Pressure changes associated with substorm depolarization in the near Earth plasma sheet

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115,, doi: /2010ja015608, 2010 Pressure changes associated with substorm depolarization in the near Earth plasma sheet Y. Miyashita, 1 S. Machida, 2 A. Ieda, 1 D. Nagata, 3 Y. Kamide, 1 M. Nosé, 4 K. Liou, 5 T. Mukai, 6 S. P. Christon, 7 C. T. Russell, 8 I. Shinohara, 6 and Y. Saito 6 Received 28 April 2010; revised 27 August 2010; accepted 7 September 2010; published 17 December [1] We have studied plasma (ion) pressure changes that occurred in association with the dipolarization in the near Earth plasma sheet around substorm onsets. Using Geotail data, we have performed a superposed epoch analysis in addition to detailed examinations of two individual cases with special emphasis on the contribution of high energy particles to the plasma pressure. It is found that, unlike previously reported results, the plasma pressure does increase in association with the initial dipolarization at X > 12 R E and 2 <Y <6R E, with the increase largely due to high energy particles. Outside the initial dipolarization region, particularly tailward and duskward of this region, the plasma pressure begins to decrease owing to the magnetic reconnection before onset or before the dipolarization region reaches there. At later times, the plasma pressure tends to increase there, related to the expanding dipolarization region, but the contribution of high energy particles is not very large. These observations suggest the following. The rarefaction wave scenario proposed in the current disruption model is questionable. The radial and azimuthal pressure gradients may strengthen between the initial dipolarization and outside regions, possibly resulting in stronger braking of fast earthward flows and changes in field aligned currents. The characteristics of the dipolarization may differ between the initial dipolarization and tailward regions, which would be possibly reflected in the auroral features. Furthermore, we have examined the specific entropy and the ion b. The specific entropy increases in the plasma sheet in the dipolarization region as well as in the midtail region in conjunction with substorm onsets, suggesting from the ideal MHD point of view that the substorm processes are nonadiabatic. The ion b is found to peak at the magnetic equator in the initial dipolarization region around dipolarization onsets. Citation: Miyashita, Y., et al. (2010), Pressure changes associated with substorm depolarization in the near Earth plasma sheet, J. Geophys. Res., 115,, doi: /2010ja Introduction [2] The triggering mechanism of a substorm expansion onset is continually a major controversial issue in magnetospheric research. Various models have been proposed, among which the near Earth neutral line (NENL) model 1 Solar Terrestrial Environment Laboratory, Nagoya University, Nagoya, Aichi, Japan. 2 Department of Geophysics, Kyoto University, Kyoto, Japan. 3 Air Liquide Engineering Japan, Harima cho, Kako gun, Hyogo, Japan. 4 Data Analysis Center for Geomagnetism and Space Magnetism, Graduate School of Science, Kyoto University, Kyoto, Japan. 5 Johns Hopkins University Applied Physics Laboratory, Laurel, Maryland, USA. 6 Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, Sagamihara, Japan. 7 Focused Analysis and Research, Columbia, Maryland, USA. 8 Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California, USA. Copyright 2010 by the American Geophysical Union /10/2010JA [e.g., Baker et al., 1996] and the current disruption (CD) model [e.g., Lui, 1996] are thought to be the principal candidates. These models predict different observational features in the initial process, its location, and the propagation direction of the resultant flows or waves. In the NENL model, the magnetic reconnection must occur first in the midtail at X 20 R E, generating a tailward moving plasmoid and a fast earthward flow. The fast earthward flow propagates to the near Earth tail at X 10 R E, causing the dipolarization or the current disruption. On the other hand, the CD model predicts that the current disruption first occurs in the near Earth tail at X 10 R E, resulting in the dipolarization. This process is accompanied by a plasma pressure reduction, generating a rarefaction wave. It then propagates tailward and leads to plasma sheet thinning and weakening of the north south magnetic field. As a result, the magnetic reconnection does take place in the midtail at a later time [Lui, 1991]. [3] Here we should clarify the term dipolarization used in the present paper. There are two types of northward B z increase in association with substorm onsets. One type accompanies the fast earthward flow (or bursty bulk flow) 1of17

