Flux transport, dipolarization, and current sheet evolution during a double onset substorm

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116,, doi: /2010ja015865, 2011 Flux transport, dipolarization, and current sheet evolution during a double onset substorm R. Nakamura, 1 W. Baumjohann, 1 E. Panov, 1 A. A. Petrukovich, 2 V. Angelopoulos, 3 M. Volwerk, 1 W. Magnes, 1 Y. Nishimura, 3 A. Runov, 3 C. T. Russell, 3 J. M. Weygand, 3 O. Amm, 4 H. U. Auster, 5 J. Bonnell, 6 H. Frey, 6 D. Larson, 6 and J. McFadden 6 Received 26 June 2010; revised 21 February 2011; accepted 25 February 2011; published 27 May [1] We study a substorm with two onsets (at 0220 and 0243 UT) that occurred during a gradual northward interplanetary magnetic field (IMF) turning on 16 February At these times, Time History of Events and Macroscale Interactions during Substorms (THEMIS) and GOES spacecraft were distributed between 6.6 and 18 R E downtail. Prior to the weak auroral electrojet onset at 0220 UT, a thin current sheet was extended near R E. After the onset, Earthward fast flows with dipolarization fronts followed by signatures of magnetic flux pileup were detected in this region. The 0243 UT onset disturbances were more intense and centered at higher latitudes. The reconnection region tailward of 18 R E became activated and reached the lobe flux. We suggest that activations of reconnection Earthward of 18 R E associated with the 0220 UT event led to pileup of flux and redistribution of B Z to form a thin current sheet with small B Z in the midtail region. This made conditions favorable for reconnection tailward of 18 R E involving lobe flux for the 0243 UT onset. The reconfiguration process in the current sheet between the two onsets possibly enabled a relatively strong, high latitude substorm despite the rather weak IMF driver. The near Earth dipolarization observed after the 0243 UT onset was accompanied by more localized Earthward flows and flow reversals. Differences in dipolarization signatures could be caused by ambient plasma condition and field configuration between these two events. Our observations of the double onset substorm suggest that the plasma sheet can be preconditioned by both the IMF driver and internal magnetotail processes. Citation: Nakamura, R., et al. (2011), Flux transport, dipolarization, and current sheet evolution during a double onset substorm, J. Geophys. Res., 116,, doi: /2010ja Introduction [2] A magnetospheric substorm is the sudden release of magnetotail energy that was transported from the dayside into the auroral ionosphere and nightside magnetosphere. During substorms, high speed plasma flows, called bursty bulk flow (BBF), and magnetic field dipolarization, enhancement(s) in B Z, are observed in the near Earth tail. These processes change the distribution of the tail current locally and/or globally. [3] One of the most widely accepted mechanisms for creating fast flows is magnetic reconnection. Although the 1 Space Research Institute, Austrian Academy of Sciences, Graz, Austria. 2 Space Research Institute, Russian Academy of Sciences, Moscow, Russia. 3 Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California, USA. 4 Finnish Meteorological Institute, Helsinki, Finland. 5 Institut für Geophysik und Extraterrestrische Physik, Technische Universität Braunschweig, Braunschweig, Germany. 6 Space Science Laboratory, University of California, Berkeley, California, USA. Copyright 2011 by the American Geophysical Union /11/2010JA occurrence rate of BBFs, which indicate enhanced magnetic flux transport, is known to be well correlated with substorm activity [Baumjohann et al., 1990], the role of the flow in substorm evolution, in particular, the timing, location, and presence of the reconnection site [e.g., Baumjohann et al., 2007; Lui et al., 2008; Angelopoulos, 2008], is still debated. Some observations support reconnection as the cause of substorm initiation; others consider it to be the result of substorm initiation. It should be noted that although BBFs also exist during nonsubstorm intervals, they are always associated with some auroral precipitation [Nakamura et al., 2001] and a distinct ionospheric equivalent current pattern [Juusola et al., 2009]. [4] Two types of dipolarization have been identified in the near Earth magnetotail. One, also called a dipolarization front, is reported to be associated mainly with Earthward flows exceeding 100 km/s. Such a dipolarization, attributed to magnetic flux transported Earthward [e.g., Angelopoulos et al., 1994; Sergeev et al., 1996; Nakamura et al., 2002; Sigsbee et al., 2005; Runov et al., 2009], is considered to be a rather thin front structure that precedes the fast flow. The other type of dipolarization, associated with recovery from a thin current sheet state to a dipolar configuration, propagates 1of19

