lines of HLPS and some of these particles have access into the cusp. As mentioned, such reconnection events and similar (e.g., Zong et al. [25]) are often observed near the cusp magnetopause when IMF By dominates suggesting that energetic particle fluxes in the cusp might be dependent on IMF direction.

29

Energetic Particles in the Cusp: A Cluster/RAPID View

421

Fig. 29.3 Close-up of flux transfer events observed by Cluster-4 at 2 February 2003 22:50–23:30. Panels from top to bottom show magnetic field components in boundary normal coordinate system (BN is the normal component), energetic electron spectrum (RAPID) and energetic proton spectrum (RAPID)

29.5 Statistics of Energetic Particles in the HLPS and the Cusp In order to understand the origin of the energetic particles in the cusp we have to study the energetic particle fluxes in the adjacent HLPS also. To this end, we performed an extensive statistical analysis of energetic particles in the HLSP and the cusp which was presented in Asikainen and Mursula [1]. Since then we have revised and extended our database to cover the January–April in 2002 and 2003. For this study on each Cluster orbit in this time range we identified the HLPS (22 observations) and cusp (26 observations) regions. We calculated average energetic proton and electron fluxes measured by the RAPID at different energy channels. In addition we calculated the simultaneous average solar wind and IMF parameters as well as Kp and AE indices. For solar wind and IMF we used the high resolution (1 min) OMNI data from OMNIWeb (see http://omniweb.gsfc.nasa.gov/) in which the data have been time shifted to the bow shock nose (for the details of the time shifting procedure please refer to the OMNIWeb website). Assuming a connection between

422

T. Asikainen

the near-Earth tail and the high-latitude dayside we have to take into account the propagation time of energetic particles from the tail into the dayside. For 10s to 100s of keV particles the drift time was estimated to be at most 3 h. When we calculated the AE index averages we took the average from up to 3 h before the Cluster observation time to account for the particle drift. We found a good correlation between the energetic ion fluxes and AE/Kp indices in the HLPS at all energy channels. The correlation coefficients between ion fluxes and AE and Kp indices were between 0.52 and 0.75 at all energy channels and they were all statistically significant. Electron fluxes in the HLPS were found to correlate with solar wind speed with correlation coefficients at different energy channels ranging from 0.51 to 0.6. The electron fluxes also correlated with the magnitude of IMF By component (i.e., IMF jBy j). The correlation coefficients with IMF jBy j ranged from 0.55 to 0.6, thus the electron fluxes tend to be higher in the HLPS when IMF jBy j is small. These correlations suggest that the energetic particles observed in the HLPS adjacent to the cusp come from the inner magnetosphere and/or near-Earth magnetotail where they are accelerated by substorms (thus the ion flux-AE correlation) or processes responsible for the acceleration of radiation belt electrons (thus the electron flux-solar wind speed correlation). The transport of energetic particles especially during substorms from the near-Earth magnetotail to the HLPS/cusp has been shown possible by Delcourt and Sauvaud [7]. The negative correlation between IMF jBy j and the electron fluxes may be understood in terms of reconnection. When By dominates the reconnection has been observed to occur near the cusp edges, i.e., at the closed HLPS field lines (see the discussion in the previous section). At these times electrons which are faster than ions and bound to the field lines more tightly than ions escape along the opened field lines more readily than ions. Turning on to the cusp we found significant differences in the behavior of energetic particle fluxes compared to the HLPS region. We found that ion fluxes in the cusp correlate mainly with IMF jBy j and AE index and that the correlation depends strongly on ion energy. Figure 29.4 shows the partial correlation coefficients between energetic ion flux and log(AE) and IMF jBy j. Because partial correlation measures linear correlation we calculated the coefficient for log(AE) which displays better linear relation with ion flux than just AE. One can see that at lower energies the fluxes correlate more with IMF jBy j and at higher energies more with AE index. The crossover takes place at the second energy channel which corresponds to about 75 keV. These relationships show that the reconnection near the cusp when By dominates has a strong and statistically significant role in the release of energetic ions into the cusp from closed field lines. It seems that reconnection is the dominant factor for ions below 75 keV. At higher energies the fluxes depend more on the AE index which suggests that ions above 75 keV can more or less directly diffuse into the cusp from closed field lines. Furthermore, transport of energetic particles with 100s of keV energy from the near-Earth tail to the high-latitude dayside regions has been shown to be enhanced during substorms [7]. Energetic electrons in the cusp behave totally differently from the ions. They show correlation only with IMF Bz and the correlation coefficients are in the range

29

Energetic Particles in the Cusp: A Cluster/RAPID View

423

Fig. 29.4 Partial correlation coefficients between energetic ion flux in the cusp and log(AE) index and IMF jBy j as a function of ion energy

0.51 to 0.65 at all energy channels. Thus the electron fluxes are higher in the cusp when IMF is northward. The electrons in the cusp do not show statistically significant dependence on AE or Kp indices or solar wind speed (as was observed in the HLPS region). To understand the correlation with IMF Bz let’s consider the cusp geometry during northward and southward IMF. During northward IMF the reconnection preferentially occurs poleward of the cusp producing a more closed field line topology than during southward IMF when subsolar reconnection opens the cusp field lines. At 10s to 100s of keV energy electrons are relativistic and in an open field line topology they are lost within seconds. A more closed field line topology is expected to trap them for a longer time and more efficiently. Thus regardless of the strength of the source producing the electrons, their fluxes in the cusp are mainly determined by how efficiently they can be confined there, i.e., how closed the cusp field line topology is. To reveal a possible dependence of the electron fluxes on a source process inside the magnetosphere we separated from the data those events where IMF was northward. Figure 29.5 shows the lowest energy channel electron flux (energy about 40 keV) with respect to AE index. One can see that taking all the observations (left hand side of Fig. 29.5) there is no correlation. However, taking only events with northward IMF (right hand side of Fig. 29.5) when the cusp electron trapping is expected to be more efficient, a weak but statistically significant correlation of 0.41 between the electron fluxes and AE index is revealed.

424

T. Asikainen Correlation coefficient 0.15

102

101

100

Correlation coefficient 0.41

103 electron flux. [1/cm2 s sr keV]

electron flux. [1/cm2 s sr keV]

103

102

101 0

200

400 600 AE index [nT]

800

0

200

400 600 AE index [nT]

800

Fig. 29.5 Relationship between energetic electron flux (energy approx. 40 keV) in the cusp and AE index. Left hand side: All events included, right hand side: only events with northward IMF included

This suggests that also electrons in the cusp come from within the magnetosphere and are transported into the cusp from the near-Earth magnetotail especially during substorms.

29.6 Conclusions The results reviewed above suggest that the origin of energetic particles in the high-latitude dayside regions, both the HLPS and the cusp is in the near-Earth magnetotail. During substorms the transport of energetic particles from the tail is enhanced and this is reflected in the positive correlation between the energetic ion fluxes and AE index in the HLPS and the cusp. Detailed case studies and large scale statistics showed that two mechanisms operate in transporting energetic particles into the cusp: release by reconnection from the adjacent closed field lines and direct diffusion. For ions below 75 keV energy release by reconnection seems to dominate and for more energetic ions direct diffusion dominates. Electrons in the cusp are also transported there from within the magnetosphere but their flux is mostly determined by cusp magnetic field topology, i.e., whether cusp field lines are open (southward IMF) or closed (northward IMF). Electron fluxes are higher when the cusp field lines are closed. These observations provide a consistent explanation for the cusp energetic particles without a need for significant local acceleration of shocked solar wind plasma to MeV energies. While some energy transfer from the electromagnetic waves and turbulence does occur in the cusp it cannot explain the observations presented here.

29

Energetic Particles in the Cusp: A Cluster/RAPID View

425

Acknowledgements The author would like to thank K. Mursula for fruitful co-operation and P. Daly for useful discussions regarding the RAPID instrument. The Cluster Active Archive, FGM, CIS and RAPID PI-groups are greatly acknowledged for providing high quality data used in this paper.

References 1. Asikainen, T., Mursula, K.: Energetic particle fluxes in the exterior cusp and the high-latitude dayside magnetosphere: Statistical results from the Cluster/RAPID instrument, Annales Geophysicae, 23, 2217–2230 (2005) 2. Asikainen, T., Mursula, K.: Reconnection and energetic particles at the edge of the exterior cusp, Annales Geophysicae, 24, 1949–1956 (2006) 3. Balogh, A., Dunlop, M. W., Cowley, S. W. H., Southwood, D. J., Thomlinson, J. G., et al.: The Cluster Magnetic Field Investigation, Space Science Reviews, 79, 65–91 (1997) 4. Chang, S.-W., Scudder, J. D., Fuselier, S. A., Fennell, J. F., Trattner, K. J., Pickett, J. S., Spence, H. E., Menietti, J. D., Peterson, W. K., Lepping, R. P., Friedel, R.: Cusp energetic ions: A bow shock source, Geophys. Res. Lett., 25, 3729–3732 (1998) 5. Chen, J., Fritz, T. A., Sheldon, R. B., Spence, H. E., Spjeldvik, W. N., Fennell, J. F., Livi, S., Russell, C. T., Pickett, J. S., Gurnett, D. A.: Cusp energetic particle events: Implications for a major acceleration region of the magnetosphere, J. Geophys. Res., 103, 69–78 (1998) 6. Chen, J., Fritz, T. A., Sheldon, R., Pickett, J., Russell, C.: The discovery of a new acceleration and possible trapping region of the magnetosphere, Adv. Space Res., 27, 1417–1422 (2001) 7. Delcourt, D., Sauvaud, J.-A.: Populating of Cusp and Boundary Layers by energetic (hudreds of keV) equatorial particles, J. Geophys. Res., 104, 22 635–22 648 (1999) 8. Fritz, T. A., Chen, J., Sheldon, R., Spence, H., Fennell, J., Livi, S., Russell, C., Pickett, J.: Cusp Energetic Particle Events Measured by POLAR Spacecraft, Phys. Chem. Earth, 24, 135–140 (1999) 9. Fritz, T. A., Chen, J., Siscoe, G.: Energetic ions, large diamagnetic cavities and ChapmanFerraro cusp, J. Geophys. Res., 108, 1028, (2003) 10. Kremser, G., Woch, J., Mursula, K., Tanskanen, P., Wilken, B., Lundin, R.: Origin of energetic ions in the polar cusp inferred from ion composition measurements by the Viking satellite, Ann. Geophys., 13, 595–607 (1995) 11. Meng, C.-I., Case studies of the storm time variation of the polar cusp, J. Geophys. Res., 88, 137 (1983) 12. Onsager, T. G., and Lockwood, M.: High-latitude particle precipitation and its relationship to magnetospheric source regions, Space Sci. Rev., 80, 77–107 (1997) 13. Palmroth, M., Laakso, H., Pulkkinen, T., Location of high-altitude cusp during steady solar wind conditions, J. Geophys. Res., 106, 21109–21122 (2001) 14. Reiff, P. H., Hill, T. W., Burch, J. L.: Solar wind plasma injection at the dayside magnetospheric cusp, J. Geophys. Res., 82, 479–491 (1977) 15. R`eme, H., Bosqued, J. M., Sauvaud, J. A., Cros, A., Dandouras, J., Aoustin, C., et al.: The Cluster Ion Spectrometry (CIS) Experiment, Space Sci. Rev., 79, 303–350 (1997) 16. Roederer, J. G., Dynamics of geomagnetically trapped radiation, Springer-Verlag Berlin Heidelberg New York, pp. 66 (1970) 17. Russell, C., POLAR eyes the cusp. In R. Harris (Ed.), Proceedings of the Cluster-II Workshop: Multi-scale/Multipoint Plasma Measurements, ESA SP-44, vol. 90, pp. 47–55, European Space Agency, Noordwijk, 2000 18. Shabansky, V. P.: Some processes in the magnetosphere, Space Sci. Rev., 12, 299–418 (1971) 19. Sheldon, R., Spence, H., Sullivan, J., Fritz, T. A., Chen, J.: The discovery of trapped energetic electrons in the outer cusp, Geophys. Res. Lett., 25, 1825–1828 (1998) 20. Trattner, K., Fuselier, S., Peterson, W., Chang, S.-W., Friedel, R., Aellig, M.: Origins of energetic ions in the cusp, J. Geophys. Res., 106, 5967–5976 (2001)

426

T. Asikainen

21. Wilken, B., Axford, W. I., Daglis, I., Daly, P., Guttler, W., Ip, W. H., Korth, A., et al.: RAPID: The Imaging Energetic Particle Spectrometer on Cluster, Space Science Reviews, 79, 399–473 (1997) 22. Wing, S., Newell, P. T., Onsager, T. G.: Modeling the entry of magnetosheath electrons into the dayside ionosphere, J. Geophys. Res., 101, 13155–13167, (1996) 23. Zhang, H., Fritz, T. A., Zong, Q.-G., Daly, P. W.: Stagnant exterior cusp region as viewed by energetic electrons and ions: A statistical study using Cluster Research with Adaptive Particle Imaging Detectors (RAPID) data, J. Geophys. Res., 110, A05211, doi:10.1029/ 2004JA010562, (2005) 24. Zong, Q.-G., Fritz, T. A., Korth, A., et al.: Energetic electrons as a field line topology tracer in the high latitude boundary/cusp region: Cluster RAPID observations, Surveys in Geophysics, 26, 215–240, (2005a) 25. Zong, Q.-G., Fritz, T. A., Spence, H., Zhang, H., Huang, Z. Y., Pu, Z. Y., Glassmeier, K.-H., Korth, A., Daly, P. W., Balogh, A., R`eme, H.: Plasmoid in the high latitude boundary/cusp region observed by Cluster, Geophys. Res. Lett., 32, doi:10.1029/2004GL020960, (2005b)

Part VII

Observations of Magnetospheric Tail

Chapter 30

Plasma Flow Reversals in the Magnetotail and Their Implications on Substorm Models A.T.Y. Lui

Abstract Intervals of plasma flow reversals in the magnetotail are considered as times when the observing satellite encounters the process responsible for substorm disturbances. We consider the expected signatures during plasma flow reversals to evaluate the validity of two prevalent substorm models.

30.1 Introduction Magnetospheric substorms represent an explosive energy release phenomenon in space plasmas that can be investigated with in situ and remote sensing measurements. This unique property of magnetospheric substorms thus constitutes a window for exploration on explosive energy release phenomena in the plasma universe that cannot be studied with in situ measurements. At present, there are two prevalent substorm models, namely the near-Earth initiation (NEI) model with the substorm process located in the downstream distance of 6–15 RE and the mid-tail initiation (MTI) model with the substorm process located in the downstream distance beyond 15 RE [6]. These two models predict different substorm disturbances in the plasma sheet, in particular, around the site of plasma flow reversal of significant speeds above the ‘noise level’ of flow measurements. In this paper, we shall examine what different plasma behavior is expected around plasma flow reversals from these two models to guide us in evaluating the validity of these models.

30.2 Features Around Plasma Flow Reversal for MTI Model In the MTI model, plasma flow reversal arises from an X-line moving tailward [4]. Flow reversal implies the encounter of the X-line during its tailward motion. Figure 30.1 depicts the essential features of the model and the associated X-line A.T.Y. Lui () The Johns Hopkins University Applied Physics Laboratory, Laurel, MD 20723, USA e-mail: [email protected] H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 30, c Springer Science+Business Media B.V. 2010

429

430

A.T.Y. Lui

Near-Earth Neutral Line Model 4 Substorm current wedge

3 Braking & dawnward current

2 High-speed flow

u

1 NENL

E

By

By

v By

By

u

E

Fig. 30.1 Schematic illustrations of some major characteristics of the MTI model and the associated magnetic reconnection

reconnection. There is a quadruple magnetic perturbation perpendicular to the plane of magnetic reconnection, i.e., tailward of the X-line, Bx By < 0 in both north and south of the neutral sheet, while similarly earthward of an X-line, Bx By > 0. If plasma flow reversal were due to tailward motion of an X-line, then one would expect the product Bx By < 0 before flow reversal and Bx By > 0 after the passage of the X-line. The plasma flow pattern is well organized by the magnetic field configuration. Tailward of the X-line, the Lorentz force has a component pointing tailward, thus accelerating plasma tailward. This component of the Lorentz force changes sign at the X-line location. Therefore, plasma flow reversal should be accompanied by the x-component of the Lorentz force changing its sign. Plasma flow reversal due to

30

Plasma Flow Reversals in the Magnetotail and Their Implications

431

tailward motion of an X-line would predict the x-component of the Lorentz force changing sign simultaneous with the flow reversal. Magnetic reconnection requires dissipation, which is assumed to take place around an X-line. One therefore would expect to find the y-component of the dot product between current density and electric field jy Ey to be positive in the region. The y-component of this dot product is expected to be the largest in comparison with the other components since the current density and the electric field have the largest component in this direction normally.

30.3 Features Around Plasma Flow Reversal for NEI Model In the near-Earth initiation model, current disruption is a turbulent phenomenon and causes reduction of the cross-tail current in the magnetotail. Current disruption also reduces the magnetic field normal to the current sheet, i.e. the Bz component, at tailward and azimuthally adjacent locations [9]. This magnetic field reduction causes thinning of the current sheet, instigating further development of the physical process for current disruption. The tailward motion of current disruption initiated in the near-Earth region arises from progressive tailward thinning of the plasma sheet, possibly by the launching of rarefaction waves downstream. Plasma flow reversal during substorms corresponds to the passage of this region of multiple current disruption sites to further downstream in the magnetotail. The current disruption region is highly turbulent, with highly variable magnetic field especially near the neutral sheet, as illustrated in Fig. 30.2. The magnetic field fluctuations reveal multiscale phenomena up to the ion cyclotron frequency [3, 8]. No clear quadruple magnetic perturbation is anticipated in association with this region. However, there is a Hall current system earthward of the current disruption region. This arises from the fact that ions are unmagnetized while electrons are still magnetized and follow the collapsing field line motion in dipolarization. Consequently, electrons move Earthward with the dipolarizing magnetic field lines while the unmagnetized ions cannot follow the electron motion completely, giving rise to a Hall current system and a dynamo in the magnetotail [7]. In contrast to magnetic reconnection, the plasma flow is not organized by the magnetic field. The Lorentz force is highly dependent on changes of current density and magnetic field locally as well as its vicinity. Therefore, the x-component of the Lorentz force does not necessarily change direction with the passage of the current disruption region. Based on the particle simulation of the cross-field current instability in Lui [6], the turbulence associated with current disruption shows highly fluctuating current density with occasional reversal in its direction, so is the electric field along the current direction. Overall, the turbulence region is generally a dissipative region, i.e., jy Ey > 0, although it exhibits occasional dynamo behavior, i.e., jy Ey < 0, as well.

432

A.T.Y. Lui

1 Plasma process on auroral field line causes current disruption (CD) and substorm current wedge

E

Ict

.

.

2 rarefaction wave causes CD to spread tailward

V

V 3

Multiple CD sites 4

An X-type magnetic neutral line forms at a CD site

Fig. 30.2 Schematic illustrations of some major characteristics of the NEI model and the current disruption region

30.4 Comparison of Observations with Model Predictions One notable work using the above characteristics to evaluate the two substorm models was reported by Lui et al. [11]. They investigated the possible signatures expected from these models around a plasma flow reversal event (from Vx < –800 km=s to Vx > 800 km=s) observed by Cluster satellites during a substorm on 22 August 2001. They found (1) no clear quadruple magnetic perturbation associated with this flow reversal, (2) the x-component of the Lorentz force does not reverse simultaneously with flow reversal, (3) the presence of episodic dynamo effect, i.e., jy Ey < 0, in the general region of flow reversal, and (4) the expected sequence of flow reversal among the Cluster satellites to be consistent with multiple tailward moving plasma flow sources and not with a single tailward moving plasma flow source. These features are all consistent with the NEI model and not with the MTI model. However, this is only one example and the general applicability of this result needs to be examined with statistical study of such plasma flow reversal events.

