1. Storm overview. Measured and predicted Dst. Special substorm.
2. Substorm phases according to IMAGE data.
3. Solar wind parameters and dayside activity . Start of magnetospheric compression at 21:30.
4. Full data set. The location of spacecrafts and IMAGE stations.
5. LANL data. Inflation. Expansion. Injection.
6. INTERBALL-Tail data. Spectrograms and parameters. Surge propagation along magnetopause and LLBL.
7. Geotail data. Recovery phase after surge propagation.
8. INTERBALL-Auroral data. Three regions corresponding to 3 phases of the substorm.
9. Scenario. Timing Graph.
10. Conclusions.
11. Additional study
12. Acknowledgements

Fig. 1 shows the storm overview. The storm is caused by high amplitude (up to -10 nT) prolonged southward IMF. Such conditions are produced by IMF compression in interplanetary interaction region between high density - low speed and low density - high speed streams in SW. Fig. 2 (IMAGE magnetometer data) shows the set of substorms during the storm main phase. The most pronounced substorm onsets are at 22:22 and 02:15.

Fig. 1. SW parameters and DST profile

Fig. 2. IMAGE data set.

Fig. 3. Predicted and measured Dst0 profile and corresponding SW parameters.

Fig.4. Dst profiles calculated with different initial values of Dst0.

Fig. 5. IMAGE network magnetograms.

Fig. 5. shows magnetograms from several IMAGE stations located at about midnight meridian for the period of interest. There is strong substorm at 21:55 - 23:00 and then the high magnetospheric activity with the set of small substorms. This substorm is the first occurred after IMF turned southward. Before there was prolonged northward IMF. Substorm phases: 21:55 - start of strong convection --- growth phase. Increasing of westward current at low latitudes. 22:08 - small onset (not important) 22:22 - start of rapid poleward expansion (main onset of substorm). The westward current increases (the current wedge), and shifts to high latitudes. 22:45 - start of recovery phase. 23:00 - end of the recovery, but on the background of the disturbed conditions.

• Fig.6 shows solar wind parameters obtained by WIND and shifted by 1300 sec (time of solar wind propagation from WIND to subsolar magnetopause).
• WIND is far from the Earth (X=112 Re Y=-66 Re) so we use CANOPUS data to check time lag and to get a proof that the disturbance of the LLBL and whole magnetosphere occurred due to IMF turning. (see Fig.7 )
• There is small peak of ram pressure at 21:35 UT. Probably it is stronger near the Earth (bad correlation WIND - Earth). But compression of magnetosphere happens even in case of southward turning of IMF only so this item is not so important.

Fig. 6. WIND: IMF Bz, SW velocity and ram pressure

Fig. 7. The CANOPUS data and corresponding IMF Bz profile

• Fig.7 shows the CANOPUS magnetometers profiles, the location of 3 Canopus stations (on the ionospheric precipitation map after Newell and Meng, 1994) and IMF Bz profile.
• There are 2 short southward turning of IMF at 20:30 and at 21:30. They coincide with positive spikes in magnetograms exactly at the same time. It means the time lag was chosen correctly. CONT is in the cusp projection and feels the begin of reconnection at subsolar point better than other stations. RANK, which is in LLBL, as well as CONT show the increasing of westward ionospheric current while IMF Bz is negative. The depression of magnetic field saturates when IMF Bz becomes close to zero. DAWS projected onto CPS shows the recovery of the CPS at 22:40 UT.
• The peak of increasing of the ground field means compression of magnetosphere due to southward IMF and (possibly) pressure pulse.
• At 21:30 UT the BIG disturbance (wave of compression) starts to propagate through LLBL from dayside point downtail. Due to increasing the FAC, and then cross-tail current, magnetosphere begins to compress.

Fig. 8. The location of satellites at 22:00 UT. The IMAGE network and AURORAL are shown at ionospheric projection.

