Pages 923-931

FIELDS AND FLOWS AT GEOTAIL DURING A MODERATE SUBSTORM 

R. L. McPherron1,2,6, R. Nakamura2, S. Kokubun2, Y. Kamide2, K. Shiokawa2, K. Yumoto2, T. Mukai3, Y. Saito3, K. Hayashi4, T. Nagai5, S. Ables6, D. N. Baker7, E. Friis-Christensen8, B. Fraser6, T. Hughes9, G. Reeves10,
H. Singer11 

1Inst. of Geophys. and Planet. Phys. Univ. of California, Los Angeles, Los Angeles, CA 90095-1567,   
E-mail: rmcpherron@igpp.ucla.edu
2Solar-Terrestrial Environment Laboratory, Nagoya University, Toyokawa, 446, Japan
3ISAS, Yoshinodai, Sagamihara, Kanagawa 229, Japan
4Department of Earth and Planetary Physics, University of Tokyo, Tokyo, Japan
5Department of Earth and Planetary Science, Tokyo Institute of Technology, Meguro, Tokyo 152, Japan
6Department of Physics, University of Newcastle, NSW, AU
7Laboratory for Atmospheric and Space Physics, Univ. of Colorado, Boulder, CO 80309-0392
8Danish Meteorological Institute, Copenhagen, Denmark, 45-39 15470
9Solar Terr. Phys. Sect., Herzberg Inst. of Astrophys., 100 Sussex Drive, Ottawa, Ontario, Canada,    K1A 0R6
10Los Alamos National Laboaratory, Los Alamos, NM 87545
11NOAA R/E/SE Space Environment Labatory, Boulder, CO 80303

ABSTRACT

The behavior of the tail field and plasma during a moderate substorm with onset at ~1120 UT on Dec. 13, 1994 is examined using data from the ISTP spacecraft GEOTAIL and WIND, synchronous orbit and the ground and other locations. Five substorms were observed on this day while GEOTAIL was located near the center of the tail at Xgsm = -46 Re. In each substorm the field and plasma variations were similar to those observed during substorms by spacecraft closer to the earth. A southward turning of the IMF caused leads to the accumulation of lobe flux, development of a more tail-like field, and eventually to an expansion phase and its consequences. The ~1120 UT onset immediately followed a northward turning which ended a lengthy interval in which the IMF was alternately northward and southward. Shortly after the onset a flux rope passed GEOTAIL with a delay consistent with its formation at 20-30 Re several minutes earlier than a Pi 2 burst began at midnight. Immediately afterSubsequent to the onset the lobe field decreased and the plasma sheet disappeared. During the substorm recovery phase the plasma sheet reappeared with plasma movingconvecting earthward. The plasma data show that the tailward flow is a combination of convecting and streaming plasma. All of the substorms exhibited multiple onsets. The main onset of each can be determined by a combination of negative bay onsets, Pi 2 bursts, synchronous field-aligned currents, dispersionless particle injection, and midlatitude positive bays. In one event the flux rope at 46 Re arrived before dispersionless injection at synchronous orbit suggesting that reconnection in the tail begins at or before major onsets. In fact most of the major onsets were preceded by pseudo breakups early in the growth phase, and weak tailward flows carrying a weak vertical field fluctuating about zero. These observations suggest that reconnection begins in the middle tail early in the growth phase and that it is substantially intensified or begins again at another location at the expansion onset.

INTRODUCTION

The study of substorms has a long and controversial history during which many substorm models have been formulated (McPherron, 1991; Rostoker, 1996). The oldest and most successful model is the near-earth neutral line model (Baker et al., 1996) which postulates that magnetic reconnection is the process responsible for both the growth and expansion phase of a substorm. Although it is generally accepted that the substorm growth phase is caused by dayside reconnection, there is considerable argument about the cause of the expansion phase. The near-earth neutral line model postulates that the onset of reconnection on closed field lines in the midnight plasma sheet beyond 15 Re is responsible for expansion onset and subsequent effects such as tailward flows and particle injection at synchronous orbit. Another explanation for expansion onset is provided by the current sheet disruption model (Lui, 1991; Lui, 1996) which argues that a process other than reconnection disrupts the cross-tail current sheet near the earth (~7 Re) and diverts it through the auroral ionosphere. In this model disruption expands in both the azimuthal and radial direction until magnetic perturbations from the current disruption eventually create an x-line tailward of the disruption region. At this later time magnetic reconnection begins and eventually cuts through the plasma sheet to form flux ropes or plasmoids that move down the tail. 