2 that is generated by the magnetic reconnection in the midtail at X 20 R E and propagates earthward to the near Earth tail. The flow is fairly localized, with a dawn dusk width of a few R E [Angelopoulos et al., 1997; Nakamura et al., 2004]. The B z increase is transient and seen only during fast earthward flows. After the passage of the flows, B z usually returns to the initial background level. This phenomenon is often called dipolarization front [Nakamura et al., 2002; Runov et al., 2009]. The other type of northward B z increase is the dipolarization that is related to the current disruption [Lui, 1996] and initiates in the near Earth tail at X 10 R E and whose region widely expands mainly tailward and in the dawn dusk directions in the course of an expansion phase. The northward B z persistently increases accompanied by rapid fluctuations and does not return to the initial background level shortly after the net increase, even after fast flows cease. Although there may be a causal link between these two types of northward B z increase, we think that they are morphologically different as mentioned above. Therefore, to avoid confusion, in the present paper we use the term dipolarization solely for the latter near Earth persistent B z increase, not for the former reconnection generated fast earthward flow. We also do not use the term dipolarization front for the former. [4] To understand relative timing and causal relationship between the magnetic reconnection and the current disruption, a statistical study by Miyashita et al. [2009] has revealed an overall picture of substorm associated evolution of the near Earth magnetotail. Namely, the magnetic reconnection occurs at X 16 to 20 R E at least 2 minutes before onset to create a plasmoid tailward of X 20 R E. Almost simultaneously with the magnetic reconnection, i.e., within 2 minutes, the dipolarization begins at X 7to 10 R E. [5] Machida et al. [2009] also statistically studied substormassociated evolution of the magnetotail. They proposed a new model called a catapult (slingshot) current sheet relaxation model, in which the initial action occurs between the regions of the magnetic reconnection and the initial dipolarization, leading to these two processes. Namely, magnetic field lines are highly taillike between the regions of the magnetic reconnection and the initial dipolarization. When the cross tail current in the current sheet of this midway region is strengthened by the enhancement of the Poynting flux toward the magnetic equator, magnetic flux tubes slip earthward due to the domination of the earthward J B force over the tailward pressure gradient force. The earthward flow generated by this process triggers the current disruption/dipolarization in the near Earth region. At the same time, the earthward convection of the catapult current sheet produces the thin current sheet at its tailward edge at X 20 R E, leading to the magnetic reconnection. [6] Further detailed studies, however, are needed to understand the causal relationship between the magnetic reconnection and the current disruption as well as processes midway between the regions of the two processes, such as fast earthward flows and rarefaction waves. There are possibly various approaches toward understanding the causal relationship, such as timing analysis of multispacecraft observations [e.g., Angelopoulos et al., 2008; Lui et al., 2008]. In the present study, we focus on plasma pressure changes that occur in association with the dipolarization in the near Earth plasma sheet around substorm onsets, since the plasma pressure is one of the key parameters for the substorm dynamics. [7] A few previous papers reported plasma pressure changes related to the dipolarization. For example, Lui et al. [1992] and Lyons et al. [2003] reported that the plasma pressure decreased during the dipolarization. On the other hand, Kistler et al. [1992] and Birn et al. [1997] reported that the plasma pressure increases, resulting from the energetic particle injection. It was not clearly shown, however, whether this change was related to the dipolarization, since they did not show the magnetic field data explicitly. Sergeev et al. [1998] showed that the ion and electron pressures increased deep in the inner magnetosphere at a dipolarization onset, but these pressures were partial, contributed by only high energy particles. Recently, Miyashita et al. [2009] showed that the total pressure (the sum of the ion and magnetic pressures), which is equivalent to the plasma pressure at the magnetic equator if the pressure balance holds in the Z direction, increases in the near Earth region in association with the dipolarization. In contrast, it was found to decrease tailward of this region. [8] In the present study, we have performed a superposed epoch analysis of the plasma (ion) pressure and have examined two individual cases using Geotail data. (Miyashita et al. [2009] showed the results of the total pressure, not those of the plasma pressure.) We then discuss the rarefaction wave scenario proposed in the CD model in connection with the triggering of the magnetic reconnection in the midtail. 2. Superposed Epoch Analysis [9] First, we show the results of superposed epoch analysis. The data set and the method are basically the same as those of Miyashita et al. [2009]. We utilized a total of 1,287 substorm events determined from the auroral breakup: 397 substorms from the Polar ultraviolet imager (UVI) [Torr et al., 1995] from March 1996 to December 1999 [Liou et al., 2000] and 890 substorms from the IMAGE far ultraviolet imager (FUV) [Mende et al., 2000a, 2000b, 2000c] from May 2000 to December 2005 [Frey et al., 2004; Frey and Mende, 2007]. The typical size of the auroral bulge reaches at least a few degrees in magnetic latitude and several hours in magnetic local time. In the case of a multiple onset event, we selected not only the first onset but also the following onsets for the Polar list, while only the first onset was selected for the IMAGE list. The substorm expansion onset times (t = 0) were determined with an accuracy of less than (Polar) or equal to (IMAGE) 2 minutes. During each of the selected events, Geotail was located in the magnetotail in 5 X 31 R E and Y 15 R E in GSM coordinates. Here we focus our examination on the near Earth region (X > 20 R E ), where the dipolarization occurs within 10 minutes after onset [Miyashita et al., 2009], but the midtail region (X < 20 R E ) is also included in the analysis for reference. The distribution of the Geotail locations at the substorm expansion onsets is shown in Figure 1. [10] For the Geotail data, the ion moments and the magnetic field were measured by the energy per charge analyzer (EA) of the low energy particle experiment (LEP) [Mukai et al., 1994] and the fluxgate magnetometer of the magnetic field experiment (MGF) [Kokubun et al., 1994], respectively. The time resolution of these data is 12 s, which 2of17

3 Figure 1. Locations of Geotail at the substorm expansion onsets used in the present study. corresponds to a four spin period. The ion moments were calculated from ions with an energy per charge range from a few tens of ev/q to 40 kev/q under the assumption that all ions are protons. To consider the contribution of highenergy particles beyond the instrumental range of LEP to the plasma pressure, we combined the data of energetic protons with an energy range of 44 to 265 kev, obtained from the suprathermal ion composition spectrometer (STICS) of the energetic particles and ion composition instrument (EPIC) [Williams et al., 1994]. The contribution of electrons to the plasma pressure was neglected; it is actually small, 8 to 15% [e.g., Baumjohann et al., 1989]. These data were averaged in 2 minute intervals from 11 minutes before onset to 11 minutes after onset for each event. We chose this time resolution in view of the 2 minute time accuracy of the auroral onset times. [11] The data of the plasma sheet were selected, judging from the ion b (= NkT/(B 2 /2m 0 )) for each data point: b b 1, where the boundary value between the plasma sheet (PS) and the plasma sheet boundary layer (PSBL) b 1 was defined as b 1 = 1 for X 15 R E, and log 10 b 1 = 0.14X 2.1 for X > 15 R E [Miyashita et al., 2000]. Here, while the spacecraft may remain in the plasma sheet throughout for some events, it may be away from the plasma sheet for some duration owing to the plasma sheet thinning or the flapping motion of the magnetotail for some events. The spacecraft may also stay in the lobe throughout for some events. Figure 1 shows the distribution of the spacecraft locations for all the selected events, but the actual number of the plasma sheet data points at each time can be reduced for the above reason. (The exact number of the data points at each time, although only from the Polar list, can be seen in the two dimensional plots of Miyashita et al. [2003].) [12] Figure 2 (left) shows the deviation of the north south magnetic field, DB z (t) =B z (t) B z, in the PS, where B z was defined as the average value over the interval from t = 11 to 7 minutes calculated for each event. (The deviations of the parameters shown below were also defined in the same way.) As shown by Miyashita et al. [2009], the positive DB z begins to grow, i.e., B z substantially increases northward first at X 7to 10 R E and Y 3to5R E 2 minutes before onset in association with the dipolarization. The dipolarization then occurs in the surrounding regions successively. [13] Figure 2 (middle) and Figure 2 (right) show the plasma (ion) pressure P p and its deviation DP p, respectively, in the PS, where the contributions of both high and lowenergy particles are included. It is noticeable that the plasma pressure increases rather than decreases in the initial dipolarization region at X > 12 R E and 2 <Y <6R E simultaneously with or just after the beginning of the dipolarization. Outside this region, particularly tailward and duskward of this region, the plasma pressure decreases just before or after onset, i.e., before the expanding dipolarization region reaches there. This decrease is caused by the magnetic reconnection that occurs in the midtail at X 20 R E. Namely, behind the fronts of fast earthward and tailward flows, the bulk and thermal energies converted from the magnetic energy are carried away to the flow fronts. Some diffusive particle transport may also occur to cancel a pressure gradient between the pressure decrease region and the surrounding regions [see Miyashita et al., 1999, 2000]. When the dipolarization occurs there at later times, the plasma pressure tends to increase. Note that in the region of 14 > X > 18 R E and 6 < Y <10R E, the plasma pressure is seen to decrease at t = 6 minutes, but it generally increases at later times (not shown). [14] Figure 3 shows the deviations of the partial plasma pressures contributed by high energy (EPIC STICS) and low energy (LEP) particles as well as the magnetic pressure. In the initial dipolarization region, both the plasma pressures from high and low energy particles generally increase. In particular, high energy particles largely contribute to the plasma pressure increase in association with the dipolarization. Furthermore, the plasma pressure from high energy particles increases just outside the initial dipolarization region, but it does not significantly change further away from this region. The plasma pressure from low energy particles generally decreases outside the initial dipolarization region, particularly duskward and tailward of this region. [15] The plasma pressure from low energy particles increases or decreases, depending on the location and event. Lyons et al. [2003] showed that lower energy particle fluxes decrease during the dipolarization, while higher energy particle fluxes increase; the transition from decrease to increase occurs at an energy between several and a few tens of kev, i.e., within the instrumental range of Geotail LEP. If the transition occurs at a low energy level, the plasma pressure from LEP low energy particles must increase, and vice versa. [16] Figure 4 shows the results of superposed epoch analysis of the deviations of the north south magnetic field and the total and partial plasma pressures, and the ratio of the plasma pressure from high energy particles to the total plasma pressure in the initial dipolarization region at X > 12 R E and 2 <Y <6R E. It is confirmed from Figures 4a, 4b, and 4c that, in general, the total plasma pressure increases in association with the dipolarization as well as the plasma pressure from high energy particles. Figure 4d shows that the deviation of the plasma pressure from low 3of17