The list of data used in this study: IMAGE magnetometers data LANL-80,84,95 geosynchronous satellites data (electron flux in energy range 50 - 75 keV). INTERBALL-TAIL plasma and magnetometer data INTERBALL-AURORAL plasma and magnetometer data GEOTAIL plasma and magnetometer data Fig.8 shows the location of spacecrafts for 22:00 UT in the X-Y (GSM) projection. All spacecrafts are near the plane Zgsm = 0. Magnetopause position was calculated according to Shue model (1997) with SW parameters - P = 4nPa, Bz = -10nT. Field lines correspond to Tsyganenko96 model for the moment 22:00 UT. Field lines drawn pass through the points with coordinates X=Xgsm_Interball-Tail, Z=Zgsm_Interball-Tail, Y varies from -22Re to 22Re. The location of INTERBALL-AURORAL and IMAGE network is shown at additional frame. INTERBALL-AURORAL is on the field lines, projected to the flank of magnetosphere at terminator. See Fig.17.

We investigate the data from spacecrafts in order corresponding to increasing the distance from the subsolar point of magnetopause to satellite. We take into account only significant events. For instance: substorm onset registered by IMAGE at 22:22.

Fig.9. shows electron fluxes in the energy range 50 - 75 keV obtained by LANL-80,84,95. The brown curve corresponds to LANL-80, which was located closer to the midnight than other LANL satellites.

Fig. 9. LANL-80,LANL-84 and LANL-95 electron flux data.

We distinguish three phases according to LANL-80 data: Start of inner magnetosphere inflation at 21:47. It means the increasing of cross-tail current at the inner boundary of PS, stretching of field lines tailward and increasing of the volume of inner magnetosphere. This leads to adiabatic cooling of electrons and decreasing of particle density. 21:47 is labeled as the begin of growth phase for this region. Begin of the expansion - 21:57 (minimum of electron flux and start of increasing). The cross-tail current and the volume of magnetosphere begin to decrease (dipolarization). Strong injection at 22:25. We associated this event with dipolarization and recovery phase of substorm. Different LANLs registered injection at different times. The explanation is shown at low panel of Fig.9. The distance between LANL-84 and LANL-80 for electrons drifting around the Earth is 20 times less, than distance LANL-95 - LANL-80. The corresponding time lags have the same ratio.

INTERBALL-Tail data is shown at Fig.10. Spectrograms represent 2 angular channels of CORALL (ion spectrometer) and ELECTRON, looking sunward (2 upper panels) and tailward. Plasma parameters (obtained from CORALL experiment) are shown at Fig.11.

Fig.10. INTERBALL-TAIL measurements

Fig.11. CORALL parameters

• INTERBALL-TAIL enters MS twice at 22:15 UT and at 22:23 UT. Then it is crossing LLBL. 22:30 - 22:42 - outer LLBL with mostly magnetosheath-like ions. 22:43 - 23:27 - inner LLBL with large contribution of PS ions.
• PS before LLBL crossing is cold and dense (thin PS) and after crossing becomes hot and low-dense. ( Fig.11. ).
• The mark points are following:
• First entering MS. Magnetosphere compression reaches INTERBALL : 22:15. Here LLBL should be relatively thin and INTERBALL-TAIL finds itself in the magnetosheath practically without LLBL crossing.
• PS ions appears in LLBL at 22:42. INTERBALL penetrates deeper into inner LLBL. The peak of disturbance passed ( see Fig.17. ) and continues its propagation further downtail. According Lundin et al., 1991 the increasing FAC from this region into ionosphere looks like expansion (poleward shift of the position of westward current).
• PS again (Recovery) at 23:28

Fig.12. Cartoon of INTERBALL-TAIL MP and LLBL crossing

Note the clear velocity shear in the inner LLBL. Due to small plasma velocity at the inner boundary of LLBL, the foothold at this boundary can move much slower than magnetopause disturbance. This explains the long duration of LLBL. Fig.12. shows the possible picture of INTERBALL-TAIL magnetopause and LLBL crossing. This picture assumes LLBL surge propagation downtail during strong changes of IMF Bz. LLBL in this case forms thick layer between two waveforms (magnetopause and inner LLBL boundary) moving with different velocities. This cavern in magnetosphere can be created due to the Kelvin-Hemholtz instability and flank reconnections.

Fig.13. GEOTAIL data

Fig. 13 shows GEOTAIL data (available from 22:30 UT). As well as INTERBALL-TAIL GEOTAIL observes firstly dense and cold PS and about 23:00 the sharp increasing of Bz (dipolarization). Simultaneously PS becomes hot and low dense. GEOTAIL sees nothing (no plasmoid, only thinning of CPS) before dipolarization which we identify as recovery phase. In our ideology it is the time of the end of surge propagation in this region.