There is now considerable evidence that the auroral substorm onset projects to a region in the plasma sheet that is well inside the closest point (~13 Re) at which signatures of magnetic reconnection have ever been observed (c.f. Baker et al. 1996). However, as discussed by Birn and Hesse (1996) it is possible to explain this may be explained by indirect effects of the plasma flow from a more distant x-line. Proponents of the current disruption model argue that reconnection does not begin soon enough to account for the phenomena that occur near synchronous orbit at expansion onset. The fundamental question is therefore, when does magnetic reconnection begin on the night side relative to expansion phase onset? 

The purpose of this brief report is to determine as precisely as possible when magnetic reconnection begins on the night side relative to expansion phase onset. To do this we use data from a variety of spacecraft and ground observatories to show that unique signatures of magnetic reconnection are observed by the GEOTAIL spacecraft at Xgsm ~ -46 Re within minutes after a major onset. The data are consistent with an onset of reconnection at X ~ -25 Re several minutes before any effects are seen on the ground or at synchronous orbit. However, the results do not rule out the possibility of a multi-step process in which reconnection is initiated by some other process provided this process begins to create a flux rope within two minutes after particles appear at synchronous orbit.

BRIEF OVERVIEW OF THE SUBSTORMS OF DECEMBER 13, 1994

December 13, 1994 was a relatively quiet day magnetically. Ap was less than 30 throughout the day and Dst averaged about -10 nT. The solar wind was monitored by the WIND spacecraft which at the beginning of the day was in the magnetosheath near the dawn terminator. WIND passed through the bow shock about 0600 UT and thereafter was in the solar wind. The solar wind speed was relatively stable throughout the day at about 550 km/s and its density was about 3-4 particles per cc. The dynamic pressure was more variable fluctuating slowly about 2 nP by about 50%. Peaks in pressure occurred about 1100 and 1640 UT. The IMF was predominantly oriented along the Parker spiral toward the Sun, although in two intervals By became small or positive. There were at least five major intervals of southward IMF near 0200, 0500, 1000, 1600 and 2200 UT. Each interval drove a moderate substorm. Major onsets can be identified by a variety of indicators at about 0126, 0526, 1120, 1642, and 2330 UT. A pseudo AL index created from available magnetograms indicates that all but the last substorm had peaks in |AL| ~ 500 nT. The final substorm was weaker at about 200 nT. These substorms were well observed by a constellation of five synchronous spacecraft. For substorm onsets at ~1120 and 1640 synchronous spacecraft were within minutes of local midnight and observed dispersionless injection of both electrons and protons. For the 0126 and 0526 UT onsets the spacecraft were past midnight and recorded dispersive injection. The 2330 UT onset was not observed by any of the five spacecraft. 