4 Figure 2 4of17

5 energy particles scatters to both positive and negative large values after onset. Namely, the plasma pressure from lowenergy particles increases after onset in some events, while it decreases in other events, as mentioned above. Figure 4e indicates that the contribution of high energy particles to the total plasma pressure becomes quite high (a few tens of percentage points) after onset. [17] In our statistics shown in the right column of Figure 3, it is evident that the magnetic pressure generally decreases in the initial dipolarization region. Lopez et al. [1988] reported that while the magnetic pressure increases near the magnetic equator, it decreases away from the magnetic equator. The magnetic pressure decrease obtained here may well occur because the events were observed in the plasma sheet but away from the magnetic equator. It is also likely that the plasma sheet expansion causes a spacecraft to enter the central plasma sheet, where the magnetic pressure is smaller. [18] Furthermore, we also examined the specific entropy in the plasma sheet and the ion b in the vicinity of the magnetic equator. Figure 5 (left) shows the specific entropy (s = P i /N 5/3 i ) in the PS, where P i and N i are the ion pressure and number density, respectively. Here we assumed isotropic plasmas. In general, the specific entropy is relatively low in the near Earth region at X > 12 R E as well as near the tail flanks throughout the interval, compared with the midnight to premidnight sector of the midtail. Figure 5 (right) (Ds) shows that the specific entropy increases in the midnight to premidnight sector from the near Earth region to the midtail region after onset. In the near Earth region the specific entropy decreases before onset and increases after onset, associated with the dipolarization. The magnetic reconnection occurs in the midtail at X 20 R E [e.g., Miyashita et al., 2009], so the increase in the specific entropy in the midtail region is related to the magnetic reconnection. In any case, the increasing entropy indicates from the ideal MHD point of view that the substorm processes are nonadiabatic, as shown by previous studies [e.g., Huang et al., 1992; Baumjohann, 1993; Miyashita et al., 2001]. [19] Figure 6 shows the ion b in the vicinity of the magnetic qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiequator. Here we used only the data satisfying B 2 x þ B2 y < 10 nt. The ion b tends to increase to values of several to a few tens between t = 6 and 6 minutes in X > 12 R E and 5 <Y <10R E, i.e., around the time of the dipolarization onset in the initial dipolarization region, although some fluctuations are seen. It is quite important to note that this high b condition is favorable for the growth of the ballooning instability [e.g., Cheng and Lui, 1998; Saito et al., 2008]. 3. Case Studies [20] In this section, we show two individual events observed by Geotail as examples of the dipolarizationassociated plasma pressure increase at and near the magnetic equator of the near Earth tail. In these events, Geotail continuously remained at almost the same position relative to the magnetic equator, so that the plasma changes can be regarded as not being due to the Geotail motion relative to the magnetic equator September 1998 Event [21] In this event, Geotail continuously stayed in the close vicinity of the magnetic equator at (X,Y ) ( 9,2) R E around the substorm expansion onset. (This event was examined also by Saito et al. [2008] from the viewpoint of the ballooning mode wave, although the ballooning mode wave was not identified for this event.) Auroral images from Polar UVI show that the auroral breakup was initiated at 68 AACGM latitude and 22 h MLT at 0234:13 UT ± 18 s. The geomagnetic field model T96 [Tsyganenko, 1995] indicates that the Geotail footprint was located at 67 latitude and 23.1 h MLT, eastward and probably slightly equatorward of the initial auroral brightening region. It took 3 or 5 minutes for the auroral bulge to reach the Geotail footprint. [22] As shown in Figure 7, B x remained around zero throughout the interval from 0220 to 0255 UT, indicating that Geotail stayed at the magnetic equator throughout this interval. The B z component began to increase at 0239 UT, 5 minutes after the auroral onset, associated with the dipolarization. This delay probably resulted from the auroral bulge expansion [Liou et al., 2002]. [23] At or just after the beginning of the dipolarization, the total ion pressure from high and low energy particles (the middle thick line in the third panel from the bottom of Figure 7) considerably increased as well as the total pressure (the sum of the ion and magnetic pressures; the upper line). Since the partial ion pressure from low energy particles (the lower line) did not change appreciably, the increase in the ion pressure was largely contributed by high energy particles. It is confirmed from the energy time spectrograms in Figure 8 and ion spectra in Figure 9 that the fluxes of highenergy ions beyond 20 kev increased at the dipolarization onset at 0239 UT. The ion spectra change is consistent with the results of Lyons et al. [2003]. It is likely that the transition of particle fluxes from decrease to increase happened to occur so that the ion pressure from LEP lowenergy particles did not change very much. The contribution of the EPIC STICS high energy particles to the ion pressure was large ( 20 to 30%), although the fluxes of the highenergy particles were much smaller than those of the lowenergy particles (Figures 8 and 9) and the number density was hardly enhanced by the high energy particles (several percentage points at most; Figure 7). The ion pressure as well as the total pressure also increased 2 minutes before the dipolarization onset, owing to the increase in the number density of low energy particles. It is not clear whether this prior change was related to the dipolarization, since fast earthward flows were not seen just before the dipolarization [cf. Dubyagin et al., 2010]. It is possible that the prior Figure 2. Two dimensional plots of (left) the deviation of the north south magnetic field DB z, (middle) the plasma (ion) pressure P p, and (right) the deviation of the plasma pressure DP p on the GSM X Y plane in the plasma sheet (PS) in 4 R E 4R E bins from t = 6 to 6 minutes. The bins are slid by 2 R E on the X Y plane, and their central part of 2 R E 2R E is shown. The times shown are the centers of the averaging intervals. 5of17