30

Plasma Flow Reversals in the Magnetotail and Their Implications

433

30.5 Summary and Discussion This paper discusses the various features that one may use on observations of plasma flow reversals to evaluate the two prevalent substorm models. Since both current disruption and magnetic reconnection sites are known to be very localized, the choice of plasma flow reversal observations eliminates the possible argument that negative results of a model prediction are due to the observing site not being at the location where these localized phenomena occur. Recently, observations of highly variable magnetic field components, especially the Bz component, accompanied by tailward plasma flow are interpreted as undeniable evidence of magnetic reconnection (e.g., Sergeev et al., [14]). However, this interpretation ignores the fact that current disruption can produce tailward plasma flow and highly fluctuating magnetic field as well. Lui et al. [10] have provided a criterion to determine the direction of plasma acceleration due to the force imbalance resulting from current disruption. Furthermore, a crude agreement of the quadruple magnetic perturbation in the current disruption region can result from the mere fact that electrons, being lighter than ions, can leave the current disruption region faster than the ions, producing a crude trend of the quadruple magnetic perturbation similar to the magnetic reconnection region but in addition superposed by large deviations from this trend. It should be noted that almost all simulations of magnetic reconnection, especially those two-dimensional ones, produce very laminar magnetic field configuration, nothing resembling the high magnetic fluctuations seen in current disruption events. In terms of physical process for the two substorm models, tearing instability is presumed to be the agent for the MTI model to create the magnetic field geometry. However, the viability of tearing instability onset is still an open question. Although there are recent efforts to enable its onset [15, 17], the proposed current sheet features required to initiate the instability have not been verified observationally at the presumed magnetic reconnection site in the magnetotail. Assuming the existence of magnetic reconnection without addressing how it occurs is highly unsatisfactory in understanding the substorm onset process. For the NEI model, there are many potential candidates for the onset process, which include the various forms of ballooning instability, magnetosphere-ionosphere coupling, cross-field current instability, entropy anti-diffusion instability, current-driven Alfv´enic instability, convection reduction, and “akis” structure in a thin current sheet. There are some efforts already attempting to identify precisely the onset process from observations [1, 2, 5, 12, 13, 16]. Acknowledgements This work was supported by the NASA grant NNX07AQ50G.

References 1. Chen, L-J, Bhattacharjee A, Sigsbee K, Parks G, Fillingim M, Lin R (2003) Wind observations pertaining to current disruption and ballooning instability during substorms, Geophys. Res. Lett., 30, 68–1, doi 10.1029/2002GL016317.

434

A.T.Y. Lui

2. Cheng, C Z and Lui A T Y (1998) Kinetic ballooning instability for substorm onset and current disruption observed by AMPTE/CCE, Geophys. Res. Lett., 25, 4091–4094. 3. Consolini, G, Kretzschmar M, Lui A T Y, Zimbardo G, Macek W M (2005) On the magnetic field fluctuations during magnetospheric tail current disruption: A statistical approach, J. Geophys. Res., 110, A07202, doi:10.1029/2004JA010947. 4. Hones, E. W., Jr. (1979), Transient phenomena in the magnetotail and their relation to substorms, Space Sci. Rev., 23, 393–410. 5. Lui, A T Y (1996) Current disruption in the Earth’s magnetosphere: Observations and models, J. Geophys. Res., 101, 13067–13088. 6. Lui, A. T. Y. (2001), Current controversies in magnetospheric physics, Rev. of Geophys., 39 (4), 535–564. 7. Lui, A T Y and Kamide Y (2003) A fresh perspective of the substorm current system and its dynamo, Geophys. Res. Lett., 30 (18), 1958, doi:10.1029/2003GL017835. 8. Lui, A T Y and Najmi A-H (1997) Time-frequency decomposition of signals in a current disruption event, Geophys. Res. Lett., 24, 3157–3160. 9. Lui, A T Y, Chang C-L, Mankofsky A, Wong H-K, and Winske D (1991) A cross-field current instability for substorm expansions, J. Geophys. Res., 96, 11389–11401. 10. Lui, A T Y, Yoon P H, and Chang C-L (1993) Quasi-linear analysis of ion Weibel instability, J. Geophys. Res., 98, 153–163. 11. Lui, A T Y, Zheng Y, Zhang Y, R`eme H, Dunlop M W, Gustafsson G, Mende S B, Mouikis C, and Kistler L M (2006) Cluster observation of plasma flow reversal in the magnetotail during a substorm, Ann. Geophys., 24, 2005–2013. 12. Roux, A, Perraut S, Robert P, Morane A, Pedersen A, Korth A, Kremser G, Aparicio B, Rodgers D, Pellinen R (1991) Plasma sheet instability related to the westward traveling surge, J. Geophys. Res., 96, 17697–17714. 13. Saito, M H, Miyashita Y, Fujimoto M, Shinohara I, Saito Y, and Mukai T (2008) Ballooning mode waves prior to substorm-associated dipolarizations: Geotail observations, Geophys. Res. Lett., 35, L07103, doi:10.1029/2008GL033269. 14. Sergeev, V, Semenov V, Kubyshkina M, Ivanova V, Baumjohann W, Nakamura R, Penz T, Runov A, Zhang T L, Glassmeier K-H, Angelopoulos V, Frey H, Sauvaud J-A, Daly P, Cao J B, Singer H, and Lucek E (2007) Observation of repeated intense near-Earth reconnection on closed field lines with Cluster, Double Star, and other spacecraft, Geophys. Res. Lett., 34, L02103, doi:10.1029/2006GL028452. 15. Sitnov, M I, Malova H V, and Lui A T Y (1997) Quasineutral sheet filamentation instability induced by electron preferential acceleration from stochasticity, J. Geophys. Res., 102, 163–173. 16. Voronkov, I, Rankin R, Frycz P, Tikhonchuk V T, and Samson J C (1997) Coupling of shear flow and pressure gradient instabilities, J. Geophys. Res., 102, 9639–9650. 17. Zelenyi, L, Artemiev, A, Malova H, and Popov V (2008) Marginal stability of thin current sheets in the Earth’s magneotail, J. Atmos. Solar-Terr. Phys., 70, 325–333.

Chapter 31

Spatial and Temporal Structures in the Vicinity of the Earth’s Tail Magnetic Separatrix Cluster Observations E.E. Grigorenko, L.M. Zelenyi, M.S. Dolgonosov, and J.-A. Sauvaud

Abstract The processes of non-adiabatic particle acceleration occurring in the vicinity of magnetic neutral lines produce highly accelerated (up to 2,500 km/s) field-aligned ion beams. Ion beams are usually observed in the Plasma Sheet Boundary Layer (PSBL) i.e. in the vicinity of magnetic separatrix between open and closed magnetic field lines. These acceleration processes have different spatial and temporal scales depending on the topology of the magnetic field and local plasma conditions in the Current Sheet (CS) and their peculiarities produce distinct manifestations in various characteristics of accelerated PSBL ion structures. Multipoint Cluster observations have led to a new understanding of these phenomena since they give an opportunity to separate spatial and temporal effects in registration of accelerated plasma structures. Our observations in the PSBL of the Earth’s magnetotail revealed two different types of ion beam populations. The first type represents energy collimated ion beams having a duration of several minutes and energies < 20 keV. Along with such ion beams isotropic electron distributions are observed. Multipoint observations of such type of beams revealed that they represent long-living plasma filaments. We suggest that such ion beams were accelerated by the quasi-steady dawn-dusk electric field in a number of spatially localized regions in the CS located in wide vicinity earthward from the distant X -line. Electron observations confirm that these phenomena occur in a region with closed magnetic field geometry. The simultaneous operation of, at least, two resonant sources is evident from the observation in PSBL of intersection of two ion beams with very different parallel velocities. Another type of distributions represents powerful (with energies up to 10s of keV) ion beams having velocity distributions wide in parallel energies. They are observed along with the anisotropic electron velocity distributions forming by counterstreaming electron beams. This feature is peculiar for the magnetic separatrix and indicates that the spacecraft crossed yet open or recently closed magnetic field

E.E. Grigorenko (), L.M. Zelenyi, and M.S. Dolgonosov Space Research Institute of RAS, Moscow, Profsoyuznaya str, 84/32, Russia e-mail: [email protected]; [email protected] J.-A. Sauvaud Centre d’Etudie Spatial des Rayonnements, Toulouse cedex4, France

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 31, c Springer Science+Business Media B.V. 2010

435

436

E.E. Grigorenko et al.

lines. Such ion beams may be produced by spatially extended “merged” acceleration source which is formed due to the overlapping of initially spatially separated resonances near X-line.

31.1 Introduction The manifestations of powerful acceleration processes occurred in the Current Sheet (CS) of the Earth magnetotail are often observed in Plasma Sheet Boundary Layer (PSBL) as field-aligned ion beams streaming from the source with high velocities ( 1;000 km=s). It is widely believed that PSBL ion beams are non-adiabatically accelerated in the CS in the region where ion motion already cannot be described by the guiding center approximation and experiences a chaotic character. This phenomenon has been addressed in many theoretical papers which considered ion beam acceleration by the dawn–dusk electric field either near the quasi-stationary distant X -line [8, 13, 21, 23] or in the regions where the normal component of the magnetic field, BZ , has a finite constant or slowly varying positive value [4, 12, 27]. In the last case, it was suggested that the magnetotail has a “stretched” magnetic configuration with the X -line located in its distant part (at jX j 100 RE ) so a wide region of finite positive BZ may exist in the distant CS earthward from the X -line. In such case, ion acceleration takes place in the region with closed magnetic field lines. Later, the resonant nature of the non-adiabatic ion interaction with the CS has been revealed q [1, 5]. Particle motion depends on the so-called parameter of

adiabaticity æD R0 introduced by Buchner and Zelenyi [4] (where R is the radius of curvature of the magnetic field line crossing the CS and 0 is the maximum value of particle’s Larmor radius). If æ 1:0 ion motion consists of fast oscillations across the neutral plane (along Z) and of slow gyration in Bn(X ) field in (XY) plane. Particle motion can be described by the action invariant over the fast variH able IZ D 1=2 zPdz. When a particle crosses the CS plane (Z D 0) the value of IZ experiences stochastic jumps and IZ D 3=2 æ.X / ln j2 sin j, (where is the phase of fast component of particle motion) and IZ has very sensitive dependence on . Since each particle intersects the CS plane twice: at the entry to and at the exit from the CS, the IZ experiences two jumps during each CS crossing. If the phase gain between entry to and exit from the CS becomes equal N (where N is an arbitrary integer number called the resonance number), the remarkable effect happens: IZ 0 and the dependence on random phase disappears. So for local CS parameters which support the resonant condition D N stochastic jumps are exactly compensated and motion of particle becomes effectively regular. So there are certain regions in the CS where particle dynamics escapes chaotization and particles follow the almost unperturbed classical Speiser orbits and thus, form field-aligned PSBL ion beams. The dependence on æ is very non-monotonic and can result in a phase space structuring. In this case, ion distribution consists of small-scale ion beams which are localized in phase and physical space. Such smallscale spatially localized ion beams were initially predicted in a model developed

31

Spatial and Temporal Structures in the Vicinity

437

by Ashour-Abdalla et al. [1] and were called “beamlets”. According to this theory, beamlets are produced by the spatially localized resonant sources simultaneously operating in the distant CS. Later observations of beamlets in auroral PSBL and in the magnetotail confirmed this theory [10, 29]. It is natural to assume that the properties of velocity distributions of charged particles observed in PSBL may have an imprint of spatial and temporal characteristics of the magnetic and electric fields in the acceleration source(s). To explain the peculiarities of ion velocity distributions Onsager et al. [16] suggested field-aligned ion acceleration at an extended CS source with approximately a constant value of the magnetic field everywhere in the source. Thus, ion distributions in the nearEarth PSBL are controlled only by the velocity-filter (VF) effect due to magnetotail convection which produces the equatorward drift of accelerated ions while they are moving from the source to the observation point. Ashour-Abdalla et al. [1], on the contrary, proposed that generation of field-aligned beamlets may simultaneously takes place in several spatially localized resonances located in the wide vicinity earthward of distant X -line. The efficiency of ion acceleration at each resonance depends on local characteristics of the magnetic field BZ .X / which generally depends on the distance from the Earth. As a result energy gain of ions may vary with the location of their acceleration site (so-called place of birth effect). This scenario simultaneously accounts for two effects: (1) the global latitude dispersion of emerging structure qualitatively similar to the one considered for example by Takahashi and Hones [24]; (2) fragmentation of this structure on a set of a small-scale substructures each of them might experience their own dispersion [9]. An estimation of temporal scales of PSBL ion beam is another difficult task, especially for one-point observations. Earlier measurements of plasma characteristics in PSBL reported the short duration of ion beams (few minutes). It was assumed that ion beams have transient nature which could be addressed either to the transient character of their acceleration in the source or to the spatial effect due to the relative motion of PSBL boundary (so-called flapping). Some authors considered ion beams as mostly spatial (i.e. very long living) structures [1] and explained their short duration by fast magnetotail flappings. Other authors attributed short observation of PSBL beams to their impulsive generation in the CS with the characteristic time scales 1–2 min [3]. The impulsive character of acceleration provides time of flight effect (TOF) which spreads ions having different energies along the magnetic field line so that more energetic ions are observed first. This phenomenon was considered as origin of Time-Dispersed Ion Structures (TDIS) which are often registered in auroral PSBL during active periods [18]. But, in presence of convection the TOF unavoidably produces the latitudinal (spatial) energy dispersion. As a result, it is almost impossible to separate spatial and temporal effects when a spacecraft moves in one direction towards the NS. The temporal nature of TDIS is easier to identify when a spacecraft moves from the NS towards the lobe. In this case VF and TOF effects produce an opposite character of ion energy dispersion. This analysis was used by Sauvaud and Kovrazhkin [19] in order to identify spatial and temporal characteristics of ion dispersed structures observed in PSBL at low altitudes. Grigorenko et al. [7] applied the similar analysis to Cluster observations in magnetotail PSBL

438

E.E. Grigorenko et al.

and showed that the typical duration of PSBL ion beam during quiet and moderately disturbed periods is 5–15 min. Because of the slow motion of spacecraft in apogee of its orbit, rapid cuts of the PSBL-Plasma Sheet (PS) region are nonavailable. Thus for reliable identification of spatial and temporal characteristics of ion beam one should compare the values of ion field-aligned velocity observed at the lobeward edge of PSBL in its successive crossings by the spacecraft. In case of frequent flapping oscillations such analysis may help to identify transient effects in ion acceleration. In this paper we present two examples of PSBL ion beams which are the manifestations of transient and quasi-steady character of ion acceleration. We will show that the presence of transient effects in ion velocity distributions observed in PSBL may be related with the proximity of acceleration source to the magnetic separatrix. In our analysis, we will use measurements of ion three-dimensional velocity distribution functions collected every 12 s by the hot ion analyzers (HIA) onboard Cluster-1 (Cl-1) and Cl-3 and the HC velocity distribution functions compiled for 8 s by composition and distribution function analyzer (CODIF) onboard Cl-4 [17]. The magnetic field data were obtained with 4 s resolution by the fluxgate magnetometer FGM [2]. We will also use electron data provided by Plasma Electron and Current Experiment (PEACE) [9].

31.2 Observations of Magnetic Separatrix Crossing and Transient Effects in PSBL Plasma Distributions On 12.09.2001 at 15:05 UT Cluster was located in the southern lobe at [19.2, 2.3, 1.8] RE (GSE) and at 15:07 UT all Cluster spacecraft encountered PSBL. During the period of interest the magnetic activity was moderate (AE 300 nT). Cl-4 observations from the CODIF, PEACE and FGM instruments between 15:05 and 15:12:30 UT are shown in Fig. 31.1. Ion and electron velocity distribution functions displayed in (V? ; Vjj ) plane for selected time sequences indicated by arrows are presented in the upper part of the Fig. 31.1. HC and electron Energy-Time spectrograms and three components of the magnetic field are presented in the three upper panels. The bottom panel of the figure shows the values of HC field-aligned velocities Vjj corresponding to the maximum of velocity distribution functions which are displayed by black dots together with a half-maximum widths of the corresponding velocity distribution functions displayed by the vertical lines crossing each dot. jBX j=jBj time profile displayed by grey line is also shown in this panel. At 15:07 UT Cl-4 crossed the lobeward edge of PSBL and observed high-velocity field-aligned ion beam streaming earthward with Vjj D 1;700 km=s (see “1i” velocity distribution) and having the half-maximum width of velocity distribution function Vjj 600 km=s Vjj =Vjj D 0:35 . Before this moment at 15:06:14 UT a tailward electron beam started (see “1e” velocity distribution). Three spins later, at 15:06:26 UT the earthward streaming electron beam with higher energy appeared (2e) and became stronger at 15:06:30 UT (3e). Simultaneously the energy

Spatial and Temporal Structures in the Vicinity 3000

0

15:07:40 (2i)

0

–3000 –3000 0 –3000 3000 –3000 0 V⊥, km/s V⊥, km/s

V||, km/s

V||, km/s

2x104

0

15:06:26 (2e)

0

3000

2x104

V||, km/s

15:06:14 (1e)

2x104

e-

3000

15:10:52 (3i) Log (Count) 2.0 1.7

0

1.3

–3000 –3000

0 3000 V⊥, km/s

15:06:30 (3e)

2x104

V||, km/s

15:07:08 (1i)

439

V||, km/s

3000

V||, km/s

Cl-4 H+

V||, km/s

31

0

15:07:38 (4e) dEF 10

0

10

7

6

–2x104 –2x104 –2x104 0 2x104 –2x104 0 2x104 –2x104 0 2x104 –2x104 0 2x104 V⊥, km/s V⊥, km/s V⊥, km/s V⊥, km/s 4e 2i 1e 2e 3e 1i 3i 4 Log (Count) 10

Energy, eV

–2x104 –2x104

103 102

H+ CODIF

104 103 102

2.5 1.8 1.1 0.4

dEF 2 eV/cm s str eV 7

10

6

10

nT10

BX BY BZ

0 –10 –20

3000

V||, km/s

1.0 2000

0.8 0.6

1000 0 15:05

|Bx|/|B|

0.4 15:06

15:07

15:08 15:09 Time (UT)

15:10

15:11

15:12

Fig. 31.1 Experimental observation of magnetic separatrix crossing on 12.09.2001 between 15:05–15:12:30 UT Cluster was located at [19.2, 2.3, 1.8] RE (GSE). From top to bottom: ion and electron velocity distribution functions in (V? ,Vjj ) plane, measured by CODIF and by PEACE instruments (onboard Cl-4) respectively at the moments indicated by the arrows; E-T spectrograms of ions and electrons; three components of the magnetic field and values of ion field-aligned velocity corresponding to the maximums of velocity distribution functions and displayed by black dots. Half maximum widths of velocity distribution functions along the magnetic field direction are shown by vertical lines crossing the corresponding dots. In the same panel the time profile of BX =jBj is shown by grey line. The intervals of field-aligned ion beam registrations are shaded by pink color

of tailward electron beam increased, so that, by the moment of field-aligned ion beam registration the counterstreaming velocity distribution consisting of two fieldaligned electron beams was formed. It became more hot and isotropic inside the PSBL region (4e). These features of electron velocity distributions indicate on the magnetic separatrix crossing by Cluster spacecraft.

440

E.E. Grigorenko et al.

Inside the PSBL but still close to its lobeward edge (jBX j=jBj was of the order of 1.0) field-aligned ions with lower parallel velocities were registered. Then, 40 s after the first encounter with the ion beam, when Cl-4 was already inside PSBL (jBX j=jBj decreased) it observed sudden increase of ion beam energy (Vjj D 1;900 km=s) followed by the widening (Vjj D 1;460 km=s and Vjj =Vjj 0:8) of its velocity distribution function (see “2i” distribution measured at 15:07:40 UT). At this time Cl-1 and Cl-3 also observed the same energetic ion beam (not shown). Simultaneously, electron energy increased up to a few keV, which is typical for the PS. Then, between 15:07:56–15:08:28 UT and between 15:08:36–15:11:41 UT Cl-4 registered more isotropic ion velocity distributions without signatures of field-aligned beams, like one observed at 15:10:52 UT (3i). It is worth to note that Cl-3 which was located farther from the NS registered field-aligned ion beam at 15:08:01, 15:08:37, 15:09:49, 15:11:01 UT, i.e. then Cl-4 observed only PS-like ion velocity distribution. This ion beam had the characteristics of velocity distribution function (Vjj and Vjj ) similar to ones observed by Cl-4 at 15:07:48 UT. This indicates that ion beam acceleration probably had rather quasi-steady character. However the observations of sudden increase of ion beam energy registered by all spacecraft at 15 W 07 W 40 UT may be related with transient effects superimposed on quasi-steady acceleration.