The ISTP spacecraft GEOTAIL was initially located at Xgsm = -46 Re, almost at midnight, and close to the nominal neutral sheet. As the dipole rotated to minimum tilt (1040 UT) the neutral sheet moved upward above the spacecraft leaving the spacecraftit close to the edge of the nominal quiet plasma sheet. By the end of the day it had moved to Ygsm ~ -7 Re. Rotation of the dipole, flapping of the tail, and substorm dynamics moved the relative location of the spacecraft from the neutral sheet to the tail lobe many times during the day. Each of the substorms exhibited the classic near-earth magnetotail signatures of substorms reported from spacecraft closer to the earth (McPherron, 1973, 1991; Hones, 1973, 1977; Nishida and Nagayama, 1973). A southward turning of the IMF leads to a gradual increase in lobe field (inferred from the measured total pressure), and a decrease in the vertical component, Bz. A major expansion onset then causes the lobe field to decrease. Following two of the onsets a flux rope passed the spacecraft causing an apparent increase in lobe field prior to the prolonged decrease. After every onset strong tailward flows and southward magnetic field were observed in the plasma sheet. In several cases these flows were observed at the neutral sheet and were convective. At other times the flows were field-aligned. Neither the tailward flows nor the southward fields were observed for prolonged intervals because in every case the plasma sheet disappeared from the spacecraft and the flow and fields of the plasma sheet were not observedcould not be measured. The plasma sheet normally reappeared in the recovery phase of each substorms and was dominated by strong convective earthward flows. 

Figure 1.: Magnetic activity near the 210° meridian. Top panel shows the Z, H and D components of the Macquarie Island fluxgate magnetometer. Middle panel presents Macquarie Island induction coil magnetograms. Bottom panel is the polarized Pi 2 power at three midlatitude stations normalized by the peak power at each station. Vertical lines are the start times of major bursts of pulsation activity

The existence of multiple onsets in each substorm complicates the interpretation considerably. Pseudo break-ups during substorm growth phase appear to initiate weak tailward flows. In addition it appears that reconnection can also occur at a distant x-line during the growth phase producing earthward flows at the center of the plasma sheet, or counter-streaming ion beams in its boundary layer. In following sections we justify this overview of mid-tail substorm observations by a detailed presentation of the ~1120 substorm on this day.

210° MERIDIAN OBSERVATIONS OF ~1120 UT SUBSTORM ONSET

The substorm reported here was well monitored by stations of the 210° meridian array. Observations near the southern auroral zone were made at Macquarie Island by a fluxgate magnetometer operated by the University of Nagoya, and by induction coils operated by the University of Newcastle. The top two panels of Figure 1 contain the data from these instruments. Weak negative bay activity began in the Macquarie fluxgate after 1000 and was suddenly enhanced at ~1120. Subsequent intensifications can be detected at 1145 and after 1210. The induction magnetograms indicate a very sudden change in activity at ~1120 suggesting that this was the major expansion onset of the substorm. Higher resolution plots show that the first deflection in the induction coils occurred at 11:19:15-30.

 Figure 2.: Electron fluxes at the synchronous spacecraft 1989-046 in the five lowest electron energy channels labeled at left. Dotted vertical line at 1100 UT is local midnight. Dashed vertical lines are onset times for midlatitude Pi 2 bursts.

Midlatitude Pi 2 pulsations on the 210° meridian were recorded by many stations including Ewa Beach, Hawaii, Moshiri, Japan, and Learmonth, Australia. The horizontal components measured by fluxgate magnetometers at these three stations have been analyzed to determine the low latitude polarization properties of the pulsation bursts. The bottom panel of Figure 1 presents a superposition of the polarized power recorded by each station. The polarized power was obtained from a running coherency analysis done in the following manner. Data from each station were multiplied by a Hamming function and then Fourier transformed. All negative coefficients as well as positive coefficients outside the Pi 2 frequency band were set to zero. The signal was inverse transformed to the time domain giving a complex (analytic) time series. A moving window of length two minutes was advanced by 15 seconds through the time series. For each window the coherency matrix was calculated. The 2 by 2 submatrix corresponding to the horizontal plane was analyzed to obtain the total power, polarized power, ellipticity, and azimuth of the major axis of polarization. A universal time corresponding to the center of the window was assigned to each determination. The maximum polarized power at each station was used to normalize the traces for ease of comparison in the final plot. Vertical dashed line have been drawn at times determined from the original filtered data traces so as to maximize time resolution in the onset of pulsation bursts. For most intensifications these times can be determined to better than one half cycle (< 30 s). 