6 Figure 3. Two dimensional plots of the deviations of the partial pressures contributed by (left) EPIC STICS high energy, (middle) LEP low energy particles, and (right) the deviation of the magnetic pressure in the PS in 4 R E 4R E bins from t = 6 to 6 minutes. 6of17

7 Figure 4. Results of superposed epoch analysis of the deviations of (a) the north south magnetic field, (b) the total plasma pressure, and the partial plasma pressures contributed by (c) EPIC STICS highenergy and (d) LEP low energy particles, and (e) the ratio of the plasma pressure from high energy particles to the total plasma pressure in the plasma sheet at X > 12 R E and 2 <Y <6R E. The dots indicate the 2 minute values of these parameters. The middle thick line indicates the average, whereas the upper (lower) thin line indicates the sum (difference) of the average and the standard deviation. change was caused by a small enhancement of the solar wind dynamic pressure (from 2 to 3 npa) that affected the magnetosphere several minutes after the auroral onset, although it is difficult to estimate the exact time when the dynamic pressure effect arrived at the Geotail location. The specific entropy increased by a factor of 2 at the dipolarization onset. The ion b increased around the dipolarization onset. [24] As mentioned above, the solar wind dynamic pressure enhancement affected the magnetosphere around the dipolarization onset in this event. Hence one may think that the dynamic pressure enhancement caused a plasma sheet compression, resulting in the increases in the plasma pressure and B z. However, Miyashita et al. [2010] showed that the plasma sheet compression caused by a sudden enhancement of the solar wind pressure is adiabatic; that is, the specific entropy should be constant, and changes of the number density and the temperature should be correlated. In contrast, substorm processes are nonadiabatic from the ideal MHD point of view; that is, the specific entropy should significantly increase and changes of the number density and the temperature should be anticorrelated. In the present event, the specific entropy significantly increased, and the changes of the number density and the temperature were anticorrelated at the beginning of the B z increase. Therefore, we conclude that the increase in the plasma pressure is attributed to the dipolarization process. On the other hand, the plasma pressure increase just before the dipolarization was accompanied by a nearly constant specific entropy, implying that this pressure increase corresponded to the solar wind dynamic pressure enhancement December 2001 Event [25] In this event, Geotail continuously stayed slightly off but nonetheless close to the magnetic equator at (X,Y ) ( 11,4) R E around the substorm expansion onset. Auroral images from Polar UVI show that the auroral breakup was initiated at 68 AACGM latitude and 22 h MLT at 1844:35 UT ±18 s. The Geotail footprint was located at 67 latitude and 22.6 h MLT, slightly eastward of the initial auroral brightening region, according to the T96 model. It took 1 minute for the auroral bulge to reach the Geotail footprint. [26] As shown in Figure 10, B x remained around 5 nt throughout the interval, indicating that Geotail continuously stayed near the magnetic equator. The B z component began to increase at 1845:30 UT, 1 minute after the auroral onset, associated with the dipolarization; B z had a negative spike just before the dipolarization onset. The total ion pressure from both high and low energy particles as well as the total pressure already began to increase 2 minutes before the dipolarization onset. This prior change is not clearly seen in the statistical results based on the 2 minute resolution data, possibly because the change was not large enough, although it was observed in some other events as well. The ion pressure increase was contributed mainly by low energy particles for the early part of the event from 1844:30 to 1847:30 UT and by high energy particles after that. As shown in Figure 11, earthward and dawnward fluxes with an energy of 20 kev were enhanced for the early part. Fluxes of high energy ions beyond 20 kev increased after the dipolarization onset (Figures 11 and 12), which largely contributed to the ion pressure ( 30%), in spite of very small contribution to the number density ( 5%). The specific entropy increased by a factor of 2 at the dipolarization onset. In contrast, it was constant during the ion pressure increase before the dipolarization, suggesting adiabatic compression in front of the fast earthward flow [Dubyagin et al., 2010]. The ion b enhanced around the dipolarization onset. The solar wind dynamic pressure was steady throughout the interval, so that the ion pressure increase observed by Geotail is attributed to the dipolarization. 4. Summary and Discussion 4.1. Summary of Observations [27] We have studied pressure changes that occurred in association with the dipolarization in the near Earth plasma sheet around substorm onset. Here we have considered the contribution of high energy particles to the plasma pressure. The significant findings are summarized as follows. Unlike previously reported results, the plasma pressure does increase in association with the initial dipolarization at X > 12 R E and 2 <Y <6R E, with the increase largely contributed by high energy particles. Outside the initial dipolarization region, particularly tailward and duskward of this region, the plasma pressure begins to decrease owing to 7of17

8 Figure 5. Two dimensional plots of (left) the specific entropy s and (right) its deviation Ds in the PS in 4 R E 4R E bins from t = 6 to 6 minutes. 8of17

9 Figure 6. Two dimensional plots of the ion b in the vicinity of the magnetic equator in 4 R E 4R E bins from t = 10 to 8 minutes. 9of17