• Fig. 14 and Fig. 15 show the INTERBALL-AURORAL ion and magnetic field data. Panels from top to bottom Fig. 14 show the following ion species: He+, He++, O+, H+. The Fig. 15 lowest panel presents the pitch-angle the of PROMICS instrument and two top panels show magnetic field (deviation from model field) is in MFL frame.

Fig.14. PROMICS-AURORAL data

Fig.15. IMAP-3 and PROMICS-AURORAL data

• The location of INTERBALL-AURORAL for 22:00 in X-Y and Y-Z GSM and corresponding (green) Tsyganenko96 field line are shown at Fig. 16 . This line passes not far from LLBL region at dawn terminator.
• Before 21:40 hot PS ions and up- and downgoing low energy ionospheric ions are observed. At 21:40 the intensity of ionospheric ions sharply increases. But there is no significant flux of magnetosheath-origin ions except small contribution of protons at 1keV energy. We interpretes this region as inner edge of LLBL. The satellite feels the southward IMF and increasing FAC. At this time By increases and Bz decreases (see Fig. 15 ) It looks like field lines stretching. Simulatneously the intensity of PS ions decreases. We are closer to magnetopause and probably the input of high energy particles decreases as disturbance propagates further downtail thinning PS.
• Between 22:00UT and 22:10 the situation changes. The energy of oxygen and He+ increases and He++ flux of high intensity shows that spacecraft is on the projection of outer part of LLBL. Accelerated ions of ionospheric origin and magnetosheath-origin particles are observed simultaneously. At 22:10 high energy ion population becomes highly structured. The low energy cutoff is about 200 ev. The maximum of surge has passed. (Expansion phase for this local point). There are accelerated ions injected from LLBL and strongly accelerated ionospheric ions (O+ up to 20 keV). Magnetic field Fig. 15 has no trend, but is highly disturbed. We see injections from thick and very disturbed LLBL.
• At 22:25 the wave along LLBL has passed and the local recovery should start. All plasma has been injected and nothing keeps the magnetosphere from collapse. By and Bz decreases. There are intense injections mostly of magnetosheath ions. (Accelerated O+ and H+ still exist). Probably these injections are connected with propagation of PC-5 pulsations.

Fig.16 Field line related AURORAL satellite

All events at AURORAL and LANL-80 occur simulatneously. Fig. 16 explains the reason. The story starts when disturbance induces the cross-tail current. At that time LANL and AURORAL feel the inflation of magnetic field. The "expansion" and "recovery" should occur simultaneously at both spacecrafts too.

According to Lundin et al., 1991, Lundin et al., 1992 the following scenario can be applied (see Fig.17. ) :
• The compression due to southward IMF (and probably pressure pulse) propagates tailward. Probably K-H instability and erosion of magnetopause due to flank reconnections are the reasons of such a large disturbance. There are two waves: along magnetopause and along inner edge of LLBL . They have different velocities. LLBL becomes thicker dramatically and at its inner boundary surge with large charge is created.
• Deeper penetration of LLBL into magnetosphere creates local increasing of cross-tail field in the tail and of FAC in the morning and evening sectors.
• Surge reaches inner (near-Earth) boundary of PS. Cross-tail current increases. LANL-80 and Auroral feel it simultaneously (Fig. 16 ). Inflation of magnetosphere for LANL. FAC increasing for INTERBALL- AURORAL.
• Surge reaches auroral band (about X = -10Re) -- increasing FAC, thinning PS, and in maximum, generates auroral events, probably current disruption. Ionospheric conductivity increased due to injections and cross-tail current diverses into ionosphere. The substorm current wedge is established.
• Surge passes the auroral band and moves downtail. The field-aligned current shifts to higher latitudes . It looks like expansion. Very strong FAC at AURORAL. Accelerated particles from LLBL and from ionosphere (O+ ions accelerated up to 20 keV). Decreasing of cross-tail current near LANL-80 location and resuming of the magnetosphere.
• When main part of disturbance leaves this tail point (and the corresponding latitude) the expansion begins. The main current moves poleward from this point.
• Surge passes INTERBALL-TAIL and it observes the LLBL and magnetopause. At this time other measuring points show the end of the expansion. Geotail is still in thin PS.
• Set up of background-state situation. There are no currents to keep magnetosphere. The recovery phase begins. INTERBALL-TAIL returns into PS. Geotail meets large Bz and hot and low-dense PS. Sharp dipolarization at IMAGE. Injections at LANL-80. Pulses like PC-5 and particle injections at AURORAL.