In addition to the major intensifications seen at all stations it is possible to pick out subordinate intensifications at individual stations, although these have not been marked in the figure. The major intesifications occurred at 10:41:30, 11:20:14, 11:36:06, 11:44:49,; 11:53:13, and 12:08:35. The second intensification was the largest, and as we will show next, can be associated with dispersionless injection at midnight in synchronous orbit. As will be shown later in Figure 4, the onset at Macquarie island just onset equatorward of the auroral zone may have been a full minute earlier than the midlatitude onsets. Note that the two-minute window used to obtain polarized power disguises this fact.

SYNCHRONOUS PARTICLES FLUXES DUDING ~11201 UT SUBSTORM

The occurrence of aA major expansion onset at about ~1120 UT is confirmed by the behavior of synchronous particle fluxes. Figure 2 shows that dispersionless injection began at a spacecraft located just past midnight shortly after the Pi 2 burst began at midlatitudes. The injection is preceded by a rapid drop and recovery of the electron fluxes. Proton channels (not plotted) also show dispersionless injection at this time verifying that the spacecraft was located inside the injection region. Effects of later intesifications identified in the Pi 2 pulsations can not be identified in the particle flux variations.

GEOTAIL OBSERVATIONS OF ~11201 UT SUBSTORM ONSET

GEOTAIL, WIND and ground observations of magnetic field and GEOTAIL observations of plasma during the ~11201 UT substorm are summarized in Figure 3. The top panel presents IMF Bz as measured by WIND on the dawn side of the magnetosphere just upstream of the bow shock. The IMF was strongly southward for two intervals, first after 0900, and later after 1000. These intervals of southward IMF apparently did not cause significant substorm activity as no large variations can be found in magnetograms (H or X in bottom panel). The absence of Pi 2 activity during these intervals, and the absence of any particle injection at synchronous orbit both support this interpretation. However, following the first southward turning the lobe field (inferred from the total pressure measured at the spacecraft and plotted in the top trace in panel 2) appears to have increased several nanoTesla. The IMF turned slightly southward again after 1100 and the inferred lobe field increased further to about 18 nT. A fourth, weak southward turning occurred at 1200. Prior to this, at ~1120, Bz turned northward as simultaneously By (not plotted) turned from negative to positive, and a major substorm expansion followed. Strong negative bays developed in the auroral zone, a sharp Pi 2 onset was observed throughout the Pacific sector (Figure 1), and dispersionless injection of both electrons and protons was recorded by a synchronous spacecraft at midnight (Figure 2). At GEOTAIL, the inferred lobe field increased rapidly to nearly 25 nT and then began a prolonged decrease that ended nearly 50 minutes later at about 10 nT. The cause of the rapid increase in Bl is apparent in the magnetic field at GEOTAIL. As shown in the bottom trace of panel 2 there occurredis a bipolar Bz signature with a strong By component (not plotted) characteristic of a flux rope. This flux rope was carried past the spacecraft in a high speed (500-1000 km/s) tailward flow evident in panel 4. Initially the flow was entirely convective (trace overlaying shading), but following the passage of the flux rope it became parallel to the magnetic field. Streaming plasma was observed for only a few minutes before the plasma sheet pressure decreased suddenly (upper trace in panel 3) and GEOTAIL entered the tail lobe. GEOTAIL remained in the lobe for more than 35 minutes where it continued to observe cold plasma streaming tailward at about 100 km/s and observing the lobe field decreasing. At ~1210 the plasma sheet recovered and plasma was observed convecting earthward at several hundred km/s. 

Figure 3.: Observations of the ~1120 UT substorm onset on December 13, 1994 by WIND, GEOTAIL and ground magnetometers. Vertical lines show Pi 2 onset times. Horizontal dashed line in panel 2 shows the level of the inferred lobe field at the time of the first southward turning of the IMF (0900 UT). From top down the panels display WIND magnetic field (Bz), GEOTAIL magnetic field (Bl, Bt and Bz), plasma pressure, Xgsm component of the ion flow velocity, and X or H components of auroral zone magnetograms.