10 Figure 7 10 of 17

11 Figure 8. Energy time spectrograms of ions obtained from Geotail EPIC STICS (> 44 kev, above the white line in each panel) and LEP (below the white line) from 0210 to 0310 UT on 7 September Ion differential fluxes are shown in units of cm 2 s 1 sr 1 kev 1. The EPIC STICS flux in each energy step is measured for 3 s every 24 s. The time resolution of the LEP data is 12 s. The left and right vertical lines indicate the substorm expansion onset time (0234:13 UT ±18 s) determined from Polar UVI and the dipolarization onset time ( 0239 UT), respectively. the magnetic reconnection before onset or before the dipolarization region reaches there; at later times, the plasma pressure tends to increase there, related to the expanding dipolarization region, but the contribution of high energy particles is not very large. Furthermore, we have examined the specific entropy and the ion b. The specific entropy increases in the plasma sheet in the dipolarization region as well as in the midtail region in conjunction with substorm onsets, suggesting from the ideal MHD point of view that the substorm processes are nonadiabatic. The ion b peaks at the magnetic equator in the initial dipolarization region around dipolarization onsets Interpretation of Previous Results [28] Recent THEMIS observations are also consistent with our results of the dipolarization associated plasma pressure increase [Xing et al., 2010; Dubyagin et al., 2010]. These results are obviously different, however, from those of Lui et al. [1992] and Lyons et al. [2003], who reported that the plasma pressure in the plasma sheet decreased during the dipolarization. [29] This discrepancy can be explained in the following way. In the 1 June 1985 event investigated by Luietal. [1992], the magnitude of the tailward component of the magnetic field increased at the time of the northward field dipolarization increase, whereas normally the plasma sheet tailward component would decrease under such circumstances. This unusual observation suggests that the spacecraft moved away from the central plasma sheet into a stronger tailward field region, causing the plasma pressure to decrease. Alternatively, the spacecraft was located outside the initial dipolarization region. Figure 7. The ion number density, the ion temperature, the three components of the ion velocity (thick lines) and the ion velocity perpendicular to the magnetic field (thin lines), the three components of the magnetic field, the total magnetic field, the total and ion pressures, the ion b, and the specific entropy obtained from Geotail from 0210 to 0310 UT on 7 September 1998 in GSM coordinates. At the top, the thick line indicates the number density calculated from both the LEP and EPIC STICS data, while the thin line indicates the number density calculated from only the LEP data. In the third panel from the bottom, the upper line indicates the total pressure, the middle thick line indicates the ion pressure calculated from both the LEP and EPIC STICS data, and the lower line indicates the ion pressure calculated from only the LEP data. The contribution of high energy particles is taken into account for the total pressure, the ion b, and the specific entropy. The time resolutions of the LEP, MGF, and EPIC STICS data used here are 12 s, 3 s, and 1 minute, respectively. The vertical solid line indicates the substorm expansion onset time (0234:13 UT ±18 s) determined from Polar UVI, while the vertical dashed line indicates the dipolarization onset time ( 0239 UT). 11 of 17

12 Figure 9. Selected ion spectra of omnidirectional fluxes from Geotail LEP (on the left side of the vertical dotted line) and EPIC STICS (on the right side of the vertical dotted line) for 24 s intervals without fast flows before (dotted line with diamonds), during (thick solid line with circles), and after (dashed line with plus signs) the dipolarization on 7 September The times shown are the start times of the intervals. [30] We have also reexamined the plasma pressure changes for the events of Lyons et al. [2003] in which the Geotail EPIC STICS data were available to sense high energy particles. For their 30 August 1996 and 4 September 1997 events, the ion pressure from high energy particles significantly increased, resulting in the total ion pressure increase in spite of the decrease in the ion pressure from low energy particles. For the 7 June 2000 and 13 November 1996 events, the transient fluctuations of B z due to fast earthward flows were observed until several minutes after the substorm onsets, followed by the net increase in B z (dipolarization). This indicates that Geotail was located tailward of the initial dipolarization region until several minutes after the substorm onsets and then entered the expanding dipolarization region. The ion pressure decreased just after the substorm onsets, even taking into account the contribution of highenergy particles, but this is consistent with our present results Rarefaction Wave Scenario [31] As mentioned in the Introduction, Lui [1991] proposed that the tailward propagating rarefaction wave, generated by the plasma pressure reduction associated with the current disruption or dipolarization in the near Earth region, leads to the magnetic reconnection in the midtail. However, there seems to be no solid evidence to support this rarefaction wave scenario, as discussed below in terms of the plasma pressure decrease, the propagation direction of the fast earthward flow, and the plasma sheet behavior. [32] In discussing the possibility of the rarefaction wave scenario, it is very important to clarify the duration of the rarefaction wave and the necessary growth time of an instability resulting in the magnetic reconnection. The thin plasma/current sheet and a very small B z for some duration are required for efficient triggering of the magnetic reconnection. In simulations by Shinohara et al. [2007], it takes 1 minute for the magnetic reconnection to be triggered. If the rarefaction wave scenario is true, the magnetic reconnection should be caused instantaneously or within 1 minute after the arrival of the rarefaction wave, considering that the propagation time of the rarefaction wave from the initial dipolarization region to the magnetic reconnection region is 1 to 2 minutes and that the two processes occur nearly simultaneously [Miyashita et al., 2009]. However, our statistical study with 2 minute resolution data and our case studies with 1 minute resolution data demonstrated that the plasma pressure increases in association with the dipolarization, without a preceding, transient decrease. Recent THEMIS observations with 3 s resolution data [Xing et al., 2010; Dubyagin et al., 2010] have also shown the same result. Hence, even if the pressure decrease occurs, it should be very transient, lasting for less than a few seconds, and the resultant rarefaction wave should also be very transient. [33] This small scale structure, however, may not be sustained long enough to propagate to the midtail and trigger the magnetic reconnection. The characteristic propagation speed of the rarefaction wave is given as the sound speed [Chao et al., 1977]: 600 to 1000 km/s for a proton temperature of 2 to 6 kev. Taking the transit time as 1 to 2 s according to the above discussion, the X length of the rarefaction wave is estimated at 600 to 2000 km. On the other hand, a typical proton Larmor radius is 500 to 1200 km for a magnetic field strength of 12 to 20 nt and a proton energy of 10 to 20 kev (see Figures 7 12). Although these values of the spatial scale of the rarefaction wave and the proton Larmor radius depend on the parameters, such as the proton temperature and the magnetic field, they are nearly comparable. Hence the rarefaction wave will vanish sooner or later owing to the finite Larmor radius effect. [34] Even if a very transient rarefaction wave can be sustained, it is questionable whether the magnetic reconnection can be effectively triggered by the short duration plasma sheet thinning and B z decrease [cf. Shinohara et al., 2007]. Otherwise, it is debatable whether waves generated at different sites successively reach the midtail region, leading effectively to the magnetic reconnection within 1 minute, even if the duration of each wave is very short. [35] Meanwhile, fast earthward flows at 400 km/s associated with the tailward propagating rarefaction wave should appear first on the earthward side and then on the tailward side. However, recent THEMIS observations [Runov et al., 2009; Takada et al., 2009] and a Geotail statistical study [Machida et al., 2009] showed that fast earthward flows appear in reverse order, i.e., first on the tailward side and then on the earthward side; the fast earthward flow propagates earthward from the tailward region, before the dipolarization region expands tailward [Machida et al., 2009; Takada et al., 2009]. [36] Furthermore, the pressure decrease begins away from the initial dipolarization region, i.e., in the midtail region, before the expected arrival of the rarefaction wave, as shown by our statistical studies. In addition, the plasma sheet expands in the Z direction, rather than thins, at R > 15 R E at or immediately after onset [Hones et al., 1984; Baumjohann et al., 1992]. [37] Thus these observations of the plasma pressure, flow, and plasma sheet behavior are not consistent with the rarefaction wave scenario. 12 of 17