31.3 PSBL Manifestations of Quasi-Steady Resonant Acceleration of Ions in the CS 31.3.1 Event of 21 September 2001 On 21.09.2001 at 19:45 UT Cluster was located in the northern lobe (at Œ16; 3; 5:5 GSE ) and moved equatorward. This was magnetically very quiet day: AE index did not exceed 50 nT. Figure 31.2 shows Energy-Time spectrograms and velocity distribution functions in (V? , Vjj ) plane of ions (HIA data) and electrons obtained by Cl-1 and Cl-3 between 19:44–19:57 UT. The observations provided by Cl-4 (from CODIF instrument) are similar to ones measured by Cl-3 and are not presented. The bottom panel of Fig. 31.2 shows the values of ion field-aligned velocities Vjj corresponding to the maximums of velocity distribution functions measured by three Cluster spacecraft and displayed by the colored dots. Although the values of Vjj provided by Cl-1 and Cl-3 were measured by HIA instruments while the values of Vjj given by Cl-4 were measured by CODIF instrument the differences in energy, angular and time resolutions between these instruments do not significantly affect our results. At 19:45 UT Cl-3 and Cl-4, which were located closer to the PS, encountered PSBL and observed field-aligned energy collimated ion beam streaming earthward with Vjj D 650 km=s.Vjj =Vjj 0:2/. During 2 min of its observation the parallel velocity of ion beam does not change (see the bottom panel of Fig. 31.2). In this

Spatial and Temporal Structures in the Vicinity 19:48:01 (2i,e) 1x10

V||, km/s

Cl-1 Ions (HIA)

3

0

–1x10

3

4

0

2x10

4

4

–2x10

4

4

0

0 4

–2x10

V⊥, km/s 2i,e 3i,e

1e

2

eV/cm s str eV

0

10

7

10

6

0 4

–2x10 4

2x10

19:51:14 (4e)

2x10

–2x10

V⊥, km/s

4

4

2x10

–2x10

V⊥, km/s 4e

0

4

2x10

V⊥, km/s

104 103 102

3

–1x10

01x10

3

3

–1x10

V⊥, km/s 4

0 4

2x10

V||, km/s

V||, km/s

2x10

– 2x10

0 4

– 2x10 4

–2x10

0

4

2x10

V⊥, km/s 1e

–2x10

0

0 3

–1x10

3

4

2x10

V⊥, km/s 2i,e 3i,e

4

3

19:48:00 (4e)

4

–2x10

0

2x10

V⊥, km/s 4e

3

3

3

–1x10

0 1x10

4

0 4

4

4

–2x10

0

4

0 4

2x10

2x10

0 4

– 2x10 4

–2x10

V⊥, km/s

4

2x10

– 2x10 4

3

0 1x10

V⊥, km/s

2x10

– 2x10 4

1.2 1.0 0.7 0.0

0

–1x10

–1x10 2x10

0

1x10

V⊥, km/s

4

– 2x10 4

0 1x10

Log (Count)

3

1x10

V⊥, km/s

V||, km/s

19:44:03 (1e)

3

19:56 19:55:04 (6i,e)

3

0

–1x10 3

–1x10

19:54

19:49:52 (5i,e) V||, km/s

1x10

0

19:52

V||, km/s

V||, km/s

1x10

19:50

19:46:51 (3i,e) 3

V||, km/s

19:48

19:45:51 (2i,e)

V||, km/s

19:46 3

V||, km/s

19:44

Cl–3 Ions (HIA)

V||, km/s

Energy, eV

dEF

4

2x10

0

–2x10

3

V⊥, km/s

V||, km/s

V||, km/s

V||, km/s

4 4

Energy, eV

3

–1x10 0 1x10

4

0

–2x10

4

3

3

0 1x10

2x10

–2x10

V||, km/s

3

V||, km/s

4

2x10

0

V⊥, km/s 19:46:32 (1e)

2x10

e–

1x10

–1x10

–1x10

e–

441

19:48:49 (3i,e)

3

V||, km/s

31

0

4

2x10

V⊥, km/s

0 4

– 2x10 4

–2x10

0

4

2x10

V⊥, km/s 6i,e 7e

5i,e

19:56:05 (7e)

104 103 102

4

–2x10

0

4

2x10

V⊥, km/s

Log(Count) 2.7 2.0 1.4 0.4

V||, km/s

1500 Cl–1 Cl–3 Cl–4

1000 500 0 19:44

19:46

19:48

19:50

19:52

19:54

19:56

Time, UT

Fig. 31.2 Experimental observation of quasi-steady energy collimated ion beam on 21.09.01 between 19:44–19:57 UT. From top to bottom: ion and electron velocity distribution functions in (V? ,Vjj ) plane, measured by HIA and PEACE instruments respectively onboard Cl-1, E-T ion spectrogram obtained by Cl-1; the same data provided by Cl-3. The bottom panel shows the values of ion field-aligned velocities (Vjj ) derived from velocity distribution functions measured by HIA instruments onboard Cl-1 and Cl-3 and by CODIF instrument onboard Cl-4 and indicated by the colored dots. Half maximum widths of velocity distribution functions along the magnetic field direction are shown by vertical lines

event, contrary to the previous one, isotropic electron velocity distribution functions were observed not only together with the ion beam but also above it, in higher latitude region. It is worth to note, that electrons observed in this PSBL crossing are colder than in the previous event and have energies <1 keV. Cl-1 was located farther from the PS than Cl-3 and Cl-4 and it started to observe PSBL ion beam only at 19:47:40 UT. By this time Cl-3 and Cl-4 stopped to observe the ion beam. Ion velocity distribution functions provided by Cl-1 between 19:47:40 and 19:50:30 UT are similar to ones obtained in the previous observations by Cl-3

442

E.E. Grigorenko et al.

and Cl-4. Since no transient effects were observed in ion beam characteristics during the interval 19:45–19:50:30 UT one may suggest that one and the same ion beam was successively observed by three Cluster spacecraft. Interestingly to note that Cl-3 and Cl-4 did not registered the ion beam during the interval of its observation by Cl-1. Between 19:46:30–19:49:30 UT, instead of high velocity field-aligned ion beam, Cl-3 and Cl-4 registered the increase of energy of cold lobe ion population. The similar increases but with lower amplitudes were also seen in E-T ion spectrogram obtained by Cl-1 within 19:45–19:47 UT and 19:49–19:52 UT intervals. Cold lobe plasma traveling tailward consists mainly of ions of ionospheric origin (e.g. [22]). Normally this ion population is very cold and can be repelled by the spacecraft potential. But sometimes when these ions experience large drifts, their energy correspondingly increases up to a few keV. These phenomena have been often found inside the PSBL and near its lobeward edge especially during the intervals of highvelocity ion beam propagation (e.g. [7, 11, 20]) and could be considered as the “remote sensing” of the PSBL boundary disturbances. Near the high-latitude edge of PSBL Kelvin-Helmholtz instability may be responsible for such transient increases of ion perpendicular velocity (e.g. [20]). Since the energy of cold ion population increases due to the increase of their perpendicular velocity this phenomenon does not influence on the characteristics of field-aligned ion beam velocity distributions which are analyzed in the present paper and it is not responsible for the formation of double-peaked field-aligned velocity distributions which will be discussed in the next section. Between 19:49:30–19:56 UT Cl-3 and Cl-4 again observed the ion beam having the same field-aligned velocity (650 km=s) as in the previous intervals. By this time Cl-1 stopped to observe it. It is worth to note, that during the entire interval of interest electron velocity distribution functions were almost isotropic and no accelerated electron beams were observed. An interesting feature is that the “ion PSBL” with the ion beam was located within the isotropic electron layer. Electron energy inside the “ion PSBL” was only slightly higher than in the adjacent high-latitude region which was void of high-energy ions. Thus, in this event the quasi-steady energy collimated ion beam was observed during at least 11 min. Grigorenko et al. [7] demonstrated that the periodic observations of this ion beam by Cluster spacecraft were due to the kinking of magnetic filed lines containing beam ions in the PSBL plane. According to their analysis this ion beam was strictly localized in the direction perpendicular to the main magnetic field (0:4 RE ).

31.3.2 PSBL Signatures of Multiple Acceleration in the CS During the quiet (AE < 100 nT) and moderately disturbed periods (AE < 300 nT) the phenomenon of ion beam intersection is sometimes observed in PSBL. In such cases the spacecraft observes simultaneously high-velocity (jVjj j 2;000 km=s) and

31

Spatial and Temporal Structures in the Vicinity

443

low-velocity (600 km=s jVjj j 1; 500 km=s) field-aligned ion beams forming “double-peaked” distributions in velocity space [29]. These ion populations are clearly separated in velocity space (with the gap of the order of one or more energy channels). One of such event observed in PSBL on 24.09.2001 between 04:38–04:46 UT is presented in Fig. 31.3. The magnetic activity from 02:00 UT until the end of the day was rather low (AE 150 nT). At 04:35 UT Cluster was located at [16, 3, 6] RE (GSE) and moved from the northern lobe towards the PS. At 04:38:30 UT Cl-3 located closer to the PS encountered PSBL and after a few seconds Cl-1 also crossed its lobeward boundary. At the lobeward edge of PSBL all spacecraft observed only high-speed ion beam streaming earthward with Vjj D 2; 100 km=s. However, as the spacecraft was penetrating deeper to the PSBL it started to observe ion beam with lower velocity (Vjj D 1; 400 km=s) simultaneously with the high-speed one. At this time (04:40–04:46 UT) Cl-3 was located in deep PSBL since it observed counterstreaming ion beams but Cl-1 was still in the outer part of PSBL and observed only earthward ions (it sank to the deep PSBL at 04:41 UT). But in distribution functions measured by both spacecraft the “double-peak” feature was clearly observed in ion component streaming earthward. It is worth to note that during this interval Cl-1 measured ion velocity distribution functions in bursty mode, i.e. every 4 s. Even if one assumes the high velocity of flapping (100 km=s) [7] then for such short interval the spacecraft passes in PSBL Z 0:06 RE . Thus with high precision one may consider this phenomenon as ion beam intersection in space. Since the parallel velocities of two ion beams forming double peaked velocity distribution functions were the same during the entire interval of interest, we may assume that this feature existed in ion distributions for at least 6 min. Thus it has rather spatial than temporal origin. Such kind of ion velocity distribution functions observed in PSBL is the manifestation of simultaneous operation in the CS of, at least, two acceleration

04:40:03

24.09.01 104

Counts

2000 0

Cl-1

–2000

80 60 40 20 0

–2000 –2000 0 2000

80 60 40 20 0

V||, km/s V⊥, km/s V⊥, km/s V||, km/s V||, km/s 04:41:20 04:40:59

0 1000 2000 V||, km/s

All ions

102 04:38

04:40

04:42

Time (UT)

04:44

04:46

0

2000 0 –2000 –2000 0 2000

V⊥, km/s

80 60

80 60 40 20 0

2000 0

2000 0 –2000 0 1000 2000 V||, km/s

–2000 0 2000

V⊥, km/s

Counts

103

2000

80 60

Counts

104

40 40 Log(Count) 20 20 3.6 –2000 –2000 0 0 2.7 –2000 0 2000 –2000 0 2000 0 1000 2000 0 1000 2000 Cl-3 1.8 V||, km/s V||, km/s V⊥, km/s V⊥, km/s 0.9 V km/s V||, km/s 04:42:08 04:45:28 ||,

Counts

102

All ions

Counts

Energy (eV)

0 0 1000 2000

–2000 0 2000

103

Cl-3 V||, km/s 04:40:44 2000

Counts

Cl-1 V||, km/s

80 60 40 20 0

0 1000 2000 V||, km/s

Fig. 31.3 Experimental observation of ion beam intersection in PSBL on 24.09.01. In the left part of the figure E-T ion spectrograms provided by HIA instruments onboard Cl-1 and Cl-3 are shown. In the right part of the figure ion velocity distribution functions in (V? , Vjj ) plane and their 1D cuts along the magnetic field direction measured by Cl-1 (two left columns) and by Cl-3 (two right columns) are presented. Vertical lines plotted in 1D cuts of ion velocity distributions indicate p statistical errors calculated as N (where N is total count number)

444

E.E. Grigorenko et al.

sources. Because, due to VF effect ion beams with significantly different velocities observed at the close Z locations can not be accelerated at the one and the same source. Below we present the simulation based on the model of ion resonant acceleration in CS and demonstrate the possibility of intersection of ion beams generating in two simultaneously operating sources.

31.3.3 Modeling of Ion Beam Intersection In this section we discuss the possibility of intersection of ion beams which were resonantly accelerated in two spatially separated sources (resonances). We assume that such resonances operate simultaneously during the time which is much larger than the ion time of flight from the source to the observation point. Thus, we can neglect TOF effect in ion beam propagation. In this case only VF and place of birth effects will influence on ion distribution in the near Earth PSBL [1]. In the equations presented below we will use cgs system of units in which plasma velocity is defined as V D cE=B; (where c is the velocity of light, E and B are the electric and magnetic fields respectively). Zelenyi et al. [30] shown that at N -th resonance (N is integer number of resonance) ions gain energy WN : WN D 2mEY

C1 ˛

4=3

1=2

N

4=3

D 2N

4=3

em1=2 EY B0 C1

!2=3 (31.1)

where m and e are proton mass and charge respectively, EY is large-scale dawndusk electric field, B0 is the value of magnetic field in the lobe region, C1 is the constant evaluated theoretically (C1 D 0:761), ’ D eL=2mc2 EY B0 and L is the characteristic spatial scale of resonance (it was taken as L D 3 RE ). It is naturally to assume that if the distance between two CS resonances is enough large, then the difference in spatial displacements of two ion beams due to different path lengths along the magnetic field lines may be compensated by the difference of their energies. In other words, the action of VF effect is compensated by the place of birth effect. This results in intersection of two beams at some (X; Z) location in PSBL. This is rather rare but very fortuitous combination of parameters allows one to detect simultaneously two ion beams accelerated at different distances from the Earth. As it was shown by Zelenyi et al. [29] the essential difference in ion beam energies may be provided by their acceleration over resonance, i.e. at resonances with either even or odd numbers. The distance X between resonances with numbers (N 1) and (N C 1) is: X D

C1 ˛

1=.qC1/ ..N C 1/1=.qC1/ .N 1/1=.qC1/ /

(31.2)

31

Spatial and Temporal Structures in the Vicinity

445

where q is free parameter of the model (q D 0:35). The time of flight of ions from the (N 1)-th and (N C 1)-th resonances to the observation point may be estimated correspondingly as: tN 1 D XN 1 =VN 1 and tN C1 D .XN 1 C X / =VN C1 , where XN 1 is the distance between the observation point and (N 1)-th resonances, VN1 and VNC1 are field-aligned velocities of ion beams accelerated at (N 1)-th and (N C 1)-th resonances correspondingly. During ion beam propagation from the source to the observation point ions drift towards the NS with the velocity VD D cEY =B0 and, thus, experience the displacements: ZN 1 D

cEY XN 1 B0 VN 1

(31.3)

cEY XN C1 B0 VN C1 where XN C1 D XN 1 C X: ZN C1 D

(31.4)

Assuming the conservation of the magnetic flux, the distance Z between the magnetic tubes containing beams accelerated at (N 1)-th and (N C 1)-th resonances (see Fig. 31.4) can be estimated under the natural condition: N 2 as: 1 Z D B0

Z

C1 N C 1 q ln Bn .X /dx D B0 .q C 1/ ˛ N 1

(31.5)

X

where Bn .X / is the normal component of the magnetic field in the CS. Finally, the condition of intersection of these beams at the certain distance from the Earth (X ) is Z C ZN 1 D ZN C1 Combining Eqs. 31.3–31.5 we could estimate the distance X as: XD

where b D

Z

˛ C1

q.N 2 1/3=2 ln.N C 1=N 1/ C b.N 1/2=3 .N C 1/ .N 1/ , D

D2

eLE2 Y

mB0 C12

1=3 .

Beam “N+1”

B0 ΔZ B0

1 , 1Cq

(31.6)

ΔZN+1

ΔZN–1 Cluster X Towards the Earth

Bn (XN–1) Beam “N–1” XN–1

Fig. 31.4 Scheme of intersection of two ion beams in PSBL of magnetotail

Bn (XN+1) XN+1

446

E.E. Grigorenko et al.

For the parameters of our simulation model, the ion beams ejected, for example, from the fourth and sixth resonances will intersect at X 14 RE from the Earth. This numerical example is just given as a possibility to illustrate that ion beams generated at different resonances can intersect in the magnetotail PSBL. Playing with the numbers of resonances one may obtain the location of ion beam intersection farther downtail. We used this simple 2D model only for qualitative comparison with Cluster observations.

31.4 Discussion and Summary The main purpose of this paper is to specify two different types of accelerated ion beams observed in PSBL and to relate them to the peculiarities of particle acceleration in the CS. We assume that ions forming PSBL ion beams were accelerated in the CS by the Speiser mechanism [12]: initially cold ions of mantle and/or ionospheric origin non-adiabatically interact withqthe CS (in that its part, where the value of local magnetic field is so small that æ D

R 0

1:0, and there acquire energy:

W .X / D E Y .X / D E 2i .X / D 2mi

cE Bn .X /

2

(where, ¡i is the ion Larmor radius at the final stage of acceleration (it should not be confused with an initial Larmor radius ¡0 of mantle ions coming to the CS); mi is ion mass; E is the value of the dawn-dusk electric field which for simplicity is assumed to be constant, Bn .X / is the value of the local magnetic field normal to the plane of the CS which on average depends on X ). Buchner and Zelenyi [4], Burkhart and Chen [5], Ashour-Abdalla et al. [1] considered the resonant character of ion interaction with the CS and revealed the existence of spatiallylocalized acceleration sites (called resonances) at which accelerated ions move along non-perturbated Speiser orbits almost without scattering and form small-scale ion beamlets in the PSBL while in the other region of the CS, accelerated ions are trapped inside the PS due to strong scattering. Thus, ion beams accelerated at spatially separated resonances for different values of Bn .X /, provide PSBL structuring in both configurational and velocity space. Zelenyi et al. [31] estimated the minimum width in velocity space (Vjj ) of ion beam accelerated at each resonance as Vjj =Vjj 0:1.Wjj =Wjj 0:2). It is worth to note that energy resolutions of HIA and CODIF instruments (W/W) are 0.17 and 0.13 correspondingly, i.e. less than the theoretical estimate of the energy width of individual ion beam accelerated at a particular resonant source. Thus the finite energy resolution of the instruments does not influence strongly on our analysis of PSBL ion beams. Let us consider the possible relation of ion beam observed on 21.09.2001 to this acceleration scenario. As it was shown above, this ion beam is characterized by energy collimated velocity distribution function (Vjj =Vjj 0:2), which may indicate

31

Spatial and Temporal Structures in the Vicinity

447

on its acceleration in spatially localized resonant CS source. Within such source the B.X/ gradient of the magnetic field d lndX should be small (at least in quiet magnetotail configuration). Thus, initially cold ions entering the CS are passing approximately the same Y across the tail and obtain approximately the same energy value (here we take into account that initial energies of ions are much smaller than their energy gain due to acceleration). The observed long duration of this ion beam (11 min) without transient effects indicates the quasi-steady character of its acceleration in the CS. This situation may be realized in the quiet magnetotail configuration which is confirmed by the fact that this event was observed during very quiet geomagnetic period. Isotropic electron velocity distributions observed on 21.09.2001 at the highlatitude edge of the PSBL and inside the PSBL, together with the ion beam, obviously indicate that ion acceleration takes place in the region of closed magnetic field lines with a small but positive BZ (as was proposed by Lyons and Speiser [12], Zelenyi et al. [27]). Energies of electrons observed together with the ion beam are significantly lower than ones observed in the Central Plasma Sheet (< 1 keV). So that we may suggest the following acceleration scenario of such quasi-steady energy collimated ion beam (see Fig. 31.5a): ions are accelerated at spatially localized resonant sources scattered in the wide vicinity of the distant X -line in the region of

a Z

PSBL X-line

X

W

b Z

PSBL

X X-line

Fig. 31.5 Two possible scenarios of ion non-adiabatic acceleration in the magnetotail CS. (a) Quasi-steady resonant acceleration at spatially localized sources (resonances) located in wide vicinity of distant X-line with finite Bz > 0. Each resonance produces energy collimated ion beam with energy W (shown by the colored line) increasing with the distance from the Earth (red color represents most energetic ion beam). Since ion acceleration takes place at closed magnetic field lines ions beams are observed inside the layer of isotropic electrons (shown by the light-yellow color). (b) Ion beam acceleration in the single spatially extended source located near X-line. The source generates ion beam consisting from ions of different energies which form in PSBL wide in parallel energy ion velocity distribution and are observed together with anisotropic electron velocity distributions (shown by the light-pink color)