The flux rope transported tailward past GEOTAIL is proof that magnetic reconnection occurred earthward of the spacecraft (Xgsm > -45 Re) sometime before 1120 (the onset of the bipolar Bz signature). Since all normal indicators of substorm onset follow this time it is reasonable to assume that near-earth reconnection began at or before this expansion onset. In fact it seems likely that near-earth reconnection began even earlier as weak tailward streaming (panel 4), and weak southward magnetic field (panel 2), began shortly after 1100, about the time of the weak southward turning of the IMF. This interpretation of tailward flow is not unique, however, as the spacecraft was inside the plasma sheet close to its lower edge (strong Bt in middle trace of panel 2). At this location it is possible that the tailward streaming was caused by reflected ion beams originating in a reconnection site tailward of GEOTAIL. However, the distribution functions for this interval provide no evidence of counter streaming beams lending additional support to the speculation that reconnection began earthward of GEOTAIL sometime after the 1041 first Pi 2 burst. 

There is good reason to believe that a distant reconnection site was active during the very early growth phase of this substorm. About 1000, shortly after the first northward turning of the IMF, strong (1000-1500 km/s) earthward flows were recorded by GEOTAIL. At first these flows were primarily field-aligned, but throughout their duration they had a significant convective component. After 1030 when Bt was very small and it is likely the spacecraft was near the neutral sheet, the flow was entirely convective. It is interesting to note that the inferred lobe field stopped increasing when the earthward flow began, and that it remained constant throughout this flow even though the IMF again turned southward. The absence of an increase in Bl could be explained by balanced dayside and nightside reconnection. 

A sudden change in the tail flow occurred at 1040 UT when a Pi 2 burst was recorded throughout the Pacific sector. The earthward flow ceased and ten minutes later weak tailward flow began. Ten minutes later still the lobe field again began to increase. Despite flapping of the plasma sheet no earthward flow was observed in stark contrast to the earlier intervals when earthward flow was seen throughout the plasma sheet. Consequently it seems unlikely that the tailward flow in this interval is simply an ion beam streaming earthward in the plasma sheet boundary layer and reflected to the spacecraft as tailward streaming. An alternative interpretation is that reconnection stopped at the distant site and began earthward of the spacecraft. However, the subsequent increase in lobe field and the slow flow suggest that the reconnection rate was slow and therefore was probably occurring on closed field lines. 

Magnetic reconnection probably began earlier than the major expansion onset at ~1120 UT since strong tailward flow, a flux rope, and synchronous particle injection were observed subsequently. The disappearance of the plasma sheet and the prolonged decrease in inferred lobe field suggest that open field line reconnection began shortly later. Although reconnection of closed field lines must have begun prior to the beginning of tailward flow at GEOTAIL (<1123:30), the time at which reconnection of open field lines began can not be constrained unless one makes certain assumptions. If we assume that tailward motion of the trailing edge of a flux rope requires at least a portion of the flux rope be wrapped by open field lines, then open reconnection began earlier than 1131. If the beginning of the decrease in lobe field is its signature then it started before 1128. If the onset of fast tailward flow is diagnostic it began before 1123:30. Finally, if dispersionless injection is also its signature then it began before 1121:24. However, these earlier phenomena could also be explained by reconnection of closed field lines provided the rate is fast enough to generate the observed injection and tailward flow. In summary, open field reconnection almost certainly began sometime within ten minutes after the major onset. However, it could have begun even before the major onset and still be consistent with the observations presented here. In the following section we develop a simple model which uses the observed arrival times to determine the time and place at which reconnection might have begun.