13 Figure 10. Ion moments and the magnetic field from Geotail from 1820 to 1920 UT on 22 December 2001 in the same format as Figure 7. The vertical solid line indicates the substorm expansion onset time (1844:35 UT ± 18 s) determined from Polar UVI, while the vertical dashed line indicates the dipolarization onset time ( 1845:30 UT). 13 of 17

14 Figure 11. Energy time spectrograms of ions obtained from Geotail from 1820 to 1920 UT on 22 December 2001 in the same format as Figure 8. The left and right vertical lines indicate the substorm expansion onset time (1844:35 UT ± 18 s) determined from Polar UVI and the dipolarization onset time ( 1845:30 UT), respectively Pressure Gradient [38] We have shown that the plasma pressure increases in the initial dipolarization region at X > 12 RE around substorm onset, while it decreases tailward of this region nearly simultaneously. The total pressure also behaves in the same way [Miyashita et al., 2009]. These results imply that the earthward gradients of the plasma and total pressures strengthen at X 12 RE. This can lead to stronger braking of fast earthward flows, unless the earthward J B force also strengthens significantly. In fact, the occurrence rate of the fast earthward flow has a minimum there [Shiokawa et al., 1997; see also Machida et al., 2009]. The presence and roles of this strengthened braking should be studied in the future from the viewpoint of the effects on the processes in the near Earth region, the energy release and transport, and field aligned currents [e.g., Pu et al., 1999, 2001; Voronkov, 2005]. [39] For the azimuthal direction, we have shown that the plasma pressure increases in the initial dipolarization region near the midnight sector at the time of the dipolarization, while it decreases dawnward and duskward of the initial dipolarization region. These pressure changes will generate enhanced azimuthal gradients of the plasma pressure. This may affect the field aligned currents into or out of the ionosphere at the dawnward and duskward edges of the initial dipolarization region, although the radial gradient of the flux tube volume should be taken into account as well as the radial gradient of the plasma pressure and the azimuthal gradient of the flux tube volume [cf. Vasyliunas, 1970; Shiokawa et al., 1998; Xing et al., 2009]. The plasma pressure decreases more significantly duskward than dawnward of the initial dipolarization region (Figure 2), which may result in a difference of the intensity of the field aligned current between the duskward and dawnward sides Difference Between Initial and Later Dipolarizations [40] Furthermore, we have shown that tailward of the initial dipolarization region (X < 12 RE), the plasma pressure first decreases before the expanding dipolarization Figure 12. Selected ion spectra for the 22 December 2001 dipolarization event in the same format as Figure of 17

15 region reaches there; the plasma pressure tends to increase when the dipolarization occurs at later times, but the contribution of high energy particles is not very large. This is in contrast to the initial dipolarization region, where the plasma pressure increase occurs in association with the dipolarization, largely contributed by high energy particles. These observations imply that the characteristics of the dipolarization differ between the initial dipolarization and tailward regions. [41] This difference may be due to the differences of the background conditions. The background B z and radial gradient of the magnetic field are larger earthward than tailward of X 12 R E [e.g., Miyashita et al., 2003]. The plasma sheet characteristics are also different between these regions. Namely, the plasma pressure, density, and temperature as well as their radial gradients are larger in the earthward region [cf. Wang et al., 2001]. Shirai et al. [2005] shows that characteristics of particle flux change from fluctuating, weak particle fluxes in the outer plasma sheet (X 15 R E ) to stable, strong fluxes in the near Earth plasma sheet (X 9 R E ). Furthermore, as shown in the present study, the ion b increases around dipolarization onset in the earthward region, whereas it does not increase very much in the tailward region. The dipolarization starts in the earthward region at the beginning of the substorm energy release, while it starts in the tailward region after the energy release. It is important to study generation mechanisms of highenergy particles under such different conditions. [42] The differences of the dipolarization and resultant high energy particles in the magnetotail are possibly reflected in auroral features as well. Namely, the latitudinal extent of the auroral appearance and development (poleward expansion) depends on the auroral emission line [Friedrich et al., 2001; Mende et al., 2003]. Different emission lines reflect different species (i.e., electron or proton) and characteristic energies of precipitating particles. Mende et al. [2003] showed that both the poleward boundaries of electron and proton auroras progress poleward after onset, but the poleward boundary of proton aurora progresses less poleward than that of electron aurora. This may correspond to the difference in the increase in the ion pressure from highenergy particles between the initial dipolarization and tailward regions that we showed in the present study. More detailed studies are definitely needed, considering the behavior of electrons and waves together Specific Entropy [43] Our two dimensional plots with 2 minute resolution from in situ observations show that the specific entropy increases in the midnight to premidnight sector from the near Earth region to the midtail region after onset. In contrast, Wing and Johnson [2009] insisted from their Figure 4 that the specific entropy is roughly conserved from the growth phase to the expansion phase, although the ratio of the specific entropy for the expansion phase to that for the growth phase appears to be less than unity in the premidnight sector in 5>X > 40 R E and 0 < Y <10R E.Herethey obtained one two dimensional plot for each of the substorm phases by calculating the plasma parameters from ionospheric particle observations and mapping them to the magnetotail using the geomagnetic field model T89 [Tsyganenko, 1989]. [44] We think that this apparent inconsistency may be due to the data selection. Wing and Johnson [2009] excluded inverted V (parallel electric field) events in their analysis, so it is possible that the corresponding important processes in the magnetotail, such as current disruption, were excluded. In our analysis, we used all plasma sheet data without excluding important magnetotail processes. In addition, Wing and Johnson [2009] used only two different geomagnetic activity levels for the mapping (Kp = 4 and 1 for the growth and expansion phases, respectively), although magnetic configuration change during the expansion phase probably progresses continuously, not stepwise. Namely, different Kp values should be used for the early, middle, and late expansion phases. Hence Wing and Johnson [2009] may have used an expansion phase value at a location more earthward than it actually is, particularly for the early expansion phase, in comparing the corresponding growth phase value. If so, this resulted in underestimating the ratio of the expansion phase value to the growth phase value, since the specific entropy decreases as distance from the Earth decreases, as seen in Wing and Johnson [2009] and the present study. [45] In spite of the above problems, looking at Figures 2a and 2b of Wing and Johnson [2009], the specific entropy roughly appears to increase in the expansion phase in the premidnight sector at X 10 to 30 R E, which is consistent with our results. [46] The increasing specific entropy during the expansion phase suggests from the ideal MHD point of view that the substorm processes are nonadiabatic. On the other hand, the specific entropy decreases in the near Earth region before dipolarization onset (Figure 5). The 7 September 1998 event also shows the decrease in the specific entropy and the anticorrelation of the changes of the number density and the temperature before the dipolarization onset (Figure 7). It is possible that these changes before dipolarization onset are still adiabatic, from a nonideal MHD point of view, where particles with different energies, not plasma as a whole, behave adiabatically [Wang et al., 2009]. 5. Concluding Remarks [47] We have shown that the plasma pressure increases in association with the dipolarization. Some substorm models, however, have been constructed, assuming the decrease in the plasma pressure. It is very important to incorporate our observational results into substorm models or to construct a new substorm model which can explain our results. [48] Acknowledgments. The Geotail MGF magnetic field data were provided by T. Nagai and S. Kokubun. The Geotail EPIC high energy particle data were provided by A. T. Y. Lui and R. W. McEntire. The Polar UVI and IMAGE FUV auroral imager data were provided by G. K. Parks and S. B. Mende, respectively. The auroral breakup list based on the IMAGE FUV data were provided by H. U. Frey. We thank M. H. Saito, K. Seki, K. Maezawa, Y. Asano, and K. Shiokawa for their valuable comments. [49] Masaki Fujimoto thanks Donald Fairfield and another reviewer for their assistance in evaluating this paper. References Angelopoulos, V., et al. (1997), Magnetotail flow bursts: Association to global magnetospheric circulation, relationship to ionospheric activity 15 of 17