448

E.E. Grigorenko et al.

already closed magnetic field lines (with finite BZ > 0). The acceleration process lasts, at least, several minutes and has rather a quasi-steady than transient character. The simultaneous quasi-steady operation of several resonances in the CS leads to the possibility of observation in the PSBL of set of localized ion beams with different velocities VN (where VN is the velocity of the ion beam accelerated in some particular N -th resonance). This was first demonstrated in numerical simulations by Ashour-Abdalla et al. [1] and finally proved experimentally by Zelenyi et al. [29]. In this paper we presented another example of observation (on 24.09.01) the double-peaked velocity distribution functions formed by two ion beams moving with different field-aligned velocities and intersecting in PSBL at X D 16 RE . It is worth emphasizing that this model implicitly assumes the operation of a quasistationary reconnection process but it does not associate any particular features of the acceleration process with steady reconnection itself. For such a model, reconnection is just a topological process which creates the magnetic geometry of a stationary model. Now let us consider ion beam observed in PSBL on 12.09.2001. The anisotropic electron velocity distributions consisting of low-energy beam streaming tailward (i.e. towards the acceleration source) and high-energy beam streaming earthward (i.e. from the source) which were observed at the lobeward edge of PSBL indicate that Cluster spacecraft crossed the magnetic separatrix. The crossing of magnetic separatrix had occurred just before the registration of accelerated ion beam, i.e. ion acceleration source was located closer to the X-line than in the previous case. The anisotropic electron velocity distribution functions consisting of a cold electron component counterstreaming in respect of the accelerated ion beam and a hot electron component moving in the same direction as the ion beam were observed in magnetotail PSBL by Nagai et al. [14], Fujimoto et al. [6], Nakamura et al. [15]. The authors related such distributions to the field-aligned current due to the Hall effect in the diffusion region of magnetic X-line which was located rather close to the observation point. In the event considered here the reflected accelerated electron beam was observed together with the “direct” ion beam. One may roughly estii , mate the distance from the point of observation to the X-line as: X D .V2XV e Vi / where X is the distance from the observation point to the Earth, Vi and Ve are correspondingly ion and electron field-aligned velocities estimated from the velocity distribution functions. For the event observed on 12:09:2001X 3 RE (so X-line was located at X 22 RE ). As we mentioned above the ion beam observed in PSBL on 12.09.2001 has comparatively wide in parallel energies velocity distribution function (Vjj =Vjj 0:35 0:8). It is natural to suggest that near the X -line, the gradient of magnetic field d ln B.X/ may be so large that its spatial scale becomes comparable or even smaller dX than the Larmor radius of accelerated ions. In this case, the initially spatially separated resonances overlap and form a single (“merged”) spatially extended acceleration source. Ions coming to such a source, propagate in the CS different distances Y and, thus, obtain different energy gains. This results in formation in PSBL of ion velocity distributions wide in parallel energies (Fig. 31.5b). Moreover, the transient increase of ion beam energy observed around 07:40 UT by all Cluster spacecraft

31

Spatial and Temporal Structures in the Vicinity

449

located inside PSBL may indicate on the importance of inductive electric fields for ion acceleration. Such energy growth along the spacecraft trajectory cannot be explained neither by VF nor by TOF effects, since VF effect results in energy decrease in the direction from PSBL to the NS. Also for the TOF effect high energy ions should arrive before the low-energy ones. The phenomenon of the appearance of more energetic ions after the low energy ones (inverse energy dispersion) was theoretically explained earlier by Taktakishvili and Zelenyi [25], Taktakishvili et al. [26] as a result of fast growth of inductive electric field at the late stage of the acceleration process. It is worth also to note that this powerful acceleration has stopped after 07:40 UT since ions with Vjj 2;000 km=s were not observed during the remaining time interval. In the conclusion we would like to notice that ion beams observed in PSBL on 12.09.01 and on 21.09.01 may be related with two types of PSBL manifestations of powerful plasma acceleration processes occurring in the CS of the Earth magnetotail. We speculate that these manifestations may be related with different magnetic topologies associated, in the first case, with X-line located in mid- or near-Earth tail and in the last case – in distant magnetotail. As a result different spatial and temporal peculiarities of particle acceleration may be realized. Although the observations of accelerated plasma in magnetotail PSBL and in auroral PSBL at low altitudes are visually dissimilar, we would like to emphasize some important common features. Quasi-steady ion beams like one observed on 21.09.01 have long duration and may be a part of Velocity Dispersed Ion Structures (VDIS) precipitating in auroral PSBL. According to the previous studies VDIS have spatial origin with durations about 10s of minutes, are produced in the sources located in the distant CS and are observed during quiet or moderately disturbed geomagnetic periods together with electrons having typical PSBL energies 1 keV [19, 28]. All these features were presented in event of 21.09.01. On the contrary, the transient features observed in PSBL ion distributions on 12.09.01 may result in formation of complicated dispersed structures in the auroral PS resembling TDIS. TDIS are registered mostly during active periods and have a source located in near-Earth or mid- magnetotail. The typical duration of these transient structures is 1–3 min. The ion energies in TDIS are 14 keV and electrons have energies several keV [18, 19]. The similar features were observed in PSBL ion beam event on 12.09.01. One may assume that two observations of PSBL ion beams presented in this paper represent two extreme cases of ion acceleration in the magnetotail: (i) near X-line which was probably located in the mid – tail and (ii) earthward from the distant X-line, in the region of closed magnetic field lines. The dynamic processes leading to switching between these two types of ion beams present special interest and require additional studies. Acknowledgements The authors would like to thank Cluster CIS and FGM teams for providing data, Dr. C. Owen for providing electron data. This work was supported by INTAS grant Nr. 06–1000017–8943; RFBR grants Nr. 06–02–72561; 07–02–00319 and by grant of HIII472.2008.2.

450

E.E. Grigorenko et al.

References 1. Ashour-Abdalla M., J.P. Berchem, J. Buchner, L.M. Zelenyi (1993) Shaping of the magnetotai from the mantle: global and local structuring, J. Geophys. Res., 98, 5651. 2. Balogh A., et al. (2001) The Cluster magnetic field investigation: Overview of in-flight performance and initial results, Ann. Geophys., 19, 1207. 3. Baumjohann W., G. Paschmann, N. Sckopke, C.A Cattell, C.W. Carlson (1988) Average ion moments in the plasma sheet boundary layer, J. Geophys. Res., 93, 11,507. 4. Buchner J. and Zelenyi L.M. (1986) Determenistic chaos in the dynamics of charged particles near a magnetic field reversal, Phys. Lett., A, 118, 395. 5. Burkhart G.R. and J. Chen (1991) Differential memory in the Earth’s magnetotail, J. Geophys. Res., 96, 14033. 6. M. Fujimoto, T. Nagai, N. Yokokawa, Y. Yamade, T. Mukai, Y. Saiti, S. Kokubun (2001) Tailward electrons at the lobe-plasma sheet interface detected upon dipolarizations, J. Geophys. Res., 106, A10, 21,255–21,262. 7. E.E. Grigorenko, J.-A. Sauvaud, L.M. Zelenyi (2007) Spatial-Temporal characteristics of ion beamlets in the plasma sheet boundary layer of magnetotail. J. Geophys. Res., 112, A05218, doi:10.1029/2006JA011986. 8. M. Hoshino, T. Mukai, T. Yamamoto, S. Kokubun (1998) Ion dynamics in magnetic reconnection: Comparison between numerical simulation and Geotail observations, J. Geophys. Res., 103, A3, 4509–4530. 9. Johnstone A.D., Alsop C., Burge S., Carter P.J., Coates A.J., Coker A. J., Fazakerley A.N., Grande M., Gowen R.A., Gurgiolo C., Hancock B.K., Narheim B., Preece A., Sheather P.H., Winningham J.D., Woodliffe R.D. (1997) PEACE: a Plasma Electron and Current Experiment, Space Sci. Rev., 79, 351–398. 10. Keiling A., H. Reme, I. Dandouras, J.M. Bosqued, G.K. Parks, M. McCarthy, L. Kistler, E. Amata, B. Klecker, A. Korth and R. Lundin (2004) Transient ion beamlet injections into spatially separated PSBL flux tubes observed by Cluster-CIS, Geophys. Res. Lett., 31, L12804, doi:10.1029/2004GL020192. 11. A. Keiling, G.K. Parks, H. Reme, I. Dandouras, M. Wilber, L. Kistler, C. Owen, A.N. Fazakerley, E. Lucek, M. Maksimovic, N. Cornilleau-Wehrlin (2006) Energy-dispersed ions in the plasma sheet boundary layer and associated phenomena: Ion heating, electron acceleration, Alfven waves, broadband waves, perpendicular electric field spikes and auroral emissions, Ann. Geophys., 24, 2685–2707. 12. Lyons L.R. and T.W. Speiser (1982) Evidence for current sheet acceleration in the geomagnetic tail, J. Geophys. Res., 87, 2276. 13. Martin R.F. Jr. (1986) Chaotic particle dynamics near a two-dimensional magnetic neutral point with application to the geomagnetic tail, J. Geophys. Res., 91 11,985. 14. T. Nagai, I. Shinohara, M. Fujimoto, M. Hoshino, Y. Saito, S. Machida, T. Mukai, Geotail observation of the Hall current system: Evidence of magnetic reconnection in the magnetotail (2001) J. Geophys. Res., 106, A11, 25,929–25,949. 15. Nakamura R., W. Baumjohann, T. Nagai, M. Fujimoto, T. Mukai, B. Klecker, R. Treumann, A. Balogh, H. Reme, J.A. Sauvaud, L. Kistler, C. Mouikis, C. J. Owen, A.N. Fazakerley, J.P. Dewhrust, Y. Bogdanova (2004) Flow shear near the boundary of the plasma sheet observed by Cluster and Geotail, J. Geophys. Res., 109, A05204, doi:10.1029/2003JA010174. 16. Onsager T.G., M.F. Thomson, R.C. Elphig, J.T. Gosling (1991) Models of electron and ion distributions in the plasma sheet boundary layer, J. Geophys. Res., 96, 2099. 17. Reme H., C. Aoustin, J.M. Bosqued, I. Dandouras et al. (2001) First multispacecraft ion measurements in and near the Earth’s magnetosphere with identical Cluster ion spectrometry (CIS) experiment, Annales Geophysicae, 19, 1303. 18. Sauvaud J.-A., D. Popescu, D.C. Delcourt, G.K. Parks, M. Brittnacher, V.A. Sergeev, R.A. Kovrazhkin, T. Mukai, and S. Kokubun (1999) Sporadic plasma sheet ion injections into the high altitude auroral bulge – satellite observations, J. Geophys. Res., 104, 28,565.

31

Spatial and Temporal Structures in the Vicinity

451

19. Sauvaud J.-A., R.A. Kovrazhkin (2004) Two types of energy-dispersed ion structures at the plasma sheet boundary, J. Geophys. Res., 109, A12213, doi:10.1029/2003JA010333. 20. Sauvaud J.-A., P. Louarn, G. Fruit, H. Stenuit et al. (2004) Case studies of the dynamics of ionospheric ions in the Earth’s magnetotail., J. Geophys. Res., 109, A01212, doi:10.1029/2003JA009996. 21. Savenkov B.V., L.M. Zelenyi, M. Ashour-Abdalla, and J. Buchner (1991) Regular and chaotic aspects of charged particle motion in a magnetotail-like field with a neutral line, Geophys. Res. Lett., 18, 1587. 22. Seki K., M. Hirahara, T. Terasawa, T. Mukai, Y. Saito, S. Machida, Y. Yamamoto, S. Kokubun (1998) Statistical properties and possible supply mechanism of tailward cold OC beams in the lobe/mantle regions, J. Geophys. Res., 103, 4477. 23. Speiser T.W. (1965) Particle trajectories in the model current sheet based on the open model of the magnetosphere, with applications to auroral particles, J. Geophys. Res., 70, 1717. 24. Takahashi K. and E.W. Hones (1988) ISEE 1 and 2 observations of ion distributions at the plasma sheet-tail lobe boundary, J. Geophys. Res., 93, 8558. 25. A.L. Taktakishvili, L.M. Zelenyi (1990) Bursts of energetic protons in the Earth’s magnetotail. Proceedings of the Joint Varenna-Abastumani-ESA-Nagoya-Potsdam Workshop on “Plasma Astrophysics” held in Telavi, Georgia. ESA SP-331, 51–55. 26. A.L. Taktakishvili, L.M. Zelenyi, E.T. Sarris, R.E. Lopes, D.V. Sarafopoulos (1993) Temporal dispersion structures of proton and electron bursts in the Earth’s magnetotail. Planet. Space Sci., 41, no. 6, 461–467. 27. Zelenyi L.M., J. Buchner and D.V. Zogin (1988) Quasiadiabatic ion acceleration in the Earth’s magnetotail, paper presented at Varenna-Abustumani Workshop on Plasma Astrophysics, Eur. Space Agency, Publ. SP-285, 227. 28. Zelenyi L.M., R.A. Kovrazhkin and J.M. Bosqued (1990) Velocity-dispersed ion beams in the nightside auroral zone: AUREOL-3 observations, J. Geophys. Res., 95, 12119 29. Zelenyi L.M., E.E. Grigorenko, J.-A. Sauvaud and R. Maggiolo (2006) Multiplet structure of acceleration processes in the distant magnetotail, Geophys. Res. Lett., 33, L06105, doi:10.1029/2005GL024901. 30. Zelenyi L.M., Dolgonosov M.S., Grigorenko E.E., Sauvaud J.-A. (2007) Universal properties of the nonadiabatic acceleration of ions in Current Sheets, JETP Letters, 85, 4, 187–193. 31. Zelenyi L.M., M.S. Dolgonosov, E.E. Grigorenko, J.-A. Sauvaud (2008), Fine structure of PSBL plasma process: filamentation, nonlinear and dispersive effects, in Future Perspectives of Space Plasma and Particle Instrumentation and International Collaborations edited by M. Hirahara, I. Shinohara, Y. Miyoshi, N. Terada, and T. Mukai, American Institute of Physics, pp.5–14, 2009.

Chapter 32

Cluster Observations of Energy Conversion Regions in the Plasma Sheet O. Marghitu, M. Hamrin, B. Klecker, K. R¨onnmark, S. Buchert, L.M. Kistler, M. Andr´e, and H. R`eme

Abstract Cluster allows for the first time a systematic examination of energy conversion, by the evaluation of the power density, E J, with E the electric field and J the current density. Following a careful inspection of the Cluster plasma sheet data from the summer and fall of 2001, we selected 43 energy conversion regions (ECRs), out of which 26 concentrated load regions (CLRs, E J > 0) and 17 concentrated generator regions (CGRs, E J < 0). As expected in the tail, at about 19 RE geocentric distance, the energy conversion is more intense for CLRs, on average some 25 pW=m3 , compared to some 5 pW=m3 for CGRs. The CLRs are located closer to the neutral sheet and dominated by E and J in the GSE y direction, unlike the CGRs, that prefer locations towards the plasma sheet boundary layer, where the deviations of E and J from the GSE y direction can be significant. The ECRs are often associated with high speed bulk flows, on average faster and hotter for

O. Marghitu () Institute for Space Sciences, Bucharest, Romania and Max-Planck-Institut f¨ur extraterrestrische Physik, Garching, Germany e-mail: [email protected] M. Hamrin Department of Physics, Ume˚a University, Ume˚a, Sweden and Swedish Institute of Space Physics, Ume˚a, Sweden B. Klecker Max-Planck-Institut f¨ur extraterrestrische Physik, Garching, Germany K. R¨onnmark Department of Physics, Ume˚a University, Ume˚a, Sweden S. Buchert and M. Andr´e Swedish Institute of Space Physics, Uppsala, Sweden L.M. Kistler Space Science Center, University of New Hampshire, Durham, USA H. R`eme CESR-CNRS, Toulouse, France

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 32, c Springer Science+Business Media B.V. 2010

453

454

O. Marghitu et al.

CLRs. The CLRs appear to be associated also with density drop and sometimes with temperature anisotropy, Tk > T? , features which are observed less frequently for CGRs.

32.1 Introduction Although the plasma sheet behaves, on average, as a load, because of the large scale dawn–dusk electric field and cross-tail current, its structure is highly inhomogeneous. The energy stored in the magnetic field is supposed to be converted into plasma kinetic and thermal energy at rather localized acceleration sites, and then transported away, by means of e. g. the bursty bulk flows (BBFs [1]). A fraction of the energy carried towards the Earth is re-converted into Poynting flux (e.g. [2]), and eventually powers the aurora (e.g. [10]). Concentrated generator regions have been recently identified in the plasma sheet boundary layer by means of Cluster data [7,8], consistent with simulation results on energy conversion and transport in the tail [4]. This paper is a first attempt towards a systematic investigation of the energy conversion in the plasma sheet, to make use of the multi-point capabilities of the Cluster mission. We focus on a set of 43 energy conversion regions (ECRs), most of them located near the spacecraft apogee (at 19RE geocentric distance), selected from the Cluster plasma sheet crossings in 2001. Energy conversion appears often to be quite structured, and the traversal of the ECRs in our set, essentially in the GSE –z direction, takes of the order of 10 min – to be compared with the several hours plasma sheet crossing. Therefore, we call the ECRs either concentrated load regions (CLRs, E J > 0) or concentrated generator regions (CGRs, E J < 0). The word ‘concentrated’ refers specifically to the z direction, but a case study analysis [8] suggests that the CGRs might be ‘concentrated’ as well in the other directions, and in time. More extended ECRs are also visible in the data occasionally, but a quantitative assessment is difficult in such cases because of the large relative errors in the small electric fields and currents. In order to evaluate E J we use data from three experiments: the ion spectrometer CIS [9], the electric field instrument EFW [6], and the fluxgate magnetometer FGM [3]. The CIS experiment provides a proxy for E, based on the frozen-in condition, E D –V B, to be derived both from the CODIF and HIA sensors (with, respectively without mass resolution). The EFW experiment measures only the electric field in the spacecraft spin plane, and uses the assumption that the parallel electric field is zero, E B D 0, in order to derive the third component of E. Since in the plasma sheet B is close to the spacecraft spin plane, the third component of E is typically not available. Although EFW does not provide the full E vector, the partial redundancy between CIS and EFW is useful to validate the events. The FGM magnetic field data allows the evaluation of the current density, via the curlometer method [5]. The reference system to be used throughout the paper is GSE. The reader is referred to [8] for a detailed discussion on the evaluation of E J and on the selection of the reference system.

32

Cluster Observations of Energy Conversion Regions in the Plasma Sheet

455

32.2 Observations

<E>

2120

2200

−19.2 0.9 1.4 19.2

−19.2 1.0 1.2 19.2

[nT]

2

J

h i j

10

7

2

10000 1000 100

10

1.0 0.8 0.6 0.4 0.2 0.0 300 100

2

SC4 SC3 SC1 Z Y

−100 −300 4000 3000 2000 1000 0

X Tperp Tpar

10 0

Z

−10 −20 −30

Y

4 2

Z

X

Y

0

X

−2 2.0 1.5 1.0 0.5 0.0 6 4 2 0

Z COD X COD Y EFW Y COD

−2 −4 4 2 0

Z COD X COD T COD Y EFW Y COD

20 0

Z COD X COD T COD Y EFW Y COD

−2 −4 −6

−20 −40 hhmm 2210 2001 Sep 19 −18.4 GSEX 5.5 GSEY −0.4 GSEZ DIST 19.2

eV/cm -s-sr-eV

+

[eV]

SC1 H +

−3

H n

[cm ] [km/s]

H

+

[mV/m]

Y COD T COD Y EFW X COD Z COD

3

<EJ>

[pW/m ]

Y COD T COD Y EFW X COD Z COD

3

<∑ EJ>

Y COD Y EFW X COD Z COD

g

∇.B/|∇?B|

∇.B/|∇?B|

X

[nA/m ]

Y

[pW/m ]

j

f

<E>

J

2

X

[mV/m]

Z

e

3

[nT]

Y

[nA/m ]

Z

h i

+

Tpar

e

g

d

<EJ>

+

[eV]

H

Tperp

H

X

[eV]

Y

[pW/m ]

+

[km/s]

H

Z

c

3

−3

SC1

d

f

b

SC3

c

CGR Event

a

<∑ EJ>

+

H n

2

10 SC4

[cm ]

2.0 1.5 1.0 0.5 0.0 600 400 200 0 −200 3000 2000 1000 0 20 10 0 −10 15 10 5 0 −5 −10 2.0 1.5 1.0 0.5 0.0 4 2 0 −2 −4 30 20 10 0 −10 400 300 200 100 0 −100 hhmm 2120 2001 Sep 07 GSEX −19.1 GSEY 0.8 GSEZ 1.5 DIST 19.2

7

2

+

[eV]

SC1 H

b

10

10000 1000 100

[pW/m ]

CLR Event

a

eV/cm -s-sr-eV

Figure 32.1 illustrates the data used for the systematic investigation, as well as some prominent features, with a CLR event on Sept 7, 2001, around 21:45 UT, and a CGR event on Sept 19, 2001, around 22:15 UT.