2 more tailward [Baumjohann et al., 1999] or in the azimuthal direction [e.g., Nagai, 1982]. The latter dipolarization may be associated with pileup of Earthward transported magnetic flux in the near Earth region where the flows encounter a strong field and a high pressure region and are decelerated (or braked) [Shiokawa et al., 1997; Baumjohann, 2002]. Pileup of the magnetic flux then leads to a tailward motion of the deceleration region [e.g., Hesse and Birn, 1991]. Instabilities, such as ballooning [Roux et al., 1991] or cross field current instabilities [Lui et al., 1991], have also been proposed to produce a dipolarization signature and subsequent tailward and azimuthal propagation [Baumjohann et al., 2007]. Recent Cluster observations [Nakamura et al., 2009] that showed systematic relationships between these two types of dipolarization; (an Earthward, BBF associated dipolarization front followed by tailward propagating dipolarization) support the pileup effect for those events. [5] The location and time, with respect to substorm onset, of near Earth magnetic reconnection and other key processes in the magnetotail have been reported in several statistical studies using the extensive data set from Geotail [Asano et al., 2004; Miyashita et al., 2004; Nagai et al., 2005; Miyashita et al., 2009]. According to Asano et al. [2004], the near Earth magnetic reconnection site is located close to the tailward edge of a thin current sheet that develops between X 5 and 20 R E [Miyashita et al., 2009]. Miyashita et al. [2009] also determined that on average, magnetic reconnection in the premidnight tail at X 16 to 20 R E and dipolarization, which occurs between X 7 and 10 R E, take place almost simultaneously, about 2 min before onset. However, for intense substorms, the reconnection site is located closer to the Earth, which is consistent with the dependence of ionospheric onset latitude on substorm intensity [Miyashita et al., 2004]. The location of the reconnection site within the magnetotail has also been found to be dependent on the solar wind electric field value, V x B s, where V x is the x component of the solar wind velocity and B s is the southward component of the interplanetary magnetic field prior to the onset [Nagai et al., 2005]. Hence, the radial distances where critical current sheet processes occur can differ for each substorm depending on internal current sheet configuration and external solar wind condition. [6] Substorms often have multiple onset signatures in both the ionosphere and the magnetotail. One distinct type of multiple onset sequence is a major onset with clear poleward expansion preceded by weak onsets with no such expansion during the growth phase. These weak onsets are referred to as pseudobreakups, or more accurately, growth phase pseudobreakups [Kullen et al., 2010]. Although pseudobreakups have typical magnetospheric signatures of substorms, their strength and magnetotail features are quite different from those of a major onset [Koskinen et al., 1993;Ohtani et al., 1993; Nakamura et al., 1994]. For example, the current wedge and the particle injections are confined to a localized region and expand very little tailward, Earthward [Koskinen et al., 1993; Ohtani et al., 1993], and in the azimuthal direction [Ohtani et al., 1993; Nakamura et al., 1994]. Auroral signatures during growth phase pseudobreakups are observed to be localized in azimuth and typically occur at a higher latitude than the major onset region [Nakamura et al., 1994]. Statistical results of substorm studies in which a stronger onset was associated with a thin current sheet and a reconnection region closer to the Earth [Miyashita et al., 2004] suggest such relationships. [7] Another suggested difference between major onsets and pseudobreakups during multionset substorms is that major onsets are associated with a lobe reconnection, whereas pseudobreakups involve only closed magnetic field lines within the plasma sheet [Russell, 2000; Mishin et al., 2001; Sergeev et al., 2008; Pu et al., 2010; Tang et al., 2010]. If closed field line reconnection evolves