16 and direct evidence for localization, Geophys. Res. Lett., 24(18), , doi: /97gl Angelopoulos, V., et al. (2008), Tail reconnection triggering substorm onset, Science, 321(5891), Baker, D. N., T. I. Pulkkinen, V. Angelopoulos, W. Baumjohann, and R. L. McPherron (1996), Neutral line model of substorms: Past results and present view, J. Geophys. Res., 101(A6), 12,975 13,010, doi: / 95JA Baumjohann, W. (1993), The near Earth plasma sheet: An AMPTE/IRM perspective, Space Sci. Rev., 64(1 2), Baumjohann, W., G. Paschmann, and C. A. Cattell (1989), Average plasma properties in the central plasma sheet, J. Geophys. Res., 94(A6), , doi: /ja094ia06p Baumjohann, W., G. Paschmann, and T. Nagai (1992), Thinning and expansion of the substorm plasma sheet, J. Geophys. Res., 97(A11), 17,173 17,175, doi: /92ja Birn, J., M. F. Thomsen, J. E. Borovsky, G. D. Reeves, D. J. McComas, and R. D. Belian (1997), Characteristic plasma properties during dispersionless substorm injections at geosynchronous orbit, J. Geophys. Res., 102(A2), , doi: /96ja Chao, J. K., J. R. Kan, A. T. Y. Lui, and S. I. Akasofu (1977), A model for thinning of the plasma sheet, Planet. Space Sci., 25(8), Cheng, C. Z., and A. T. Y. Lui (1998), Kinetic ballooning instability for substorm onset and current disruption observed by AMPTE/CCE, Geophys. Res. Lett., 25(21), , doi: /1998gl Dubyagin, S., V. Sergeev, S. Apatenkov, V. Angelopoulos, R. Nakamura, J. McFadden, D. Larson, and J. Bonnell (2010), Pressure and entropy changes in the flow braking region during magnetic field dipolarization, J. Geophys. Res., 115, A10225, doi: /2010ja Frey, H. U., and S. B. Mende (2007), Substorm onsets as observed by IMAGE FUV, in Proceedings of International Conference on Substorms 8, edited by M. Syrjäsuo and E. Donovan, pp , University of Calgary, Alberta, Canada. Frey, H. U., S. B. Mende, V. Angelopoulos, and E. F. Donovan (2004), Substorm onset observations by IMAGE FUV, J. Geophys. Res., 109, A10304, doi: /2004ja Friedrich, E., J. C. Samson, I. Voronkov, and G. Rostoker (2001), Dynamics of the substorm expansive phase, J. Geophys. Res., 106(A7), 13,145 13,163, doi: /2000ja Hones, E. W., Jr., T. Pytte, and H. I. West Jr. (1984), Associations of geomagnetic activity with plasma sheet thinning and expansion: A statistical study, J. Geophys. Res., 89(A7), , doi: /ja089ia07p Huang, C. Y., L. A. Frank, G. Rostoker, J. Fennell, and D. G. Mitchell (1992), Nonadiabatic heating of the central plasma sheet at substorm onset, J. Geophys. Res., 97(A2), , doi: /91ja Kistler, L. M., E. Möbius, W. Baumjohann, G. Paschmann, and D. C. Hamilton (1992), Pressure changes in the plasma sheet during substorm injections, J. Geophys. Res., 97(A3), , doi: /91ja Kokubun, S., T. Yamamoto, M. H. Acuña, K. Hayashi, K. Shiokawa, and H. Kawano (1994), The GEOTAIL magnetic field experiment, J. Geomagn. Geoelectr., 46(1), Liou, K., C. I. Meng, P. T. Newell, K. Takahashi, S. I. Ohtani, A. T. Y. Lui, M. Brittnacher, and G. Parks (2000), Evaluation of low latitude Pi2 pulsations as indicators of substorm onset using Polar ultraviolet imagery, J. Geophys. Res., 105(A2), , doi: /1999ja Liou, K., C. I. Meng, A. T. Y. Lui, P. T. Newell, and S. Wing (2002), Magnetic dipolarization with substorm expansion onset, J. Geophys. Res., 107(A7), 1131, doi: /2001ja Lopez, R. E., D. G. Sibeck, A. T. Y. Lui, K. Takahashi, R. W. McEntire, T. A. Potemra, and D. Klumpar (1988), Substorm variations in the magnitude of the magnetic field: AMPTE/CCE observations, J. Geophys. Res., 93(A12), 14,444 14,452, doi: /ja093ia12p Lui, A. T. Y. (1991), A synthesis of magnetospheric substorm models, J. Geophys. Res., 96(A2), , doi: /90ja Lui, A. T. Y. (1996), Current disruption in the Earth s magnetosphere: Observations and models, J. Geophys. Res., 101(A6), 13,067 13,088, doi: /96ja Lui, A. T. Y., R. E. Lopez, B. J. Anderson, K. Takahashi, L. J. Zanetti, R. W. McEntire, T. A. Potemra, D. M. Klumpar, E. M. Greene, and R. Strangeway (1992), Current disruptions in the near Earth neutral sheet region, J. Geophys. Res., 97(A2), , doi: /91ja Lui, A. T. Y., et al. (2008), Determination of the substorm initiation region from a major conjunction interval of THEMIS satellites, J. Geophys. Res., 113, A00C04, doi: /2008ja Lyons, L. R., C. P. Wang, T. Nagai, T. Mukai, Y. Saito, and J. C. Samson (2003), Substorm inner plasma sheet particle reduction, J. Geophys. Res., 108(A12), 1426, doi: /2003ja Machida, S., Y. Miyashita, A. Ieda, M. Nosé, D. Nagata, K. Liou, T. Obara, A. Nishida, Y. Saito, and T. Mukai (2009), Statistical visualization of the Earth s magnetotail based on Geotail data and the implied substorm model, Ann. Geophys., 27(3), Mende, S. B., et al. (2000a), Far ultraviolet imaging from the IMAGE spacecraft. 