2220

2230

−18.3 5.5 −0.5 19.2

−18.3 5.5 −0.6 19.1

Fig. 32.1 Left: CLR event. Right: CGR event. All quantities are averaged in time, over 24 s, and some in space, over the available spacecraft. (a) Time-energy spectrogram for SC1. (b) Density for the spacecraft where CIS was operational. (c) Bulk velocity. (d) Parallel and perpendicular temperature. (e) Magnetic field. (f, g) Current density and ratio div B/curl B, used as quality index for the current estimate. Over a good fraction of the CLR event SC3 probes the close vicinity of the neutral sheet, while the other spacecraft are still above the neutral sheet (not shown), which makes the application of the curlometer problematic. However, since divB/rotB is small, there is a good chance that J is reasonably correct. (h) Ex (blue), Ey (green), and Ez (red), as derived from CODIF data, and Ey (cyan) as derived from EFW data. (i, j) Ex Jx , Ey Jy , Ez Jz (same colors as in panel (h)), E J (black), and the cumulative sums. The vertical bands indicate the ECRs, around 21:45 UT in the CLR plot and around 22:15 UT in the CGR plot. The HIA data are similar to the CODIF data and not shown. One notes the pretty good agreement between the CODIF and EFW y quantities in the bottom three panels. The peak (absolute) value of E J is 20 pW=m3 for the CLR and 5 pW=m3 for the CGR. The CLR is located close to the neutral sheet, unlike the CGR, located near the plasma sheet boundary (e). Both cases show high-speed flow in the x direction (c). The CLR shows also density drop (b) and temperature anisotropy, Tk > T? (d)

456

O. Marghitu et al.

The CLR event was observed near midnight and close to the neutral sheet (e), during the growth phase of a modest substorm .AE 300 nT/, after a prolonged quiet time. This explains the high background density (b), n > 1 cm3 , and low ion temperature (d), T < 1 keV. E J and its cumulative sum based on CODIF (i, j), HIA (not shown), and EFW (i, j cyan) data are comparable, with maximum values of 20 pW=m3 , and step variations of 300 pW=m3 . The main contribution to E J is provided by Ey Jy . The CLR is associated with high speed plasma flow, Vx Š 500 km=s (c), density drop (b), and temperature anisotropy (d). The CGR event was observed in the pre-midnight sector, in the southern plasma sheet boundary layer. E J reaches a peak (absolute) value of 5 pW=m3 and its cumulative sum has a step variation of 30 pW=m3 . Similar to the CLR event, there is good agreement between the CODIF, EFW, and HIA data (not shown), and the dominant contribution to E J is made by Ey Jy . The CGR is associated with lower velocity flow and with little temperature anisotropy. Some of the results based on the examination of the full set of events are presented in Fig. 32.2. Each panel includes two plots, one for CLRs and one for CGRs,

1000 800

HIA/EFW

b HIA/EFW

a 200

CLR StepEyJy

150

150

CGR StepEyJy

25 CLR MaxEJ (codif+hia)/2

100 600

15

100

400

HA EFW

50

200 0 0

−50

200 400 600 800 1000

0

0 0

50

100

−50

150 200

1

CODIF

c

3 <MaxEJ> = 6 pW/m Median = 4

10 5

0 CODIF

3 <MaxEJ> = 29 pW/m Median = 23

50

HA EFW

CGR MaxEJ (codif+hia)/2

20

6

11

16

21 26

1

6

11

16

d 200 CGRStepEyJy vs StepEJ 150

1000 CLRStepEyJy vs StepEJ 800 600

800

800

600

<|MaxV|> = 500 km/s

400 200

100

600

200 0

0

400

COD HA EFW

200

COD HA EFW

50 0

0

0 0

200 400 600 800 1000

50

100

150 200

−50

e

−200 0

50

100

150

−400 −600 −800

<|MaxV|> = 250 km/s

400

−200 0

5

10

15

20

25

−400

CLR MaxV vs. MaxEJ

−600 −800

CGR MaxV vs. MaxEJ

f

7

CLR Temp. vs. MaxEJ

6

7 6

5

5

4 3

4 3

2

CGR Temp. vs. MaxEJ

100

0

50

100

150

CLR with Tpar=Tperp

20

CGR with Tpar>Tperp

CGR with Tpar=Tperp

15 10 5

= 3.0 keV

1

0

0

0

CLR with Tpar>Tperp

50

2 = 3.8 keV

1

25

150

0

5

10

15

20

25

0

1 3 5 7 9 11 13 1 4 7 10

1 2 3 45 1 3 5

7 9 11

Fig. 32.2 CLR and CGR features (left and right, respectively, in each panel). (a) Step variation in the cumulative sum (SVCS) of Ey Jy , as derived from EFW and HIA data, versus CODIF. (b) Maximum E J for each event. (c) Contribution of Ey Jy to E J. (d, e) Maximum velocity and temperature versus maximum E J. (f) Association with Tk > T? . E J, Ey Jy , and the respective SVCS are given in pW=m3 . In the two histogram type panels, (b) and (f), each event is represented by a vertical bar, of height equal to the maximum E J, while the horizontal axis shows the event count, in decreasing E J order

32

Cluster Observations of Energy Conversion Regions in the Plasma Sheet

457

in order to emphasize the differences. Since the two subsets are not very big, we have not included further refinements, e. g. with respect to geomagnetic activity. The values of the step variation in the cumulative sum (SVCS) of Ey Jy derived from CODIF, HIA, and EFW data, are compared in panel (a). The agreement is good for CLRs, where the correlation coefficient and slope of the linear regression are 0.97/0.93 and 0.96/0.94, respectively, for the CODIF–HIA/CODIF–EFW points. For CGRs, where the slope drops to 0.90/0.64, with a correlation coefficient of 0.83/0.98, the agreement is not as good. The difference is presumably related to the substantially smaller magnitude of the energy conversion in CGRs (b), which makes the respective measurements less reliable. While for CLRs the typical peak E J is a few 10 pW=m3 , for CGRs it is just a few pW=m3 . The CGRs appear to be more ‘elusive’ entities that prefer locations towards the plasma sheet boundary layer, where the fluctuations of E and J are more significant. A quantitative evaluation has not been performed so far, but based on the Bx profile, 10 out of the 17 CGRs were found to be located ‘away from the neutral sheet’ and 7 ‘close to the neutral sheet’. The relationship between the SVCS of Ey Jy , computed from CODIF/HIA/EFW data, and the SVCS of E J, computed from CODIF data (c), is not very tight, with correlation coefficient 0.75/0.74/0.76 and slope 1.55/1.68/1.03. The CLRs, on the other hand, tend to be located near the neutral sheet, where E and J are oriented predominantly in the y direction (c). 18 out of the 26 CLRs were identified ‘close to the neutral sheet’ and only eight ‘away form the neutral sheet’, six of which had very low magnitude .< 10 pW=m3 /. The prevalence of Ey Jy is well reflected by correlation coefficients and slopes equal to 0.89/0.86/0.93 and 1.00/0.96/1.07, respectively, for the dependence between the SVCS of Ey Jy and E J. Nevertheless, the CLR with the highest peak E J, not included in panel (c), is dominated by Ez Jz . Energy conversion is often associated with high speed bulk flows (d), on average faster for CLRs than for CGRs. The average speed is 500 km=s in 23 CLR events, compared to just 250 km=s in 12 CGR events; for three CLRs and five CGRs the plasma flow is rather weak. The examination of the proton distribution function in a few cases shows that, for CGRs, the bulk flow is often field-aligned, while for CLRs the E B drift prevails – consistent with the respective locations of the CGRs and CLRs ‘away from’ and ‘close to’ the neutral sheet. No obvious tendency is seen in the flow speed, with respect to the magnitude of the energy conversion, except that most of the time the speed is positive, as expected at 19RE . An outstanding exception is, again, the CLR event with the highest peak E J, coincident with a major intensification of a substorm (the AE index jumps from 750 to 1;500 nT), which makes this event a good candidate for a detailed case study. With one exception (of just 4 pW=m3 ), the CLRs associated with high speed flows show also density drops, similar to the BBFs. While this suggests a strong relationship between CLRs and BBFs, the details are still to be explored. Unlike the CLRs, only 5 CGRs of the 12 associated with high speed flows show also density drops. The average temperature (e) is also higher for CLRs than for CGRs, 3.8 keV compared to 3.0 keV. In a general sense, the larger flow speeds and temperatures for CLRs are consistent with the expectations, since the CLRs behave as sources,

458

O. Marghitu et al.

while the CGRs as sinks, for the mechanical energy, in bulk kinetic or thermal form. For 15 CLRs (more than half of the cases) and five CGRs the temperature is anisotropic, Tk > T? (f), to our knowledge a feature not pointed out so far. For CLRs, when the plasma is accelerated locally by the Lorentz force, whose work is equal to (J B/V D .–V B/J Š E J; Tk > T? could be related to a faster thermalization in parallel direction, with a corresponding increase in Tk . If the observation is made close enough to the acceleration site, the anisotropy is still visible, otherwise the distribution becomes fully thermalized, with Tk D T? . The temperature anisotropy is more difficult to explain for CGRs, although it is conceivable that energy conversion is in general associated with wave-particle interactions, that can result in Tk > T? .

32.3 Prospects For additional insight on energy conversion more case study work is required, as well as further systematic investigation, e. g. of the temperature anisotropy and density drop. In order to improve the data processing speed and the statistics, an automated ECR recognition procedure is needed. Such a tool would enable a more comprehensive examination of the plasma sheet data, as well as of data from other regions, like the boundary layers at the magnetospheric flanks. The THEMIS mission opens new possibilities for the investigation of energy conversion. The data from the current disruption region make a perfect candidate in this sense, provided that a method to derive the current from only three satellites will be developed. Acknowledgements O.M. acknowledges support through the PECS contract ECSTRA, C98048, and the kind hospitality of Max-Planck-Institut f¨ur extraterrestrische Physik, Garching.

References 1. Angelopoulos, V., Baumjohann, W., Kennel, C. F., et al.: Bursty bulk flows in the inner central plasma sheet. J. Geophys. Res. 97, 4027–4039 (1992) 2. Angelopoulos, V., Chapman, J. A., Mozer, F. S., et al.: Plasma sheet electromagnetic power generation and its dissipation along auroral field lines. J. Geophys. Res. 107 (2002). doi: 10.1029/2001JA900136 3. Balogh, A., Carr, C., Acu˜na, M. H., et al.: Overview of in-flight performance and initial results. Ann. Geophys. 19, 1207–1217 (2001) 4. Birn, J., Hesse, M.: Energy release and conversion by reconnection in the magnetotail. Ann. Geophys. 23, 3365–3373 (2005) 5. Dunlop, M., Balogh, A., Glassmeier, K.-H., Robert, P.: Four-point Cluster application of magnetic field analysis tools: The Curlometer. J. Geophys. Res. 107 (2002). doi:10.1029/2001JA005088 6. Gustafsson, G., Andr´e, M., Carozzi, T., et al.: First results of electric field and density observations by Cluster EFW based on initial months of operation. Ann. Geophys. 19, 1219–1240 (2001)

32

Cluster Observations of Energy Conversion Regions in the Plasma Sheet

459

7. Hamrin, M., Marghitu, O., R¨onnmark, K., et al.: Observations of concentrated generator regions in the nightside magnetosphere by Cluster/FAST conjunctions. Ann. Geophys. 24, 637–649 (2006) 8. Marghitu, O., Hamrin, M., Klecker, B., et al.: Experimental investigation of auroral generator regions with conjugate Cluster and FAST data. Ann. Geophys. 24, 619–635 (2006) 9. R`eme, H., Aoustin, C., Bosqued, J., et al.: First multispacecraft ion measurements in and near the Earth’s magnetosphere with the identical Cluster ion spectrometry (CIS) experiment. Ann. Geophys. 19, 1303–1354 (2001) 10. Wygant, J., Keiling, A., Cattell, C., et al.: Polar spacecraft based comparisons of intense electric fields and Poynting flux near and within the plasma sheet – tail lobe boundary to UVI images: An energy source for the aurora. J. Geophys. Res. 105, 18 675–18 692 (2000)

Chapter 33

Acceleration of > 40 keV Electrons in Near-Earth Magnetotail Reconnection Events ˚ A. Asnes, M.G.G.T. Taylor, and A.L. Borg

Abstract Reconnection of magnetic field lines have been invoked as an acceleration mechanism producing significant amounts of super-thermal electrons in the high ˚ energy range. However, in a recent paper by Asnes et al., 2008, energetic particle generation during geomagnetic active times was shown to be mainly caused by plasma sheet heating rather than reconnection driven acceleration. To examine this discrepancy we present observations from an ensemble of near-Earth reconnection events observed by the Cluster spacecraft near apogee in the years 2001–2004, and compare electron fluxes with values obtained during the surrounding time intervals and statistical results obtained in the same region in the plasma sheet. We find that observations in the proximity of the X-line only sometimes yield high fluxes of energetic electrons. The maximum flux level is always observed near the neutral sheet, and typically occurs when the distribution is near Maxwellian. It appears that although reconnection immediately heats the cold inflowing plasma, this acceleration is typically only sufficient to bring the electron fluxes up to a level approximate to the pre-existing plasma sheet levels.

33.1 Introduction Observations of solar flares indicate that as much as 10–50% of the energy released by magnetic reconnection goes to acceleration of electrons to supra-thermal energies [1], illustrating that collisionless reconnection should be considered as an important element of particle acceleration in plasmas. In the Earth’s magnetosphere reconnection takes place in many different regions, with very different boundary conditions in the distant tail, at the dayside magnetopause and the magnetosheath [2–4]. Here we focus on Cluster observations of near-Earth magnetotail reconnection, observed at ˚ A. Asnes () and M.G.G.T. Taylor ESA/ESTEC, Keplerlaan 1, 2200AG, Noordwijk, The Netherlands e-mail: [email protected] A.L. Borg Norwegian Defense research Institute, Kjeller, Norway

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 33, c Springer Science+Business Media B.V. 2010

461

462

˚ A. Asnes et al.

distances in the XGSM direction between 16 and 19RE and within a few hours of midnight local time. The onset of reconnection in this region is not well understood, but is believed to occur when the current sheet becomes sufficiently thin to allow plasma instabilities to grow and demagnetize the particles. Reconnection in the nearEarth tail must initiate within the closed field line region of the plasma sheet, but may continue onto the open field lines of the lobe. Although the geometry of the Xline region remains the same for closed and open field line reconnection, there are significant differences in the inflow (pre-reconnected) plasma, which may be decisive for the reconnection rate. Modeling (magneto-hydrodynamic (MHD), hybrid and particle in cell-PIC) of the reconnection problem indicates that reconnection rate is proportional to the Alfv´en velocity vA in the inflow region [5], which varies from 100s of km/s in the central plasma sheet up to several 1,000s km/s in the lobe region. It is likely that the direct reconnection acceleration of electrons depends strongly on the reconnection rate, which is equivalent to the “out of plane” reconnection electric field. This suggests that the most intense acceleration occurs when near-Earth reconnection has reached open field lines. The direct acceleration mechanism is a straightforward acceleration occurring as particle trajectories fall in the reconnection electric potential. The particle trajectories might be Speiser-like [6], but PIC simulations show a variety of possible trajectories where forces conspire to allow the particles to travel significant distances within the diffusion region [7]. In the case of a strong guiding field the electrons will stay magnetized, but may also undergo a significant amount of acceleration [8]. In addition to direct particle acceleration in the diffusion region, the larger scale plasma transport occurring as a result of near-Earth reconnection can accelerate the particles further, through the betatron and Fermi mechanisms [9]. The outflow regions Earthward and tailward are not symmetric, and the plasma pile up towards the dipolar field leads to more efficient betatron acceleration on the earthward side [10, 11]. Recent simulation results indicate that particles can also be accelerated within magnetic islands (or flux ropes) by the Fermi mechanism [12]. It is likely that coalescence of small-scale magnetic islands will also allow particles to be accelerated several times within reconnection diffusion regions.

33.2 Reconnection Events The 15 reconnection events used for the statistical results were observed by Cluster in the magnetotail between 19 < XGSM < 16RE at the times listed in Table 33.1. In all the events the Cluster spacecraft observe both tailward and earthward flows (usually in that sequence), which have been identified [13]. as reconnection outflow The table lists the peak differential electron fluxes s1 cm2 sr1 keV1 measured by a sample channel (E3: 70.3–96.7 keV) of the RAPID IES instrument [14]. In order to examine the effects of reconnection acceleration on the plasma sheet population the observations are separated into three intervals: (1) the period 1.5 h before the event; (2) during the event (inside the reconnection outflow) and (3) 1.5 h after

33

Acceleration of > 40 keV Electrons in Near-Earth

463

Table 33.1 Energetic electron flux in channel 70.3–96.7 keV for reconnection events from tail seasons 2001–2004 Peak flux Reconnection time Te, MK Ne cm3 Va km/s Before X-line After B nT interval (UT) 2001–08–17 16:23–16:44 2001–08–22 09:42–10:00 2001–09–10 07:48–08:04 2001–09–15 05:01–05:16 2001–10–01 09:37–09:47 2001–10–08 12:33–13:02 2001–10–11 03:25–03:42 2002–08–18 17:03–17:16 2002–08–21 07:54–08:25 2002–09–13 18:07–18:25 2002–10–02 21:20–21:23 2003–08–17 16:38–16:43 2003–08–24 18:30–19:10 2003–09–19 23:25–23:35 2004–09–14 23:01–23:07

70 230 15 15 850 200 50 175 3,300 500 600 40 200 450 500

6,000 180 75 110 600 50 80 200 200 200 700 200 4,500 450 450

NA 340 90 NA 850 1,000 150 250 500 200 500 150 2,500 2,000 600

63 28 21 31 35 32 21 21 40 25 20 46 25 24 30

1 2.3 0.9 1 2.8 5 3.5 6.5 1.3 2.0 <1 0.6 2 <6 4

0.02 0.03 0.1 0.02 0.06 0.07 0.03 0.06 0.03 0.05 0.01 0.08 0.03 0.04 0.04

9,700 3,860 1,450 4,780 3,120 2,640 2,640 1,870 5,520 2,440 4,360 3,550 3,150 2,620 3,270

the event. In the table are also listed the magnetic field strength, electron temperature and density of the reconnection inflow region, together with the resulting Alfv´en velocity (VA ). The multi-point observations provide us with coverage of the regions of interest here (central plasma sheet, inflow region and outflow region) at least for some time during all these events. The peak electron fluxes in this study are typically found at the spacecraft located closer to the neutral sheet (minimum BX ). We cannot guarantee that Cluster passes through the location of maximum flux during these events, but this is acceptable as we are mainly interested in spatially significant effects of reconnection in this study. In a majority of the events the maximum flux is observed during (five events highlighted yellow in the table) or some time after the reconnection event (seven events highlighted orange), which is consistent with acceleration/heating of the plasma sheet in relation with the reconnection (or substorm) event. However, the flux increase was only well above the typical plasma sheet level in a few of these events. In the remaining three events the maximum flux was observed prior to the event, indicating that there was no significant acceleration in these events. To compare the electron flux levels produced in the reconnection events with the general plasma sheet values, we plot the fluxes measured in energy channel E3 during the time intervals listed in Table 33.1 against BX =BLOBE in Fig. 33.1, superposed on the statistical distribution obtained by Cluster in the same region [15]. The black solid lines indicate median and quartile values of the statistical plasma sheet E3 fluxes in the midnight region (21–03 LT), including data for all levels of geomagnetic activity (Kp). The electron flux values in the reconnection events are distributed slightly above the median flux from the statistical values, but do not constitute the highest fluxes of the dataset.

˚ A. Asnes et al.