TIMING OF THE ~1120 UT SUBSTORM ONSET AND ITS MAGNETOSPHERIC EFFECTS

The fundamental question thatwhich we want to answer is, what physical process causes the onset of the substorm expansion phase? The near-earth neutral line model postulates it is the onset of reconnection in the plasma sheet. The current disruption model argues thatsays it is an instability between electrons and ions drifting in a very thin current sheet, and that reconnection is initiated later as a consequence ofby effects of the current disruption. If this is the case there should be an obvious delay in the arrival of the effects of reconnection at a spacecraft like GEOTAIL at some distance down the tail. From data presented above it is obvious that this delay can not be long. To emphasize this point we have plotted in high resolution data from various locations in Figure 4. The earliest indication of the onset is given by induction coils at the subauroral station on Macquarie Island. The first panel of Figure 4 displays the outputs of north and east coils recorded at 1.0 s resolution. Deflections begin in both components at 1119:15. The second panel presents the midlatitude response to the high latitude activity in the form of a Pi 2 burst recorded by a fluxgate magnetometer in Hawaii. The data have been band pass filtered by a zero phase filter to show only the Pi 2 band [7-20 mHz]. The first deflection appears to be delayed by as much as half a cycle from the subauroral onset, but this may be a problem of the limited resolution and noise level of the original data. The next effect is seen in panel 3 at 11:20:56 at midnight in synchronous orbit when the electron fluxes begin a sudden drop. Dispersionless injection begins less than a minute later at 11:21:32 as evident in both the electron channel (upper trace) and the proton channel (lower trace). The next obvious effect is at GEOTAIL at 11:23:26 where the ion flow (panel 5) begins to be tailward and the vertical component of magnetic field (panel 4) begins to increase rapidly with the approach of thea flux rope. The center of the flux rope passed GEOTAIL at 11:25:31. From this time until 11:30:39 the magnetic field wasis continuously southward while the flow velocity gradually grew from 500 to over 1000 km/s. At this later time the magnetic field became comes positive and the flow changeds from convective to field-aligned (note cross-over of the X component of parallel and perpendicular components of ion flow). At 1135 the plasma sheet disappeared from the spacecraft. 

In data not presented here it can be seen that during the flux rope passage Bx changed sign four times, twice at the first minima in Bt, and twice again at the two successive minima. The plasma sheet By component changed significantly at the trailing edge of the flux rope. Prior to the flux ropeits arrival, plasma sheet By had been negativesouthward as was By in the IMF. After the flux rope arrival By it was nearly zero. An examination of the IMF at this time reveals that the IMF By also changed from negative to positive, and thereafter fluctuated about zero. We consider this sudden change in orientation of By in the plasma sheet to the same direction ascorrespond to By in the IMF as an indication that field lines open to the solar wind were being convected past the spacecraft.

DISCUSSION

Where and when reconnection begins in the tail can not be precisely determined from the available data. However, if we postulate a simple model and make several reasonable assumptions we can constrain the possibilities. Thus we assume that magnetic reconnection begins in the tail at the center of the plasma sheet at an unknown distance X, at an unknown time Tx. The onset creates earthward and tailward flows that carry the effect of reconnection away from the location X. The velocity shears at the edges of the earthward jet are propagated to the ionosphere by Alfven waves that set up the field-aligned currents of the substorm current wedge and initiate the high latitude Pi 2 (Baumjohann and Glassmeier, 1984). Subsequently the flow and its imbedded magnetic flux collide with the rigid inner magnetosphere and begin to pile up. Compressional waves generated by this pile up stimulate a plasmaspheric cavity resonance that is responsible for Pi 2 pulsations at midlatitudes (Allan et al., 1996). As magnetic flux accumulates there is a strong induced electric field. Electrons and protons drifting through this region are energized and appear above the thresholds of a synchronous particle detector located inside the region (dispersionless injection). 

As the above phenomena happen close to the earth, a flux rope is forming between the location X and the distant x-line. Plasma flows tailward from the near-earth x-line at X compressing northward magnetic field tailward of the o-line. After some delay the o-line passes a spacecraft in the center of the tail and the plasma sheet magnetic field turns southward. Subsequently reconnection continues to sever closed field lines until it breaches the plasma sheet boundary at some azimuthal location. This portion of the flux rope is no longer constrained close to the earth by closed field lines. In fact, it is now wrapped by open field lines that help eject the flux rope from the tail. With time, reconnection extends the azimuthal sector in which reconnection of open field lines is occurring until it reaches the magnetosheath on either side of the tail. At these intermediate times the flux rope exhibits a very complex structure of intertwined field lines of different topologies. For our purposes we assume that the trailing edge of the flux rope defines the first fully open field lines transported tailwards. In our model we take the first arrival of tailward flowing plasma as an indication that closed field line reconnection has begun earlier, earthward of the spacecraft.