toward a lobe reconnection, then major onsets can occur poleward of the first onset in the ionosphere, which explains observations of a poleward expansion of onset disturbances in the ionosphere [Russell, 2000; Sergeev et al., 2008]. [8] While poleward expansion of the onset can be produced by evolution of the same reconnection site toward the lobe [Russell, 2000; Sergeev et al., 2008], it can be also associated with tailward excursion of the reconnection region [Tang et al., 2010]. Tailward retreat of the activation sites has been observed in the midtail region by IMP8 and Geotail [Angelopoulos et al., 1996]. Although there are a number of studies of double onset substorms, the evolution of the current sheet and flux transport processes in the magnetotail are not yet well understood. Monitoring of current sheet configuration, the location of the reconnection site and simultaneous observations of the ionosphere are essential to understand the evolution of multiple onset substorm. [9] The Time History of Events and Macroscale Interactions during Substorms (THEMIS) mission [Angelopoulos, 2008], with its five spacecraft and a dedicated array of ground observatories located in Canada and the northern United States, was designed to study substorm evolution in the midtail and inner magnetosphere on a large scale. In this paper we study midtail and inner magnetospheric disturbances during a doubleonset substorm beginning at 0220 UT on 16 February The first weak electrojet onset was located at lower latitude compared to the second substorm onset at 0243 UT, which occurred at higher latitude with a stronger activation. For both onsets, dipolarization signatures were detected by THEMIS inner probes and a comparable or stronger dipolarization was observed in the first weak onset. Multipoint magnetospheric observations together with ground based observations enable us to understand large scale current sheet evolution and the flux transport process in the course of the two onsets. Such investigations from multipoint measurements are essential to understand not only substorm dynamics but also local processes such as the dipolarization front. In this study we use THEMIS spacecraft data from the FluxGate Magnetometer (FGM) experiment [Auster et al., 2008], the Electric Field Instrument (EFI) instrument [Bonnell et al., 2008], the electrostatic analyzer (ESA) [McFadden et al., 2008], and the Solid State Detector (SST) [Sibeck and Angelopoulos, 2008]. In addition, we use data from the THEMIS Ground Based Observatories (GBO) all sky cameras (ASI) and magnetometers [Mende et al., 2008], as well as geomagnetic field data from the International Real time Magnetic Observatory Network (INTERMAG), a global network of cooperating digital magnetic observatories. The spacecraft location in the 2of19

3 Figure 1. (a) Location of THEMIS and GOES 10 and 12 spacecraft in the equatorial plane in GSM coordinates. (b) Foot point location of THEMIS, GOES 10 and 12 spacecraft, and the closest ground based magnetometers and/or all sky imagers in geographic coordinates. magnetosphere and ionospheric foot points are shown in Figure Overview [10] On 16 February 2008, two westward electrojet intensifications accompanied by Pi2 magnetic fluctuation onsets started at 0220 and 0243 UT, as shown in Figure 2. The gradual westward electrojet enhancement at 0220 UT was relatively weak (<200 nt) compared with the strong (>500 nt) enhancement of the electrojet at 0243 UT detected in the high latitude magnetogram and clear positive bay enhancement at midlatitude stations. The 0220 UT intensification took place during gradual recovery of the southward IMF, which was between B Z 2and 4 nt for the preceding 45 min and then increased to between B Z 1and 0 nt. The 0243 UT activation took place during a gradual 4 nt northward IMF turning. THEMIS spacecraft were distributed in the premidnight side of the magnetotail, as shown in Figure 1a, and their foot points were located near the local time sector where enhancement of the westward electrojet, Pi2 pulsations, and the positive bays were detected (Figure 1b). The foot points are calculated using the adapted time dependent magnetospheric model [Kubyshkina et al., 2009], in which the standard Tsyganenko model (T96) is modified to find the best fit to the observed field from all the relevant spacecraft in 5 min steps. [11] The strongest westward electrojet intensification during this period was detected at Iqaluit (IQA), which is located more than 4 northward of the foot point of P3 5(THD,E,A) and northward of the standard AE stations. Such high latitude substorms have also been called contracted oval substorms [Akasofu et al., 1973; Lui et al., 1976]. The strong electrojet and preceding southward IMF interval, however, are dissimilar to characteristics of the original contracted oval substorm events, which were associated with quiet geomagnetic activity and northward IMF periods. [12] Auroral activity and nightside magnetospheric observations during the two onsets are summarized in Figures 3 and 4. The keograms in Figures 3b and 3c show gradual equatorward motion of the arc associated with the growth phase (indicated by slanted arrows) at Kuujjuaq (KUUJ) and Sanikiluaq (SNKQ) before a weak intensification at around 0216 UT at KUUJ (indicated by the vertical arrow directed upward) and 0217 UT near SNKQ. The intensification near SNKQ cannot be easily identified in the keogram due to different longitudinal location, but can be seen in the all sky image in Figure 5, as discussed later. At 0220 UT a clear auroral onset was recorded at SNKQ (downward directed arrow) and then from 0222 UT the aurora was also observed at KUUJ (downward directed arrow) followed by both equatorward and poleward auroral expansion. As the auroral arc further intensified, another onset of poleward expansion started at 0243 UT at SNKQ (upward directed arrow) and then at 0244 UT at KUUJ (upward directed arrow). This activity also reached the field of view of Rankin Inlet (RANK), as plotted in Figure 3a. The keogram of KUUJ (Figure 3c) also shows that the aurora expanded farther equatorward than the previous onset at 0220 UT. The foot points of the THEMIS spacecraft were distributed around these auroral regions. In particular, the foot points of P3(THD) and P4(THE) were within the field of view of KUUJ during the interval (see Figure 5 and its description for details). [13] Figures 3d and 3e show the total pressure and the sum of the magnetic field pressure and ion pressure. The magnetic field pressure is calculated using only the B X and B Y components assuming pressure balance in a planar tail current sheet, as in previous studies [e.g., Xing et al., 2010]. The ion pressure is calculated using SST and ESA instrument data. No particle measurement data were available for P1(THB). To better examine the current sheet signatures associated with the two onsets, we plotted the magnetic field data for P5(THA), P3(THD), and P4(THE) in Figures 3 and 4 in a coordinate system that takes into account the tilt of the dipole and the hinging distance. The location of the hinging distance and the effective dipole tilt for the event were calculated using a model by Tsyganenko and Fairfield [2004] with IMF B Z = 3of19

4 Figure 2. Overview of IMF and ground magnetic activity between 1 and 4 UT on 16 February Shown are (a) the IMF B Z from ACE, (b) the H component from six high latitude magnetograms, (c) the Pi2 pulsations observed at five stations, and (d) the H component from two lower latitude stations. The ACE observations are shifted by 37 min to represent the IMF values at the magnetopause. The vertical lines show the two ground magnetogram onsets at 0220 and 0243 UT. Locations of the magnetogram stations used in this plot are shown in Figure 2b. 0, IMF B Y = 4 nt, solar wind dynamic pressure of 1.5 npa, and solar wind speed of 630 km/s. The hinging distance was estimated to be 11.0 R E, which indicates that the three inner magnetospheric THEMIS spacecraft, P5(THA), P3(THD), and P4(THE), were Earthward of this location. The effective dipole tilts for the three spacecraft were estimated to be 15.2, 14.7, and 15.7, respectively. To calculate the coordinate system for the three inner magnetospheric THEMIS spacecraft, we used the average tilt, 15, and plotted the magnetic field for the spacecraft in a tilted coordinate system (see Figures 3g, 3i, and 4a). The midtail data from P1(THB) and P2(THC) are plotted in GSM coordinates (see Figures 3f and 3h). For GOES 10 and 12 (shown in Figures 4b and 4c), we used the conventional dipole coordinate (HDV) system. Figure 3j shows the cumulative magnetic flux transport from P2(THC) and P4(THE), which is obtained by integrating E Y. This parameter has been found to be a useful quantity to show the enhanced flux transport rate and to detect the onset of reconnection [Angelopoulos et al., 2009; Liu et al., 2010]. We integrated the Y