464 Central Plasma Sheet 70.30 < E (keV) < 96.70

log10 j (cm−2s−1sr −1keV−1)

4 3 2

102

1 0 −1 −1.0

# Observations per bin

Fig. 33.1 Reconnection event fluxes (70–97 keV) plotted against BX =Blobe (white dots) superposed on the statistical distribution of fluxes obtained by [14]

101 −0.5

0.0

0.5

1.0

Bx/BLOBE

33.3 Discussion and Summary Comparing the energetic electron flux during reconnection, with both general plasma sheet observations, as well as pre- and post-reconnection times, we find that reconnection does not always produce significant fluxes on a large spatial scale in the proximity of the X-line. Only a few out of 15 events produce unusually high levels of energetic electrons. In all our events at least one spacecraft obtain measurements from the reconnection inflow region, identified as relatively cold, tenuous plasma with absence of reconnection outflow. The inflow conditions in these events vary considerably in density and temperature, which might be expected. NearEarth reconnection may take place either on closed (plasma sheet) or open (lobe) field lines, and these regions also have varying plasma conditions. In several of the events, the inflow density was significantly higher than typical lobe densities (0:01 cm3 ), which results in a low Alfv´en velocity and less intense reconnection. The intensity of reconnection should be reflected in the acceleration of electrons. However, we find only a weak correlation between flux level and VA in the inflow region, based on the observations listed in the table. In this study we have compared observations of newly reconnected plasma with those of the pre-existing plasma sheet and have found a modest increase of energetic electron fluxes. However, it should be remembered that the plasma sheet is typically very hot compared with the inflowing plasma being accelerated, and may consist of plasma already heated by magnetic reconnection (at the distant neutral line or in preceding near Earth reconnection activity). In order to understand the acceleration efficiency of magnetic reconnection alone, it appears necessary to separate out the effects of direct acceleration from the subsequent adiabatic acceleration, which depends primarily on the large-scale magnetic field structure. This necessitates the

33

Acceleration of > 40 keV Electrons in Near-Earth

465

examination of detailed observations at the reconnection separatrix, combined with knowledge of the inflow plasma conditions. In this region one can observe the directly accelerated particles. Acknowledgements FGM and CIS CODIF/HIA data was obtained through the Cluster Active Archive. We thank the instrument teams and PIs P. Daly, A.N. Fazakerley, E. Lucek and H. R´eme for making the data available.

References 1. Lin, R. et al., RHESSI Observations of Particle Acceleration and Energy Release in an Intense Solar Gamma-Ray Line Flare Astrophys. J. Lett., 595, L69–L76, 2003 2. Øieroset, M.; Phan, T. D.; Fujimoto, M.; Lin, R. P. & Lepping, R. P., In situ detection of collisionless reconnection in the Earth’s magnetotail Nature, 412, 414–417, 2001 ¨ Papamastorakis, I.; Sckopke, N. & Haerendel, G. Plasma 3. Paschmann, G.; Sonnerup, B. U. O.; acceleration at the Earth’s magnetopause: evidence for reconnection Nature, 282, 243–246, 1979 4. Phan, T. D. et al., Evidence for magnetic reconnection initiated in the magnetosheath, Geophys. Res. Lett, 34, L014104, 2007 5. Birn, J. et al., Geospace Environmental Modeling (GEM) magnetic reconnection challenge, J. Geophys. Res., 106, 3715–3720, 2001 6. Speiser, T. W., Particle motion in the tail current sheet, Adv. Space Res., 11, (9)151 1991 7. Hoshino, M., Electron surfing acceleration in magnetic reconnection, J. Geophys. Res., 110, 10215, 2005 8. Pritchett, P. L., Relativistic electron production during guide field magnetic reconnection, J. Geophys. Res., 111, A10212, 2006 9. Birn, J.; Thomsen, M. & Hesse, M. Electron acceleration in the dynamic magnetotail: Test particle orbits in three-dimensional magnetohydrodynamic simulation fields Phys. Plasmas, 11, 1825–1833, 2004 10. Imada, S.; Hoshino, M. & Mukai, T. Average profiles of energetic and thermal electrons in the magnetotail reconnection regions Geophys. Res. Lett., 32, L09101, 2005 11. Imada, S.; Nakamura, R.; Daly, P. W.; Hoshino, M.; Baumjohann, W.; M¨uhlbachler, S.; Balogh, A. & R`eme, H., Energetic electron acceleration in the downstream reconnection outflow region J. Geophys. Res., 112, A03202, 2007 12. Drake, J. F.; Swisdak, M.; Che, H. & Shay, M. A., Electron acceleration from contracting magnetic islands during reconnection, Nature, 443, 553–556, 2006 13. Borg, A., A study of magnetic reconnection events observed by Cluster satellites in the Earth’s magnetotail, University of Oslo, Plasma and Space Physics Group, PhD Thesis, 2006 14. Wilken, B. et al., RAPID – The imaging energetic particle spectrometer on Cluster, Space Sci. Rev., 79, 399–473, 1997 ˚ 15. Asnes, A.; Friedel, R. W. H.; Lavraud, B.; Reeves, G. D.; Taylor, M. G. G. T. & Daly, P., Statistical properties of tail plasma sheet electrons above 40 keV, J. Geophys. Res., 113, A032023202, 2008

Chapter 34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere J.F. Fennell and J.L. Roeder

Abstract In this paper we present a review of what we know about the evolution and energization of energetic electrons in the inner magnetosphere. We emphasize what we have learned since the review by Friedel et al. [17] that discussed in detail the possible acceleration mechanisms for these electrons. The primary result has been a greater focus on local acceleration processes with significant new evidence that points to processes that are acting in the region between GPS and geosynchronous altitudes. In addition to reviewing the most recent results we also use as an example the magnetosphere’s energetic electron responses to a series of storms that occurred in the July 16–25, 2004 period. The storms had gradually increasing magnitude, in terms of their minimum DST of 80, 101, 148, and 197 nT. We followed the penetration and enhancements of the electrons as a function of L value using the HEO3 observations. We found that the two smallest storms did not cause relativistic electron enhancements for L <6:5 but did cause enhancements in the electron “source” populations at >130 and >230 keV down to L 3:0. The two largest events cause electron enhancements at all energies from >130 keV to >3 MeV down to L 2:5. We found that the Cluster IES observations of the source populations in the pre midnight plasma sheet were quite intense with strong field aligned distributions during the early main phase of the storm on July 24. The combined results indicate that these early main phase electron fluxes at Cluster may be directly linked to the rapid response of electrons at low L observed by HEO3 and could have been the source for the <1 MeV fluxes observed there.

34.1 Introduction We have known for some time now that the outer zone electron fluxes respond to changes in solar wind conditions ([2–4, 7, 43, 44, 62]). In particular, high-speed solar wind streams correlate with electron fluxes increases, as do magnetic storm

J.F. Fennell () and J.L. Roeder The Aerospace Corporation, P.O. Box 92957, Los Angeles, CA 90009, USA e-mail: [email protected]

H. Laakso et al. (eds.), The Cluster Active Archive, Astrophysics and Space Science Proceedings, DOI 10.1007/978-90-481-3499-1 34, c Springer Science+Business Media B.V. 2010

467

468

J.F. Fennell and J.L. Roeder

Fig. 34.1 Example of the MeV electron flux dependence on solar wind velocity (After Paulikas and Blake [44])

occurrences. Figure 34.1 shows an example of the relationship between solar wind velocity and >1:55 MeV electron fluxes taken at geosynchronous orbit [44]. However, we also know that electron flux enhancements do not occur with all solar wind speed increases nor with all magnetic storms. To generate energetic electron flux increases may require a substantial solar wind speed increase associated with a precursor solar wind density enhancement and combined with a southward turning of the IMF as was shown in Blake et al. [7]. Reeves et al. [45] showed that magnetic storms could produce cases of energetic electron flux increases, decreases, or no change in flux. They also showed that, during an 11-year period in which they examined 226 storms, 53% were followed by flux enhancements, 28% were associated with flux decreases and 19% showed no change in the electron fluxes or an adiabatic response. They indicated that the magnitude of the storms, as measured by the minimum DST did not matter nor did the L-value of the observations for L <7 however other evidence may be in disagreement with that conclusion (see [39, 60]). However, like Paulikas and Blake [44], Reeves et al. [45] did note that the solar wind speed mattered and that the flux increases were greater for larger speeds. Studies have also been done examining the source location of relativistic electrons [49], evolution of the energetic electron pitch angle distributions during post-storm flux enhancements [5, 18], the relationship between the L value of the peak electron flux with storm magnitude [60] and plasma pause position [38] and the main phase and post storm acceleration and losses of electrons (e.g. [10, 20, 22, 34–37, 39, 48– 51, 56] etc.)

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

469

34.2 What Has Been Learned 34.2.1 Parameters Used in Paper We note, that throughout this paper we will use L to identify the McIlwain [33] L parameter that is computed by tracing a particle’s trajectory between mirror points at the observation magnetic longitude in a model field. We will use L to indicate the Roederer [46] L value that is computed by tracing the closed drift trajectory for an electron in a model field and is related to the magnetic flux enclosed by the particle’s drift path. The reason for using both L and L is that not all storm-time studies used L and the two parameters are not equivalent. We use K to represent the Kaufman [28] K parameter which is a representation of the second adiabatic invariant. Finally, we use M to indicate a particle’s magnetic moment, the first adiabatic invariant. For more details on these parameters see the references noted or refer to Schulz and Lanzerotti [47].

34.2.2 Summary of Current Knowledge From the observations quickly summarized in the introduction above, we have learned much. In particular we have learned that the storm-time electron flux characteristics can be summarized as follows: (1) During the main phase of magnetic storms there is a dropout of the relativistic electron fluxes like shown in Fig. 34.2a, which is also often energy dependent ([5, 32, 42]) while at lower energies the fluxes may actually increase. The loss was not an adiabatic response [32]. Such losses are possibly due to microbursts, EMIC waves, strong chorus and hiss waves and outward diffusion to the magnetopause or a combination thereof ([21, 22, 31, 52, 56]; Shprits et al. 2006) while the increases at lower energies may result from convection or substorm injections. (2) The relativistic electrons start to increase at the beginning of the storm recovery phase, with increases at all pitch angles but beginning earlier at the near-90ı pitch angles as shown in Fig. 34.2b [9]. The pitch angle distributions are more peaked during the acceleration phase than at other times, with the steepest distributions at injection/acceleration onset during the flux rise. (3) The increase in electron intensity and the flattening of the pitch angle distributions are relatively rapid at 1 MeV with both features changing more slowly with increasing energy (see Fig. 34.2b). (4) The steepness of the pitch angle distributions remains energy-dependent even after the intensity of the relativistic electrons ceases to increase (again Fig. 34.2b).

470

J.F. Fennell and J.L. Roeder 105

104 L = 3.5 L = 3.7 L = 3.9 L = 4.1 L = 4.3 L = 4.5

1000

100 2.7 MeV

10 320

325

b

330 335 Day of 1997

340

Max intensity (3
106

0 −50

105

Dst

Electron Flux, /cm2 sec sr McV

a

−100

104

100

1.1 MeV

1000 10000 B (nT)

−150

Max intensity (3
105

0 −50 Dst

104

−100

103

100

1000 10000 B (nT)

−150

Max intensity (3
2.2 MeV

104

0 −50 Dst

103

−100

102 54

100

3.3 MeV 56

58

60

1000 10000 B (nT)

62 64 Day of 1997

66

−150 68

Fig. 34.2 Examples of MeV electron flux dropouts and post storm gains over a range of L values (panel a) and the evolution of the pitch angle distributions during the flux rise for three electron energies (panel b). The DST history is over plotted on the pitch angle plots for reference. The pitch angle dependence of the fluxes is color coded to represent the field intensity at the particle mirror points. The pitch angle distribution is relatively flat when the red and blue points are close together (i.e. have nearly same flux) and the distributions are peaked at 90ı when the red and blue points are widely separated (After Blake et al. [9])

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

471

(5) The electron fluxes increase most rapidly in the vicinity of the L value where the peak flux is finally established [9, 60]. (6) Bursty, strong precipitation is intimately associated with the injection and acceleration of relativistic electrons [39].

34.2.3 Radial Transport and M Violation Processes Friedel et al. [17] in their review of electron dynamics in the inner magnetosphere cited at least nine different mechanisms for generating relativistic electrons in the outer radiation belts. In all cases the models generally assume a seed population of 100 keV electrons provided by substorms and convection. A coarse summary of the processes is that they either conserve or violate magnetic moment (M) conservation, the first adiabatic invariant. Simple radial diffusion [14, 53] by ULF waves from high phase space density (PSD) regions into low PSD regions conserves M. Similarly, the enhanced radial diffusion discussed by Elkington et al. [12, 13] and Hudson et al. [24, 25] conserves M, however they used the known day-night asymmetry of the geomagnetic field to obtain faster transport rates than is predicted for azimuthally symmetric field models. The main issue is whether radial transport alone is sufficient to account for the population and maintenance of the electron radiation belts. Fujimoto and Nishida [18] tried to address this issue with a variation on radial diffusion that allowed pitch angle transport and recirculation of electrons to enhance the energy gained. That processes essentially required multiple cycles of radial transport inward near the equator to gain energy and outward well off the equator while conserving energy in the pitch angle transport process. In the trapping region, violation of M can occur by electron cyclotron resonance with different plasma waves or by field aligned curvature scattering. Doppler shifted cyclotron resonance of electrons with ULF waves can occur during their bounce motion as discussed by Summers and Ma [53]. Doppler shifted cyclotron resonance with ELF/VLF waves from chorus, EMIC, and AKR plus Landau resonance with magnetosonic waves, can violate M ([21–23, 35, 36, 57]; and others). Young et al. [63,64] discussed and modeled the violation of M by changes in field line curvature during storm times. In general, violation of M can lead to either acceleration or loss of electrons but the field line curvature violation is only a loss process.

34.2.4 Peak Electron Enhancements and Phase Space Density Dependences with L It has not been demonstrated, at this point, whether the M conserving and violating processes act alone or in concert to generate the resulting electron flux enhancements and the post storm PSD profiles that are observed. However, radial transport

472

J.F. Fennell and J.L. Roeder

Fig. 34.3 Cartoon showing examples of relativistic electron PSD radial profiles for different transport and acceleration assumptions. ts represents the pre storm profile and te the post-storm flux peak PSD profile. (a) Profile expected for pure inward radial transport (gray arrow) from large L source region of high PSD towards a smaller L region of lower PSD. (b) Profile from a combination of inward radial transport and internal acceleration with latter resulting in a peak in PSD. (c) Profile that starts with inward radial transport from a region of high PSD at large L (blue arrow) but ends with outward radial transport (orange arrow) into a region of lower PSD at large L because of a delayed but dramatic PSD decrease at large L during storm recovery phase Fig. 34.4 Energy gained by equatorially mirroring electrons starting near geosynchronous orbit as they are adiabatically transported to smaller L

and diffusion provides a good starting place from an observational perspective. It predicts a form for the electron PSD radial profiles that is easily observed by spacecraft that cover a significant range of L and L values. This classic PSD versus L

form is shown schematically in Fig. 34.3a. There is some long standing evidence that diffusive and convective radial transport does occur as evidenced by the many observations of a positive PSD gradient from small to large L ([49] and references therein). The convective transport is effective only for electrons with energies less than a few hundred keV. The energy gained from M conserving radial transport can be significant, as is shown in Fig. 34.4. The curves in this figure show how equatorially mirroring electrons that have the labeled energy at L 6:6 gain energy as they are transported to lower L values. This is an ideal depiction and does not take into account the pitch angle dependence of the energy gain, the time frame required for the transport, nor whether sufficient wave power is present at the electron drift frequencies to support the transport to small L values. Such issues are discussed in the paper by O’Brien et al. [39].

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

473

HEO3 5

SAMPEX

L of Maximum Flux

Lmax 1.3Lpp 4

3

40

50

60 70 80

100 | min Dst | (nT)

200

300

Fig. 34.5 Location of the peak MeV electron fluxes as a function of storm minimum DST . The solid black curve is the LMAX function from Tverskaya et al. [60] and the dashed green curve is a scaled plasmapause position model LPP from O’Brien and Moldwin [38]

As noted above, the electron flux rise is observed to occur most rapidly at the L position that ultimately corresponds to the peak in the enhanced fluxes for MeV electrons. Tverskaya et al. [60] showed that the L position of the post-storm MeV flux peak .LMAX / could be organized in terms of the storm-time DST minimum. O’Brien and Moldwin [38] showed that the plasmapause position, LPP , was well correlated with DST . The Tverskaya et al. [60] result was replotted by O’Brien et al. [39] with the estimated plasmapause position from DST used to fit the L dependence of the flux peak. Those results are reproduced here as Fig. 34.5. It shows that there is a strong correlation between the magnitude of a storm, as represented by DST , and the L value where the strongest enhancements in the MeV electron fluxes occur. O’Brien et al. [39] and later Li et al. [29] inferred that this relationship was the result of the erosion of the plasmapause with the deepest penetration and position of maximum acceleration of the electrons occurring just outside the estimated eroded plasmasphere boundary as indicated by the dashed LPP curve in Fig. 34.5. This is where it is expected that the VLF chorus is cutoff by the rising plasma density. So the picture that arises is one in which the source population is provided by convective and diffusive transport from large L and those source particles are further energized by M breaking in the VLF chorus wave fields outside the plasmasphere. The source population contains the free energy that generates the waves and the particles to be energized. This is an example of the organization of the electron enhancements by the strength and penetration of substorm and storm-time electric fields during the main and early recovery phases of a magnetic storm.

474

J.F. Fennell and J.L. Roeder

34.2.5 Source Populations The source populations consist of the suprathermal electrons of the plasma sheet. It is thought that these electrons are brought into the inner magnetosphere, to geosynchronous and lower altitudes, from the near-earth magnetotail by substorm injections and the storm-time convection that also brings the ring current ions into the inner magnetosphere ([5, 8, 27] and references therein). The Blake et al. [8] paper showed a linkage for the mid 10s to 100s keV electron fluxes between geosynchronous altitude and the Polar, Chandra and Cluster satellite apogee altitudes (all well outside of geosynchronous) on the night side during one substorm. Other authors have shown similar results (e.g. Baker et al. [5]). There are still some questions about the relative timing of the arrival of such source or seed populations and the observed onsets of reconnection in the magnetotail. An issue is whether the PSD in the plasma sheet is sufficient to provide the PSD gradient for classic diffusive transport to the inner magnetosphere. Li et al. [30] had already determined that the solar wind electron PSD was not sufficient to account for the electron PSD in the inner magnetosphere via a radial transport process alone. However, those electrons would be a seed population for other processes that could increase their energy as they transitioned from the solar wind to the plasma sheet before being further transported to trapping regions of the inner magnetosphere. Polar and Cluster observations were used to investigate whether the plasma sheet source, exterior to the trapping regions, was a sufficient source to supply electrons to the inner magnetosphere by radial transport. Taylor et al. [54] compared the PSD levels for electrons at the same M (0.6 MeV/G) at Cluster, Polar, geosynchronous and GPS positions. They concluded that there did appear to be sufficient electron PSD in the plasma sheet out to 18 Re to support the classic diffusive transport process but at the same time they noted that the PSD was enhanced or peaked in the inner regions and could not rule out the action of local acceleration mechanisms. They did note that the PSD levels measured at 18 Re are highly dependent on where in the plasma sheet they are taken. The PSD values change quite dramatically from the plasma sheet boundary to the central plasma sheet and through the neutral sheet regions. Taylor et al. [54] examined an interval of weak to moderate magnetic activity .DST 40 nT/, not representative of your typical magnetic storm. The question remains as to whether the plasma sheet PSD was high enough prior to large storms to supply the electrons for those events and whether the electrons measured at 18 Re in the magnetotail are on closed drift trajectories. This issue of closed drift trajectories will be discussed in greater detail below. There is some evidence by Shprits et al. [52] that the dramatic losses of electrons during the storm main phase could be partially caused by outward diffusion because of a dramatic drop in the PSD at high altitudes. They argue that the high pressure generated by the high-speed, high-density solar wind associated with magnetic storm onset brings the magnetopause boundary earthward. The low PSD exterior to the dayside boundary sets up the conditions for outward diffusion that lowers the PSD in the outer and inner regions of the magnetosphere for several hours. Other processes must then raise the PSD in the distant magnetosphere and magnetotail

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

475

to supply the source populations for the ensuing PSD enhancements in the inner magnetosphere, if radial transport is the dominant processes. The M breaking processes would lead to a different PSD radial profile than that expected for pure diffusive transport. A possible example of this is shown in Fig. 34.3b and is much like the peak in the PSD radial profiles that were observed by Green and Kivelson [19], Iles et al. [26], Chen et al. [11], and Fennell and Roeder [15]. The PSD peak observed in the Polar HIST data by Green and Kivelson occurred at L 5 at small equatorial pitch angles (large K values) for MeV electrons. The PSD peaks observed in the CRRES data by Iles et al. [26] occurred for near equatorially mirroring MeV electrons in the L 5–5:5 region. Chen et al. [11] used multiple satellite observations in a superposed epoch study covering 2 years of data and got a result that showed a peak in MeV electron (M 2:1 MeV/G) PSD profile at near equatorial pitch angles near L 5:5. Recently, Fennell and Roeder [15] observed PSD peaks in near equatorially mirroring electrons but at <MeV energies. They did observe a peak in the MeV electrons near L 5:5 but only for electrons mirroring well off the equator. Near the equator, they observed a flat or negative slope in the radial profile for 4:7 < L < 7:5 In fact, the PSD peak observed by Fennell and Roeder appears to be of the type shown in Fig. 34.3c for the off equatorial pitch angles. A summary PSD profile from Chen et al. [11] is shown in Fig. 34.6. In general, it shows a negative PSD gradient for L 6:5 for storm main phase and for the recovery and average conditions. This would seem to be inconsistent with the results of Taylor et al. [54] that indicated the plasma sheet PSD was consistent with that needed to support radial transport as a source of relativistic electrons. The PSD values at large L in Fig. 34.6 would not support radial transport as a major mechanism for generating and maintaining the radiation belts, at least at the one M value Chen et al. [11] studied. The differences between the limited observations indicate that further examination of this issue is needed.