Fig.17. Cartoon of scenario.

Fig.18 Timing graph of "substorm phases"

1. The increasing of ring current intensity during storm is well enough described by direct influence of SW conditions. The single exception occurred at the time of one substorm with onset at 22:22. This first substorm gives a significant input into ring current.
2. All events of this substorm (especially behaviour of ions populations measured by INTERBALL-TAIL, INTERBALL-AURORAL and GEOTAIL) show that substorm was initiated by large disturbance moving along magnetopause.
3. There are many evidences that this substorm is LLBL-driven one. The sequence of events corresponds to propagation of disturbance blob along LLBL.
4. Different parts of disturbance blob move with different velocities. This leads to different speeds of propagation of substorm "phases" across the points of measurements.

1. Problems to clarify:
   • How could such a big cavern in magnetosphere be created? Study of the magnetopause - LLBL crossings of INTERBALL-TAIL probe.
   • What part of LLBL or magnetosphere are projected onto INTERBALL-AURORAL probe? How are INTERBALL-AURORAL measurements connected with the processes it the vicinity of flank magnetopause?
   • Understanding speed of tailward propagation of disturbance (100 km/s - 50 km/s)
   • Expected signatures at Geotail. Relation between "Recovery" at Geotail and new substorm onset at 22:56
   • Study of possible time delays for each event due to finite Alfen speed.
2. To Be Done
   • Equivalent current system for different phases of substorm.
   • Use all-sky camera to check our assumption about disturbance propagation.
   • To define the location of source of injection observed by geosynchronized satellites.
   • To define disturbance boundary velocities using MAGION-INTERBALL-TAIL measurements
3. To find and to study the similar cases.

We thank for providing data:

WIND:

K. Ogilvie - Solar Wind experiment, Key parameters

R. Lepping - Magnetic Field Investigations, Key parameters

GEOTAIL:

T. Mukai - LEP

S. Kokubun - Magnetic Field instrument

INTERBALL-TAIL:

J.-A. Sauvaud, N. Borodkova - ELECTRON

S. Romanov - MIF

M. Nozdrachev, S.Skalsky - FM-3

Yu. Yermolaev - CORALL

K.Kudela, V.Lutsenko - DOK-2

INTERBALL-AURORAL:

V.Styazhkin, V.Petrov, A.Bochev - IMAP-3

J.-A.Sauvaud, R. Kovrazhkin - ION

LANL:

D.Belian, G.Reeves - Energetic Particles

IMAGE magnetometer data used were collected as a German-Finnish-Norwegian-Polish project conducted by the Technical University of Braunschweig .

CANOPUS network was supported by the Canadian Space Agency.

We are grateful to Dr. V. Sergeev for help in interpretation of IMAGE data and for his outstanding remark: "I do not understand ..."

We thank Yu. Galperin, A. Petrukovich, L. Zelenyi, E.Antonova, S.Savin, R. Kovrazhkin,T.Bosinger, A. Jahnin, S. Perraut for discussions.

We would like to emphasize that the main conductor of idea of LLBL-driven substorms is S.P.Savin.

Burton, McPherron, Russel, An empirical relationship between interplanetary conditions and Dst
J. Geophys. Res., 80, 4204, 1975

Lundin, R., I. Sandahl, J. Woch, R. Elphinstone, The contribution of the boundary Layer EMF to Magnetospheric Substorms,
Magnetospheric Substorms, Geoph.Monograph 64, 355, AGU 1991

Lundin, R., I. Sandahl, J. Woch, M. Yamamauchi, R. Elphinstone, J.S. Murphree, Boundary Layer Driven Magnetospheric Substorms,
Substorms 1, ESA SP-335 193, 1992

Newell P.T. and C.-I. Meng, Ionospheric projections of magnetospheric regions under low and high solar wind pressure conditions,
J. Geophys. Res., 99, 273, 1994

Iijima and Poterma, Large-scale characteristics of FAC associated with substorms.,
J. Geophys. Res., 83, 599, 1978

Lui, A.T.Y., Extended Consideration of a Synthesis Model for Magnetospheric Substorms,
Magnetospheric Substorms, Geoph. Monograph 64, 43, AGU 1991