FFigure 4.: An expanded time scale plot of important substorm variables. From the top down they include Macquarie Island induction coils (panel 1); filtered fluxgate magnetometer data at Hawaii (panel 2); electron and proton flux variations at a midnight synchronous spacecraft 1989-046 (panel 3); inferred lobe field magnitude Bl, observed Bt and Bz at GEOTAIL (panel 4); and gsm X components of total ion velocity and its components parallel (medium trace near zero) and perpendicular to magnetic field (panel 5). Vertical dashed lines show times (hhmmss) of significant events in different panels.

 We thus have three known times: the beginning of auroral zone Pi 2 (T0), the onset of dispersionless injection (T1), and the start of tailward flow (T2). We assume that each effect started simultaneously at X and propagated to the observation points at average velocities (V0, V1, V2). There are thus five unknowns (X, Tx, V0, V1, V2). We now constrain the possible solutions by requiring that the Pi 2 travel faster than either the inward or outward flows, and that the outward flow speed is greater than the inward. The latter is almost certainly true since V = E/B and B is increasing inward and decreasing outward. It is also likely that Alfven waves will carry the effect of the inward flow to the auroral zone before any significant dipolarization and hence induced electric field (dispersionless injection ) can be produced at synchronous orbit. Thus V0<V1 and V1<V2. There are thus three observations and two constraints to determine five average quantities.

Figure 5.: Average velocities of Pi 2 (V0), dispersionless injection (V1), and tailward flow (V2) as a function of the location X at which reconnection begins. The lines in each family are for an assumed onset at X at successive minutes before the Pi 2 begins in the auroral zone. The heavy trapezoid constrains possible solutions for V2 through model described in text.

 The model results are presented in Figure 5. The three sets of lines originating at the earth (X = 0), at synchronous orbit ( X = 6.6), and at GEOTAIL (X = 47 Re) give possible average velocities for different assumed delays of the onset (Tx) before the arrival at earth of the Pi 2 (T0). The top line is for a one minute delay, the second for two minutes, etc. The constraint V2 > V1 defines the right side of the region of thick lines, and the constraint V0 > V1 defines the left side. Together the heavy lines show of possible solutions for the average tailward flow V2, and define a possible source between synchronous orbit and about 26 Re. For example, if the source were at 20 Re behind the earth and the delay wasis one minute (top line), then V2 = 5.2 Re/min. The Pi 2 velocity is off scale and obviously 20 Re/min, while V1 is 4 Re/min. The velocity in km/s is approximately 100 times as large. Thus these solutions predict an earthward wave velocity of about 2000 km/s and a tailward flow of 500 km/s. These are quite reasonable values, but other solutions are possible as emphasized by the trapezoidal region. Note, however, that all solutions lie earthward of 25 Re, and that delays longer than a few minutes lead to average velocities that are quite low compared to the observed tailward flow. We thus conclude that observations during this event are consistent with the onset of reconnection near earth a few minutes before a Pi 2 burst begins in the auroral zone. The time at which open field line reconnection begins is not determined by this analysis. However, if the sudden change in By at the trailing edge of the flux rope (1130 UT) is a signature of open field lines, and if these field lines moved tailward at 5 Re/min, then it began 12 minutes after the onset of closed field line reconnection at 1118 UT. 