Fig. 34.6 PSD radial profiles from Multisatellite observations during storm main and recovery phases plus comparisons to average PSD profiles (After Chen et al. [11])

476

J.F. Fennell and J.L. Roeder

Why would a peak of the type shown in Fig. 34.3c form? This could occur if the electrons observed at large L are not on closed drift shells or the PSD dropped dramatically near the last closed drift shells during the storm recovery period. The use of the adiabatic invariants to organize the electrons and to remove the drift-shell splitting [46] assumes that they move on closed drift trajectories. Those on open drift trajectories violate the third adiabatic invariant and may violate conservation of the second and possibly even the first invariant in extreme cases. If the drift trajectories are open, one should not assume there is a linkage between the PSD’s of electrons on those trajectories and those on closed drift trajectories. However, one does require that the electrons on open drift trajectories must be transported onto closed trajectories in order to provide a source for all processes that operate to reestablish and maintain the radiation belts. Fennell and Roeder [15] observed that the off equatorial mirroring electrons showed a strong gradient in PSD at higher L ’s .6/. In contrast, the simultaneously observed equatorially mirroring electrons had PSD’s that were nearly constant for L > 6:5. An example from their paper, presented in Fig. 34.7, shows this result. The model field [58,59] predicted that the electrons were on closed drift shells for both large and small K values but the observations indicated that they most likely were not on closed drift shells at K 0:25 otherwise the PSD profiles should not have exhibited nearly an order of magnitude drops within a narrow L region (delta-L < 0:5 Re.) at the higher L ’s. Fennell and Roeder [15] did not observe such steep PSD gradients for equatorially mirroring electrons out to L 7:5. However, it is not clear how far to larger L one could go along the magnetic equator and guarantee the electrons are on closed drift trajectories. Based on the Fennell and Roeder [15] observations, it is clear that magnetic field models may indicate electrons are on closed drift shells when observations of very steep electron PSD gradients indicate the drift shells are not closed.

Fig. 34.7 PSD radial profiles for small (panel a) and moderate K values (panel b). Note the difference in vertical scales for (a) and (b) (After Fennell and Roeder [15])

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

477

34.3 Energetic Electron Responses During July 2004 Storms In the sections summarizing what we have learned we did not provided an example of what a storm really looks like, beginning to end, from the perspective of satellites in the inner magnetosphere and plasma sheet. We do so here in an attempt to show the relationship between the near-earth magnetotail electron fluxes and the changes deep in the inner magnetosphere with emphasis on a moderately large storm. To do this, we use HEO3 and Cluster data from the period around 24 July 2004. For the HEO3 data we extend the observations in the inner magnetosphere to include events that preceded and followed the 24 July storm by a few days. For Cluster we focus on the early main phase portion of the July 24 storm.

34.3.1 Data Sources for Example Storms HEO3 is a non-spinning satellite in a 12 h high-altitude high-inclination orbit that crosses the same L and L values at two different MLT regions twice a day. The sensors on HEO3 measure the integral electron fluxes in six channels: E1 > 130 keV, E2 > 230 keV, E3 > 45 keV, E4 > 630 keV, E5 > 1:5 MeV and E6 > 3 MeV. E4 to E6 are omni directional sensors while E1-E2 and E3 are two telescopes. (Note: The HEO3 E1-E2 telescope temperature rose rapidly with altitude for the regions with L 5 during the period of the study, because of increased solar illumination as its attitude changed with altitude. The >130 and >230 keV electron channels became noisy as the temperatures rose. This raised their background levels and limits their usefulness at large L.) The HEO3 trajectory is close to the magnetic equator from L 1:75 to 3 but significantly far from it by L 6:5 (B/Bo 6:6 in the Olson-Pfitzer [40, 41] field model). For more details about HEO3 see Fennell et al. [16]. The Cluster data is from the RAPID IES sensor [61]. IES measures electrons with energies of 40–400 keV in eight channels. It obtains 3D angular distributions in two broad energy channels. We will show the angular data for the 42–53 keV electrons. The data used here were taken when the Cluster spacecraft were traversing the plasma sheet near their apogee in the pre midnight region on July 24, 2004.

34.3.2 HEO3 Observations of July 2004 Storms There was a sequence of magnetic storm events starting late on July 16 with a 80 nT event followed by events on July 22, 24, and 26 with DST minima of 101, 148 and 197 nT respectively, as shown in Fig. 34.8. Note that the DST minima were reached at 0300 UT on July 17, 0300 UT on July 23, 1200 UT on July 25, and 1400 UT on July 27. Examples of the electron responses in the inner magnetosphere are shown in Fig. 34.9 using HEO3 data. Figure 34.9 shows the electron rates

478

J.F. Fennell and J.L. Roeder

Fig. 34.8 DST , LPP , and KP for the period 14 July to 8 August in 2004. The red bar indicates the period of the Cluster data used. The HEO3 data spanned the full time interval of the plots. LPP is calculated from DST using the O’Brien and Moldwin [38] formulation

Fig. 34.9 Coded electron intensity plots. The left and right hand vertical axes have different ranges for the channels indicated. Dashed vertical lines identify the time of the DST minimum for each event. The L, L , and MLT are labeled at the top of each plot. The legends relate the colors to the electron energy channels with E1, E2, E3, E4, E5, and E6 having energies >0:13 (red), >0:23 (gray), >0:45 (blue), >0:63 (purple), >1:5 (green), and >3 MeV (black), respectively

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

479

measured at nine different L and L values for the interval from 14 July to 8 August 2004. (L and L were obtained from the HEO3 magnetic ephemeris data base which was generated using the Olson-Pfitzer field model [40,41].) Each panel is annotated with the L, L and MLT where the data were taken and the data from the individual energy channels are color-coded (see legends and caption for Fig. 34.9). We binned the HEO3 data into L-bins corresponding to L D 1:75, 2.0, 2.25, etc. up to L D 10 and show only a subset in Fig. 34.9 to show the trends in the data. The vertical dash lines in Fig. 34.9 correspond to the times of the DST minima for the storms in the 16–26 July 2004 interval. Figure 34.10 shows the solar wind parameters for only

Fig. 34.10 Interplanetary magnetic field (top four panels), solar wind density, speed and proton temperature (bottom three panels) on July 24, 2004. The arrow marks the interval where the intense energetic plasma sheet electron fluxes were observed

480

J.F. Fennell and J.L. Roeder

July 24. There was a sudden commencement-like positive excursion in DST prior to the main phase of the July 24 event as seen in Fig. 34.8. This was associated with a rotation in the interplanetary magnetic field and the solar wind speed jump near 1200 UT on July 24 as shown in Fig. 34.10. The solar wind speed was above 550 km/s the whole day and, after the field rotation, the interplanetary field intensity was relatively constant for the remainder of the day. We start by using the HEO3 data to examine the electron responses to magnetic activity preceding 24 July 2004 to see if there are residual effects in the inner magnetosphere. The data in Fig. 34.9 show that very deep in the inner magnetosphere at L D 2 .L D 1:9/ the electron fluxes did not start rising until about 0000 UT on 25 July and then only in the lowest energy channels (see panel a). The 450 and 630 keV fluxes started rising near 0000 UT on 27 July and the >1:5 MeV fluxes rose about a day later. At L D 2 there was not an electron response to the 16 and 22 July 2004 events and there wasn’t a response in the >3 MeV channel to any of the events. The electrons with energies that would correspond to a “source” population were observed to rise about half way through the main phase of the 24 July event at L D 2. The higher energies did not start rising until 48 h later during the secondary main phase that started late on 26 July. Thus while there is evidence that “source” population electrons reached L D 2 during the 24 July event there was not a corresponding increase in the relativistic electron population that deep in the inner magnetosphere. At L D 2:5 (L 2:41; see panel b) the electron responses to the 24 July event were significantly stronger than at L D 2. However, except for a gradual rise in the 130 and 230 keV electron fluxes over the 14–23 July interval, there was no clear higher energy electron response to the July 16 and 22 storms at L D 2:5. The 130 and 230 keV electron fluxes started a gradual rise near 0000 UT on July 24 that steepened near 1200 UT on 25 July. The 450 and 630 keV fluxes rose sharply near 1200 UT (i.e. near DST minimum) on 25 July at L D 2:5. The higher energy, >1:5 and >3 MeV, flux rises were more delayed, as is generally observed for all stormtime events, with the >3 MeV fluxes starting to rise after 0800 UT on July 27 during the main phase of the July 26 storm. We note that the time history of the estimated L position of the plasma pause .LPP / shown in Fig. 34.8 had values LPP 3:15 for the July 16 and July 22 events but had values of LPP D 2:9 and 2.6 during the July 24 and July 26 events, respectively. (The LPP values were obtained from DST using the O’Brien and Moldwin [38] formulation.) The fact that relativistic electron fluxes were not observed at small L in response to the July 16 and July 22 storms may be consistent, if the LPP values are accurate, with chorus waves not being present at these L values because they are inside the plasmapause boundary. Separately, we note that if we used the LMAX curve in Fig. 34.4 we would expect the peak in the relativistic electron flux to be at L 4:3 and 4:1 for the July 16 and July 22 events respectively. Observationally, the peak fluxes occurred at L 4 for the July 16 event at the four lowest energies with no response at the two highest energies. The peak fluxes were at L 3:75–4 at the three lowest energies and at L 4:25 for the three highest energies for the July 22 event. Similarly, the LMAX curve in Fig. 34.4 would indicate that the flux peaks for the July 24 and

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

481

July 26 events should be at L 3:7 and 3:5 respectively. The HEO3 data had the peak fluxes near L 3:5–3:75 for the two lowest energies. It was difficult to tell for the July 24 event since near their peak values these lower energy electron fluxes only had a short plateau before they continued to rise in response to the July 26 event. They had well defined temporal peaks at higher L values but those were not the maximum fluxes. For July 24 and July 26 events the higher energy channels .>450 keV/ had their peak fluxes near L 3:5 for both events. At this L value these higher energy fluxes were nearly constant for several days, following the July 26 event, after achieving their peak values (see Fig. 34.9 panels (c-e)). The largest electron responses were to the July 24 and July 26 events. These events are only 2 days apart. At the highest L values, the July 26 event main phase caused a flux drop so we are somewhat unsure of the peak flux from the July 24 event. We note that all the available L values, not just those in Fig. 34.9, were used in searching for the L of maximum flux for event. At L values from 3.0–6.5 the fluxes were observed to rise on 23 July and reached an initial peak or a plateau early on 24 July, and then further increased on 25 July or fell depending on the L value. The main response to the July 16 event was a flux drop for >450 keV electrons at all L’s and a rise in the 130 and 230 keV “source” electron fluxes only at L D 3:0–4:0 RE (Fig. 34.9, panels (c–e)). (The HEO3 low energy telescope temperature rose rapidly with altitude for the regions with L 5 because of increased solar illumination. The 130 and 230 keV electron channels started getting noisy and their background levels rose. The E1 channel was unusable while the E2 channel has a relatively high background for those L values. Thus, our measurements of the source electron fluxes are uncertain for L 4:5.)

34.3.3 Cluster Observations of July 24, 2004 Storm As noted above, the July 24 event caused a strong response in electrons deep in the inner magnetosphere. During the interval, 24–26 July the solar wind speed was enhanced and IMF Bz turned southward starting at 2100 on July 24 (see Fig. 34.10), reaching a level of 20nT early on July 25 and remained at that level until 1520 UT that day (not shown). Cluster transitioned from the near-earth tail-lobe, boundary layer and plasma sheet region prior to and during the main phase of this event, as evidenced by the IES electron fluxes in Fig. 34.11. The spin average energetic electron fluxes were very similar at all four Cluster spacecraft throughout July 24. The spacecraft were in the magnetotail on the morning side, near 3 MLT, during the onset of a 148 nT storm that started gradually near 1200 UT (ref. Fig. 34.8). At the time of the storm onset the Cluster spacecraft appeared to be in the tail lobe or near the high latitude boundary of the plasma sheet as evidenced by the lack of electron fluxes observed by the IES sensor on all four cluster satellites in Fig. 34.11 and the weak ion fluxes observed by CIS (not shown).

482

J.F. Fennell and J.L. Roeder

Fig. 34.11 Spectrograms of spin averaged electron fluxes measured by RAPID IES on all four Cluster spacecraft for 24 July 2004

As the main phase of the July 24 storm progressed a bit, the Los Alamos geosynchronous satellites near midnight observed a substorm injection around 1300 UT (not shown). The Los Alamos satellites also observed a second injection near 2320 UT but none in between. The Cluster spacecraft entered the plasma sheet about 3 h later, just before 1600 UT, while the DST 30 nT and Kp was falling from C6 to C5. The steepest decline in DST occurred after 2200 UT on July 24. Between 1700 and 2400 UT, the Cluster IES instruments observed intense fluxes of field aligned electron (FAE) distributions as shown in Fig. 34.12. These FAE distributions were not constant but occurred whenever Cluster entered the plasma sheet during this interval, as indicated by the FGM [6] Bx and BT measurements in Fig. 34.12a. These FAE fluxes, with pitch angles <30ı , mirror closer to the earth where the field intensity is more than twice that at Cluster. The electron distributions observed at Cluster can convect inwards in the storm time electric field to become the source population

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

483

Fig. 34.12 Cluster 3 magnetic field measurements (top panel), energy-time spectrogram of RAPID IES spin averaged fluxes on C3 (second panel) and RAPID IES angular distributions from all four Cluster spacecraft (bottom four panels) for the interval 1530–2400 UT on 24 July 2004

for the subsequent enhancements observed by the HEO3 spacecraft. The energetic electrons observed at cluster are not purely field aligned. The fluxes at 90ı pitch angle were at times comparable to or 50% of the FAE fluxes. In general, early in

484

J.F. Fennell and J.L. Roeder

the storm main phase, Cluster observed about an order of magnitude increase in the 40–400 keV plasma sheet electron fluxes from those prior to the main phase onset as shown in Fig. 34.11. These long duration energetic FAE are unique and not understood at this time. The existing models of magnetotail processes can support FAE from reconnection for a matter of minutes but not for very extended periods and certainly not for hours. The fact that they are present every time Cluster reentered the plasma sheet of the 6 h period shown in Fig. 34.12 requires that either they were present continuously in the plasma sheet for that long of a period or that there were at least seven or more reconnection events in that period. The period 2020–2300 UT in Fig. 34.12 shows that FAE were present essentially that whole time. This cannot be explained by the standard models. These 24 July FAE fluxes can be contrasted to those observed by ˚ Asnes et al. [1] and Taylor et al. [55], both of which were very short in duration and consistent with expectations for a reconnection event. Cluster left the plasma sheet and entered the tail lobe prior to the substorm onset observed by the Los Alamos satellites near 2320 UT on July 24. It did not reenter the plasma sheet until about 0030 UT on July 25. At that time the energetic electron fluxes observed by IES had returned to normal pre-storm levels (not shown). Enhanced levels of energetic electron fluxes were not observed again at Cluster until after 0250 UT on July 25 and they were still at levels lower than those observed prior to 2300 UT on July 24. It is not until 0630 UT on July 25 that Cluster observed energetic electron flux levels comparable to those observed in Fig. 34.10 and then for less than an hour. Cluster spent most of the rest of July 25 in the tail lobe with occasional intervals in the boundary layer and very short intervals in plasma sheet like plasma. During this period DST reached its minimum of 148 nT at 1000 UT and the recovery period started. As Fig. 34.9 shows, by mid day on 25 July the energetic electron enhancements deep in the inner magnetosphere (see Fig. 34.9a and b) were well under way. The source fluxes (130 and 230 keV fluxes) that were transported deep into the inner magnetosphere initially arrived at low L late on July 24 and early on July 25. We cannot tell exactly when from the HEO3 observations because of sparse sampling due to its 12-h orbital period. It is clear from Fig. 34.9 that the >450 keV electrons had reached low L values by the time of or just after DST minimum. The sequence is less clear for L 3 because of the preexisting enhanced >100 keV fluxes that remained from the earlier activity. However the >3 MeV fluxes are definitely rising by 1200 UT on July 25 at L D 3. So not only are the source type electrons present by the time of early recovery so are the accelerated or high energy fluxes. The fact that the intense 40–400 of keV electron fluxes were observed at Cluster during the early storm main phase, very late on July 24, combined with the clear rise of fluxes of electrons at low L with energies consistent with those observed in the plasma sheet (but raised in energy by transport as in Fig. 34.5) provides strong circumstantial evidence that the electrons observed by Cluster are very possibly the source for the subsequent enhancements of MeV fluxes deep in the inner magnetosphere.

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

485

34.4 Summary Examination of historical and current results that describe and attempt to understand what causes the enhancements of electron fluxes in the inner magnetosphere’s radiation belts has narrowed the current research focus onto basically two mechanisms. Radial transport from the near earth plasma sheet while conserving the particle’s magnetic moment has been considered the primary mechanism for a long time. More recently, in the last decade, mechanisms that break conservation of M have been strongly considered. Much of the recent observational evidence points to the necessity for M breaking mechanisms in localized regions, for example 5 L 6, if one is to explain PSD peaks there even if radial transport is also operating. However the discussion is not closed because there is conflicting evidence and the number of observational studies is still relatively small. The Cluster and HEO3 storm response example shown above hints that transport could explain the rapid appearance of particles to near 0.5 MeV energies at very low L values prior to or at DST minimum during some storms. This provides a reason to continue to look carefully at the processes and to try to find ways to test the different models. However, one major difficulty that we currently have to contend with is that it usually takes multiple observation platforms to obtain the necessary physical space and energy sampling to trace the PSD from the distant plasma sheet to low L values in the magnetosphere. This requires very careful intercalibration of disparate data sets. In addition, except for the CRRES observations [34–37], there are not simultaneous wave measurements that can be used to assess whether the M breaking mechanisms are operating or not in most of the recent studies. It is hoped that the NASA RBSP mission will help solve some of this difficulty because it covers the whole inner magnetosphere with electron instruments that have a wide range, covering plasmasheet to highly relativistic energies, has the wave and field measurements needed, and with its two platforms, will increase the sampling cadence for determining the L dependence of the PSD profiles. However the RBSP must still rely on a second set of measurements within the distant plasma sheet. Acknowledgements The authors would like to acknowledge their many colleagues at The Aerospace Corporation that have supported their efforts, provided references, and participated in discussions on the topic of energetic electron sources and process in the magnetosphere. This work was supported by grants GC189637NGA from Boston University, 64361 from University of Colorado, and by The Aerospace Corporation’s Mission Oriented Investigation and Experimentation program.

References ˚ 1. Asnes, A., M. G. G. T. Taylor, A. L. Borg, B. Lavraud, R. W. H. Friedel, C. P. Escoubet, H. Laakso, 1. Daly, and A. N. Fazakerley, Multispacecraft observation of electron beam in reconnection region, J. Geophys. Res., 113, A07S30, doi:10.1029/2007JA012770, 2008. 2. Baker, D. N., J. B. Blake, L. B. Callis, J. R. Cummings, D. Hovestadt, S. Kanekal, B. Klecker, R. A. Mewaldt, and R. D. Zwickl, Relativistic electron acceleration and decay time scales in the inner and outer radiation belts: SAMPEX, Geophys. Res. Lett., 21, 409, 1994.