1. System design, Space Sci. Rev., 91(1 2), Mende, S. B., et al. (2000b), Far ultraviolet imaging from the IMAGE spacecraft. 2. Wideband FUV imaging, Space Sci. Rev., 91(1 2), Mende, S. B., et al. (2000c), Far ultraviolet imaging from the IMAGE spacecraft. 3. Spectral imaging of Lyman a and OI nm, Space Sci. Rev., 91(1 2), Mende, S. B., H. U. Frey, B. J. Morsony, and T. J. Immel (2003), Statistical behavior of proton and electron auroras during substorms, J. Geophys. Res., 108(A9), 1339, doi: /2002ja Miyashita, Y., S. Machida, A. Nishida, T. Mukai, Y. Saito, and S. Kokubun (1999), GEOTAIL observations of total pressure and electric field variations in the near and mid distant tail associated with substorm onsets, Geophys. Res. Lett., 26(6), , doi: /1999gl Miyashita, Y., S. Machida, T. Mukai, Y. Saito, K. Tsuruda, H. Hayakawa, and P. R. Sutcliffe (2000), A statistical study of variations in the near and middistant magnetotail associated with substorm onsets: GEOTAIL observations, J. Geophys. Res., 105(A7), 15,913 15,930, doi: / 1999JA Miyashita, Y., S. Machida, T. Mukai, Y. Saito, and P. R. Sutcliffe (2001), Mass and energy transport in the near and middistant magnetotail around substorm onsets: Geotail observations, J. Geophys. Res., 106(A4), , doi: /2000ja Miyashita, Y., S. Machida, K. Liou, T. Mukai, Y. Saito, H. Hayakawa, C. I. Meng, and G. K. Parks (2003), Evolution of the magnetotail associated with substorm auroral breakups, J. Geophys. Res., 108(A9), 1353, doi: /2003ja Miyashita, Y., et al. (2009), A state of the art picture of substormassociated evolution of the near Earth magnetotail obtained from superposed epoch analysis, J. Geophys. Res., 114, A01211, doi: / 2008JA Miyashita, Y., K. Keika, K. Liou, S. Machida, Y. Kamide, Y. Miyoshi, Y. Matsumoto, I. Shinohara, Y. Saito, and T. Mukai (2010), Plasma sheet changes caused by sudden enhancements of the solar wind pressure, J. Geophys. Res., 115, A05214, doi: /2009ja Mukai, T., S. Machida, Y. Saito, M. Hirahara, T. Terasawa, N. Kaya, T. Obara, M. Ejiri, and A. Nishida (1994), The low energy particle (LEP) experiment onboard the GEOTAIL satellite, J. Geomagn. Geoelectr., 46(8), Nakamura, R., et al. (2002), Motion of the dipolarization front during a flow burst event observed by Cluster, Geophys. Res. Lett., 29(20), 1942, doi: /2002gl Nakamura, R., et al. (2004), Spatial scale of high speed flows in the plasma sheet observed by Cluster, Geophys. Res. Lett., 31, L09804, doi: / 2004GL Pu, Z. Y., et al. (1999), Ballooning instability in the presence of a plasma flow: A synthesis of tail reconnection and current disruption models for the initiation of substorms, J. Geophys. Res., 104(A5), 10,235 10,248, doi: /1998ja Pu, Z. Y., A. Korth, Z. X. Chen, Z. X. Liu, S. Y. Fu, G. Zong, M. H. Hong, and X. M. Wang (2001), A global synthesis model of dipolarization at substorm expansion onset, J. Atmos. Sol. Terr. Phys., 63(7), Runov, A., V. Angelopoulos, M. I. Sitnov, V. A. Sergeev, J. Bonnell, J. P. McFadden, D. Larson, K. H. Glassmeier, and U. Auster (2009), THEMIS observations of an earthward propagating dipolarization front, Geophys. Res. Lett., 36, L14106, doi: /2009gl Saito, M. H., Y. Miyashita, M. Fujimoto, I. Shinohara, Y. Saito, K. Liou, and T. Mukai (2008), Ballooning mode waves prior to substormassociated dipolarizations: Geotail observations, Geophys. Res. Lett., 35, L07103, doi: /2008gl Sergeev, V. A., M. A. Shukhtina, R. Rasinkangas, A. Korth, G. D. Reeves, H. J. Singer, M. F. Thomsen, and L. I. Vagina (1998), Event study of deep energetic particle injections during substorm, J. Geophys. Res., 103(A5), , doi: /97ja Shinohara, I., K. G. Tanaka, and M. Fujimoto (2007), Quick triggering of magnetic reconnection in magnetotail, Earth Moon Planet., 100(3 4), Shiokawa, K., W. Baumjohann, and G. Haerendel (1997), Braking of high speed flows in the near Earth tail, Geophys. Res. Lett., 24(10), , doi: /97gl Shiokawa, K., G. Haerendel, and W. Baumjohann (1998), Azimuthal pressure gradient as driving force of substorm currents, Geophys. Res. Lett., 25(7), , doi: /98gl Shirai, H., T. Hori, and T. Mukai (2005), Boundary between the near Earth (< 10 R E ) plasma sheet and outer plasma sheet: Equatorial observations at 9 15 R E by Geotail, Adv. Polar Upper Atmos. Res., 19, of 17