Other models for the initiation of the substorm onset are possible. However, in any of these models reconnection on closed field lines must be triggered sometime before the onset of tailward flow at the GEOTAIL spacecraft. This time is only 1.9 minutes after the onset of synchronous injection, and 4.18 minutes after the start of the auroral zone Pi 2. Reasonable assumptions about the location of this onset and delays in propagation of its effects suggest the actual commencement is earlier than any of the observed effects and quite possibly the actual cause of all of them. 

ACKNOWLEDGEMENTS

The first author would like to thank the many contributors of data some of whom are co-authors of this study. In addition he would like to thank the Solar-Terrestrial Environment Llaboratory in Toyokawa, Japan for their hospitality and support during a three month visit in the fall of 1995. He also thanks the Department of Physics of the University of Newcastle, NSW, Australia for their support during a similar visit in the summer of 1996. Partial support for the principle author has been provide by an NSF grant 443869-RM-21404, a NASA grant 443869-RM-23246 and a Los Alamos IGPP Grant 443869-RM-22935. Particular thanks go to the many government agencies accessible throughout the world via internet and the World Wide Web for their willingness to make data available automatically. Sites used in this study include NSSDC at GSFC, the SPDS through LANL, the NGDC of NOAA in Boulder, CO, and the Tromso Magnetic Observatory.

REFERENCE

Allan, W., F.W. Menk, B.J. Fraser, Y. Li, and S.P. White, Are low-latitude Pi 2 pulsations cavity/waveguide modes?, Geophys. Res. Lett., 23(7), 765-768 (1996).

Baker, D. N. ; T. I. Pulkkinen, V. Angelopoulos, W. Baumjohann, R.L. McPherron, Neutral line model of substorms: Past results and present view, J. Geophys. Res,. 101(A6), 12,975-13,010 (1996).

Birn, J. ; Hesse, M. ; Schindler, K., MHD simulations of magnetotail dynamics, J. Geophys. Res., 101(A6), 12,939-12,954 (1996).

Baumjohann W. and K.H. Glassmeier, The transient response mechanism and Pi 2 pulsations at substorm onset - Review and outlook, Planet. Space Sci., 32, 1361-1370 (1984).

Birn, J.; Hesse, M.; Schindler, K., MHD simulations of magnetotail dynamics, J. Geophys. Res., 101(A6), 12,939-12,954 (1996).

Hones, E.W., Jr., J.R. Asbridge, S.J. Bame, and S. Singer, Substorm variations of the magnetotail plasma sheet from Xsm = -6 Re to Xsm = -60 Re, J. Geophys. Res., 78(1), 109-132 (1973).

Hones, E.W., Jr., Substorm processes in the magnetotail: Comments on "On hot tenuous plasmas, fireballs, and boundary layers in the earth's magnetotail" by L.A. Frank, K.L. Ackerson, and R.P. Lepping, J. Geophys. Res., 82(35), 5633-5640 (1977).

Lui, A.T.Y., Extended consideration of a synthesis model for magnetospheric substorms, in Magnetospheric Substorms, Geophys. Monogr. # 64, edited by J. Kan, T.A. Potemera, S. Kokubun, and T. Iijima, pp. 43-60, Amer. Geophys. Union, Washington, D.C. (1991).

Lui, A. T. Y., Current disruption in the Earth's magnetosphere: Observations and models, J. Geophys. Res., 101(A6), 13,067-13 (1996).

McPherron, R.L., C.T. Russell, and M. Aubry, Satellite studies of magnetospheric substorms on August 15, 1978, 9. Phenomenological model for substorms, J. Geophys. Res., 78(16), 3131-3149 (1973).

Nishida, A., and N. Nagayama, Synoptic survey for the neutral line in the magnetotail during the substorm expansion phase, J. Geophys. Res., 78(19), 3782-3798 (1973).

McPherron, R.L,, Physical Processes Producing Magnetospheric Substorms and Magnetic Storms, in Geomagnetism, Vol. 4, edited by J. Jacobs, pp. 593-739, Academic Press, London (1991).

Rostoker, G., Phenomenology and physics of magnetospheric substorms, J. Geophys. Res. 101(A6), 12,955-12, (1996).