486

J.F. Fennell and J.L. Roeder

3. Baker, D. N., J. B. Blake, R. W. Klebesadel, and P. R. Higbie, Highly Relativistic Electrons in the Earth’s Outer Magnetosphere, 1. Lifetimes and Temporal History 1979–1984, J. Geophys. Res., 91, 4265, 1986. 4. Baker, D. N., P. R. Higbie, E. W. Hones Jr., and R. D. Belian, High-Resolution Energetic Particle Measurements at 6.6 RE, 3. Low-Energy Electron Anisotropies and Short-Term Substorm Predictions, J. Geophys. Res., 83, 4863, 1987. 5. Baker, D. N., W. K. Peterson, S. Eriksson, X. Li, J. B. Blake, J. L. Burch, P. W. Daly, M. W. Dunlop, A. Korth, E. Donovan, R. Friedel, T. A. Fritz, H. U. Frey, S. B. Mende, J. Roeder, and H. J. Singer, Timing of magnetic reconnection initiation during a global magnetospheric substorm onset, Geophys. Res. Lett., 29, 2190, doi:10.1029/2002GL015539, 2002. 6. Balogh, A., M. W. Dunlop, and S. W. H. Cowley et al., The Cluster magnetic fields investigation, Space Sci. Rev. 79, 65, 1997. 7. Blake, J. B., D. N. Baker, N. Turner, K. W. Ogilvie, and R. P. Lepping, Correlation of changes in the outer-zone relativistic-electron population with upstream solar wind and magnetic field measurements, Geophys. Res. Lett., 24, 927, 1997. 8. Blake, J. B., R. Mueller-Mellin, J. A. Davies, X. Li, and D. N. Baker, Global observations of energetic electrons around the time of a substorm on 27 August 2001, J. Geophys. Res., 110, A06214, doi:10.1029/2004JA010971, 2005. 9. Blake, J. B., R. S. Selesnick, D. N. Baker, and S. Kanekal, Studies of relativistic electron injection events in 1997 and 1998, J. Geophys. Res., 106, 19157, 2001. 10. Bortnik, J. and R. M. Thorne, The dual role of ELF/VLF chorus waves in the acceleration and precipitation of radiation belt electrons, JSAPT, 69, 378, doi:10.1016/j.jastp.2006. 05.030, 2007. 11. Chen, Y., G. Reeves, and R. H. W. Friedel, The energization of relativistic electrons in the outer Van Allen radiation belt, Nature Physics, 3, 614, 2007. 12. Elkington, S. R., M. K. Hudson, A. A. Chan, Acceleration of relativistic electrons via driftresonant interaction with toroidal-mode Pc-5 ULF oscillation, Geophys. Res. Lett., 26, 3273, 1999. 13. Elkington, S. R., M. K. Hudson, and A. A. Chan, Resonant acceleration and diffusion of outer zone electrons in an asymmetric geomagnetic field, J. Geophys. Res., 108, (A3), 1116, doi:10.1029/ 2001JA009202, 2003. 14. Falthammer, C. G., Radial diffusion by violation of the third adiabatic invariant, in: Earths Particles and Fields. Reinhold, New York, p. 157, 1968. 15. Fennell, J. F. and J. L. Roeder, Storm time phase space density radial profiles of energetic electrons for small and large K values: SCATHA results, Journal of Atmospheric and Solar– Terrestrial Physics, 70, 1760, doi:10.1016/j.jastp.2008.03.014, 2008. 16. Fennell, J. F., J. B. Blake, R. Friedel, and S. Kanekal, The Energetic Electron Response to Magnetic Storms: HEO Satellite Observations, in Physics and Modeling of the Inner Magnetosphere, AGU monograph 155, p. 87, 2005. 17. Friedel, R. H. W., G. D. Reeves, and T. Obara, Relativistic electron dynamics in the inner magnetospherea—review, J. Atmospheric and Solar Terrestrial Physics, 64, 265, 2002. 18. Fujimoto, M. and A. Nishida, Energization and anisotropization of energetic electrons in the Earth’s radiation belt by the recirculation process, J. Geophys. Res., 95, 4265, 1990. 19. Green, J. C. and M. G. Kivelson, Relativistic electrons in the outer radiation belt: differentiating between acceleration mechanisms, J. Geophys. Res., 109, doi:10.1029/2003JA010153, 2004. 20. Green, J. C., T. G. Onsager, T. P. O’Brien, and D. N. Baker, Testing loss mechanisms capable of rapidly depleting relativistic electron flux in the Earth’s outer radiation belt, J. Geophys. Res., 109, A1224, 2004. 21. Horne, R. B., N. P. Meredith, R. M. Thorne, D. Heynderickx, R. H. A. Iles, and R. R. Anderson, Evolution of energetic electron pitch angle distributions during storm time electron acceleration to megaelectronvolt energies, J. Geophys. Res., 108, 1016, doi:10.1029/2001JA009165, 2003. 22. Horne, R., R. M. Thorne, S. A. Glauert, J. M. Albert, N. P. Meredith, and R. R. Anderson, Timescale for radiation belt electron acceleration by whistler mode chorus waves, J. Geophys. Res., 110, A03225, doi:10.1029/2004JA010811, 2005.

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

487

23. Horne, R. B., R. M. Thorne, S. A. Glauert, N. P. Meredith, D. Pokhotelov, and O. Santolik (2007), Electron acceleration in the Van Allen radiation belts by fast magnetosonic waves, Geophys. Res. Lett., 34, L17107, doi:10.1029/2007GL030267. 24. Hudson, M. K., S. R. Elkington, J. G. Lyon, M. J. Wiltberger, and M. Lessard, Radiation belt electron acceleration by ULF wave drift resonance: Simulation of 1997 and 1998 storms, in Space Weather, edited by P. Song, H. J. Singer, and G. L. Siscoe, vol. 125, p. 289, AGU, Washington, D. C., 2001. 25. Hudson, M.K., S. R. Elkington, J. G. Lyon, V. A. Marchenko, I. Roth, M. Temerin, J. B. Blake, Gussenhoven, and M. S, J. R Wygant, Simulations of radiation belt formation during storm sudden commencements, J. Geophys. Res., 102, 14087, 1997. 26. Iles, R. H. A., N. P. Meredith, A. N. Fazakerley, and R. B. Horne, Phase space density analysis of the outer radiation belt energetic electron dynamics, J. Geophys. Res., 111, A03204, doi:10.1029/2005JA011206, 2006. 27. Jordanova, V. K., New insights on geomagnetic storms from model simulations using multispacecraft data, Sp. Sci. Rev., 107, 157, 2003. 28. Kaufmann, R. L., Conservation of the first and second adiabatic invariants, J. Geophys. Res., 70, 2181, 1965. 29. Li, X., D. N. Baker, T. P. O’Brien, L. Xie, and Q. G. Zong, Correlation between the inner edge of outer radiation belt electrons and the innermost plasmapause location, Geophys. Res. Lett., 33, L14107, doi:10.1029/2006GL026294, 2006. 30. Li, X., I. Roth, M. Temerin, J. R. Wygant, M. K. Hudson, and J. B. Blake, Simulations of the prompt energization and transport of radiation belt particles during the March 24, 1991 SSC, Geophys. Res. Lett., 20, 2423, 1993. 31. Lorentzen, K. R., J. B. Blake, U. S. Inan, and J. Bortnik, Observations of relativistic electron microbursts in association with vlf chorus, J. Geophys. Res., 106, 6017, 2001. 32. McAdams, K. L. and G. D. Reeves, Non-adiabatic relativistic electron response. Geophys. Res. Lett., 28, 1879–1882, 2001. 33. McIlwain, C. E., Coordinates for Mapping the Distribution of Magnetically Trapped Particles, J. Geophys. Res., 66, 3681, 1961. 34. Meredith, N., R. B. Horne, S. A. Glauert, R. M Thorne, D. Summers, J. M. Albert, and R. R. Anderson, Energetic outer zone electron loss timescales during low geomagnetic activity, J. Geophys. Res., 111, A05212, doi:10.1029/2005JA011516, 2006. 35. Meredith, N. P., R. B. Horne, R. H. A. Iles, R. M. Thorne, D. Heynderickx, R. R. Anderson, Outer zone relativistic electron acceleration associated with substorm-enhanced whistler mode chorus, J. Geophys. Res., 107, doi:10.1029/2001JA900146, 2002b. 36. Meredith, N. P., R. B. Horne, D. Summers, R. M. Thorne, R. H. A. Iles, D. Heynderickx, R.R. Anderson, Evidence for acceleration of outer zone electrons to relativistic energies by whistler mode chorus. Ann. Geophys. 20, 967, 2002a. 37. Meredith, N., R. M. Thorne, R. B. Horne, D. Summers, B. J. Fraser, and R. R. Anderson, Statistical analysis of relativistic electron energies for cyclotron resonance with EMIC waves observed on CRRES, J. Geophys. Res., 108, 1250, doi:10.1029/2002JA009700, 2003. 38. O’Brien, T. P. and M. B. Moldwin, Empirical plasmapause models from geomagnetic indices. Geophys. Res. Lett., 30, 1152, 2003. 39. O’Brien, T. P., K. R. Lorentzen, I. R. Mann, N. P. Meredith, J. B. Blake, J. F. Fennell, M. D. Looper, D. K. Milling, and R. R. Anderson, Energization of relativistic electrons in the presence of ULF power and MeV microbursts: Evidence for dual ULF and VLF acceleration, J. Geophys. Res., 108, 1329, doi:10.1029/2002JA009784, 2003. 40. Olson, W. P. and K. A. Pfitzer, A Quantitative Model of the Magnetospheric Magnetic Field, J. Geophys. Res., 79, 3739, 1974. 41. Olson, W. P. and K. A. Pfitzer, Magnetospheric magnetic field modeling, Annual Scientific Report, AFOSR Contract No. F44620-75-C-0033, 1977. 42. Onsager, T. G., G. Rostoker, H.-J. Kim, G. D. Reeves, T. Obara, H. J. Singer, and C. Smithtro, Radiation belt electron flux dropouts: Local time, radial, and particle-energy dependence, J. Geophys. Res., 107, 1382, doi:10.1029/2001JA000187, 2002.

488

J.F. Fennell and J.L. Roeder

43. Paulikas, G. A. and J. B. Blake, Modulation of trapped energetic electrons at 6.6 Re by the direction of the interplanetary magnetic field, Geophys. Res. Lett., 3, 277, 1976. 44. Paulikas, G. A. and J. B. Blake, Effects of the solar wind on magnetospheric dynamics: Energetic electrons at the synchronous orbit, in Quantitative Modeling of Magnetospheric Processes, p180, WP Olsen, ed., Am. Geophys.Union, Washington, D. C., 1979. 45. Reeves, G. D., K. L. McAdams, R. H. W. Friedel, and T. P. O’Brien, Acceleration and loss of relativistic electrons during geomagnetic storms, Geophys. Res. Lett., 30(10), 1529, doi:10.1029/2002GL016513, 2003. 46. Roederer, J. G., Dynamics of geomagnetically trapped radiation, Springer Verlag, HeidelbergNew York, 1970. 47. Schulz, M. and L. J. Lanzerotti, Physics and Chemistry in Space, vol. 7, Particle Diffusion in the Radiation Belts, Springer-Verlag, New York, 1974. 48. Selesnick, R. S., Source and loss rates of radiation belt relativistic electrons during magnetic storms, J. Geophys. Res., 111, A04210, doi:10.1029/2005JA011473, 2006. 49. Selesnick, R. S. and J. B. Blake, On the source location of radiation belt relativistic electrons, J. Geophys. Res., 105, 2607, 2000. 50. Shprits, Y. Y. and R. M. Thorne, Time dependent radial diffusion modeling of relativistic electrons with realistic loss rates, Geophys. Res. Lett., 31, L08805, doi:10.1029/2004GL019591, 2004. 51. Shprits, Y. Y., R. M. Thorne, R. B. Horne, S. A. Glauert, M. Cartwright, C. T. Russell, D. N. Baker, and S. G. Kanekal, Acceleration mechanism responsible for the formation of the new radiation belt during the 2003 Halloween solar storm, Geophys. Res. Lett., 33, L05104, doi:10.1029/2005GL024256, 2006a. 52. Shprits, Y. Y., R. M. Thorne, R. Friedel, G. D. Reeves, J. Fennell, D. N. Baker, and S. G. Kanekal, Outward radial diffusion driven by losses at magnetopause. J. Geophys. Res., 111, A11214, doi:10.1029/2006JA011657, 2006b. 53. Summers Danny and Chun-yu Ma, Rapid acceleration of electrons in the magnetosphere by fastmode MHD waves, J. Geophys. Res., 105, 15, 887–15,895, 2000. 54. Taylor, M. G. G. T., R. H. W. Friedel, G. D. Reeves, M. W. Dunlop, T. A. Fritz, P. W. Daly, and A. Balogh, Multisatellite measurements of electron phase space density gradients in the Earths inner and outer magnetosphere, J. Geophys. Res., 109, A05220, doi:10.1029/2003JA010294, 2004. 55. Taylor, M. G. G. T., G. D. Reeves, R. H. W. Friedel, M. F. Thomsen, R. C. Elphic, J. A. Davies, M. W. Dunlop, H. Laakso, B. Lavraud, D. N. Baker, J. A. Slavin, C. H. Perry, C. P. Escoubet, A. Masson, H. J. Opgenoorth, C. Vallat, P. W. Daly, A. N. Fazakerley, and E. A. Lucek, Cluster encounter with an energetic electron beam during a substorm, J. Geophys. Res., 111, A11203, doi:10.1029/2006JA011666, 2006. 56. Thorne, R. M., T. P. O’Brien, Y. Y. Shprits, D. Summers, and R. B. Horne, Timescale for MeV electron microburst loss during geomagnetic storms, J. Geophys. Res., 110, A09202, doi:10.1029/2004JA010882, 2005. 57. Thorne, R. M., Y. Y. Shprits, N. P. Meredith, R. B. Horne, W. Li, and L. R. Lyons, Refilling of the slot region between the inner and outer electron radiation belts during geomagnetic storms, J. Geophys. Res., 112, A06203, doi:10.1029/2006JA012176, 2007. 58. Tsyganenko, N. A. and M. I. Sitnov, Modeling the dynamics of the inner magnetosphere during strong geomagnetic storms, J. Geophys. Res., 110, A03208, doi:10.1029/2004JA010798, 2005. 59. Tsyganenko, N. A., H. J. Singer, and J. C. Kasper, Storm-time distortion of the inner magnetosphere: How severe can it get?, J. Geophys. Res., 108(A5), 1029, doi:10.1029/2002JA009808, 2003. 60. Tverskaya, L. V., N. N. Pavlov, J. B. Blake, R. S. Selesnick, and J. F. Fennell, Predicting the L-position of the storm-injected relativistic electron belt, Adv. Space. Res., 31, 1039, 2003. 61. Wilken, B., W. I. Axford, I. Daglis, P. Daly, W. Gttler, W. H. IP, A. Korth, G. Kremser, S. Livi, V. M. Vasliunas, J. Woch, D. N. Baker, R. D. Belian, J. B. Blake, J. F. Fennell, L. R. Lyons, H. Borg, T. A. Fritz, F. Gliem, R. Rathje, M. Grande, D. Hall, K. Kecscuemety, S. MckennaLawlor, K. Mursula, P. Tanskanen, Z. Pu, I. Sandahl, E. T. Sarris, M. Scholer, M. Schulz,

34

Evolution and Energization of Energetic Electrons in the Inner Magnetosphere

489

F. Sorass, and S. Ullaland, RAPID: The Imaging Energetic Particle Spectrometer on Cluster, Space. Sci. Rev., 79, 399–473, 1997. 62. Williams, D. J., A 27-Day Periodicity in Outer Zone Trapped Electron Intensities, J. Geophys. Res., 71, 1815, 1966. 63. Young, S. L., R. E. Denton, B. J. Anderson, and M. K. Hudson, Empirical model for scattering caused by field line curvature in a realistic magnetosphere, J. Geophys. Res., 107, 1069, 10.1029/2000JA000294, 2002. 64. Young, S. L., R. E. Denton, B. J. Anderson, and M. K. Hudson, Magnetic field line curvature induced pitch angle diffusion in the inner magnetosphere, J. Geophys. Res., 113, A03210, doi:10.1029/2006JA012133, 2008.

Island Universes (Astrophysics and Space Science Proceedings)

Read more

Island Universes (Astrophysics and Space Science Proceedings)

Read more

The Sun and Space Weather (Astrophysics and Space Science Library)

Read more

Highlights of Spanish Astrophysics V (Astrophysics and Space Science Proceedings)

Read more

Galaxies in the Local Volume (Astrophysics and Space Science Proceedings)

Read more

Principles and Perspectives in Cosmochemistry (Astrophysics and Space Science Proceedings)

Read more

Protostellar Jets in Context (Astrophysics and Space Science Proceedings)

Read more

Multiwavelength Cosmology (Astrophysics and Space Science Library)

Read more

Sirius Matters (Astrophysics and Space Science Library)

Read more

Space Plasma Simulation

Read more

Hasegawa, space plasma physics

Read more

Basic space plasma physics

Read more

Space Plasma Simulation

Read more

BASIC SPACE PLASMA PHYSICS

Read more

Basic Space Plasma Physics

Read more

BASIC SPACE PLASMA PHYSICS

Read more

Advanced space plasma physics

Read more

Advanced Space Plasma Physics

Read more

High Energy Density Laboratory Astrophysics (Astrophysics and Space Science)

Read more

High Time Resolution Astrophysics (Astrophysics and Space Science Library)

Read more

Science with the VLT in the ELT Era (Astrophysics and Space Science Proceedings)

Read more

Physics of the Space Environment

Read more

Water in the Universe (Astrophysics and Space Science Library)

Read more

Solar, Stellar and Galactic Connections between Particle Physics and Astrophysics (Astrophysics and Space Science Proceedings)

Read more

The Sun and Space Weather (Astrophysics and Space Science Library) (Second Edition)

Read more

Handbook of Geophysics and the Space Environment

Read more

Space Science

Read more

Space Science

Read more

Space Weather, Environment and Societies

Read more

Physics of Space Plasma Activity

Read more

Recommend Documents

Island Universes (Astrophysics and Space Science Proceedings)

ISLAND UNIVERSES ASTROPHYSICS AND SPACE SCIENCE PROCEEDINGS ISLAND UNIVERSES STRUCTURE AND EVOLUTION OF DISK GALAXI...

Island Universes (Astrophysics and Space Science Proceedings)

ISLAND UNIVERSES ASTROPHYSICS AND SPACE SCIENCE PROCEEDINGS ISLAND UNIVERSES STRUCTURE AND EVOLUTION OF DISK GALAXI...

The Sun and Space Weather (Astrophysics and Space Science Library)

THE SUN AND SPACE WEATHER ASTROPHYSICS AND SPACE SCIENCE LIBRARY VOLUME 277 EDITORIAL BOARD Chairman W.B. BURTON...

Highlights of Spanish Astrophysics V (Astrophysics and Space Science Proceedings)

Astrophysics and Space Science Proceedings For further volumes: http://www.springer.com/series/7395\ Highlights of S...

Galaxies in the Local Volume (Astrophysics and Space Science Proceedings)

Galaxies in the Local Volume ASTROPHYSICS AND SPACE SCIENCE PROCEEDINGS ATCA HI mosaic of the giant spiral galaxy M8...

Principles and Perspectives in Cosmochemistry (Astrophysics and Space Science Proceedings)

Astrophysics and Space Science Proceedings For further volumes: http://www.springer.com/series/7395\ Principles and ...

Protostellar Jets in Context (Astrophysics and Space Science Proceedings)

Astrophysics and Space Science Proceedings Protostellar Jets in Context K. Tsinganos Editor University of Athens, Gr...

Multiwavelength Cosmology (Astrophysics and Space Science Library)

MULTIWAVELENGTH COSMOLOGY ASTROPHYSICS AND SPACE SCIENCE LIBRARY VOLUME 301 EDITORIAL BOARD Chairman W.B. BURTON, Na...

Sirius Matters (Astrophysics and Space Science Library)

Siriu s Matters Astrophysics and Space Science Library EDITORIAL BOARD Chairman W. B. BURTON, National Radio Astronom...

Space Plasma Simulation