component of the V B electric field starting from 0200 UT. [14] Prior to the 0220 UT onset, typical growth phase signatures, such as an increase in the total pressure (Figures 3d and 3e), were observed. This suggests that magnetic flux accumulated in the tail because of enhanced dayside reconnection during southward IMF B Z. A decrease in B Z (Figures 3h and 3i) can be identified most clearly at P1(THB) and P4(THE), but not at P5(THA), suggesting progressive stretching of the tail in the premidnight region. Furthermore, the absolute value of B X increased at all spacecraft (Figures 3f and 3g), suggesting large scale thinning of the current sheet. Such a sequence of stretching and thinning has been reported as characteristic of the growth phase [Petrukovich et al., 2007]. All the three inner magnetospheric THEMIS spacecraft were located at a similar distance (about R E ) and distributed in both hemispheres. The enhanced current density in this region can be seen in the difference between B X in the northern hemisphere (average of P3(THD) and P4(THE)) and B X in the southern hemisphere (from P5(THA)), as shown in Figure 4a. Figure 4a also gives our estimate of the plasma sheet thickness, L. We estimated L using the Harris current sheet assumption with these B X values and the lobe field value obtained from the average total pressure and by assuming that the plasma pressure is negligibly small in the lobe. Before dipolarization, the plasma sheet thickness decreased gradually to about 0.9 R E, thinner than an average plasma sheet, 6 R E [Baumjohann and Paschmann, 1990] (obtained from AMPTE/IRM observations), but not thin enough for the typical near Earth current sheet instability, which requires current sheet thickness less than an ion scale. The ion inertial length and the gyroradius were in the km range for this event. 4of19

5 Figure 3. Auroral and magnetospheric observations between 2 and 3 UT. Keogram from (a) RANK, (b) SNKQ, (c) KUUJ, (d e) total pressure, (f g) B X from THEMIS spacecraft in the midtail P1 2(THB C) andintheinnermagnetospherep3 5(THD E,A), respectively, and (h i) B Z in the same format as B X. For total pressure, only data from P2(THC) are shown during this event. (j) Cumulative flux transport from P2(THC) and P5(THE) calculated by integrating the y component of the electric field determined with (V B). The flux transport value for P2(THC) is multiplied by two to show the temporal change more clearly. The vertical lines indicate the onset times at 0220 and 0243 UT. 5of19

6 Figure 4. (a) The DB X (local current density) and L (current sheet thickness) determined using THEMIS data, (b) H, and (c) D components from the GOES 10 and 12 magnetic field instruments. The vertical lines indicate the onset times at 0220 and 0243 UT. [15] At around 0218 UT, P4(THE) observed enhancements in B Z and B X, most likely associated with localized, short auroral intensifications at 0216 UT (Figure 3b). In the midtail at P2(THC), the pressure began to decrease at around 0219 UT (Figure 3d), indicating the start of unloading or magnetic flux decrease in this region. The pressure in the near Earth tail region, on the other hand, increased further (Figure 3e) at the same time. These midtail and near Earth magnetotail pressure changes were reported as a typical signature in the tail during substorm onset [e.g., Miyashita et al., 2009; Xing et al., 2010]. At P2(THC), B Z turned negative (Figure 3h) as the flux transport started to increase (Figure 3j). Such signatures are expected with onset of reconnection [Angelopoulos et al., 2009; Liu et al., 2010], and the reconnection site was located Earthward of the P2(THC) spacecraft based on the negative excursion in B Z. The sharp dipolarization at P3(THD) and P4(THE) in the near Earth tail started at around 0222 UT (Figure 3i), and flux transport at P4(THE) increased significantly (Figure 3j). This dipolarization is associated with enhanced Earthward transport of the magnetic flux. During this onset interval, no clear signature of dipolarization was observed at P5(THA)(Figure 3i) and at geosynchronous orbit (Figure 4b) except some small fluctuation at GOES 10 starting at 0219 UT. We interpret these observations to mean that for the 0220 UT onset, dipolarization was limited to a local region in the near Earth plasma sheet. [16] During the 15 min interval between the two onsets, the total pressure in the midtail (THC) decreased on average compared to its value before the 0220 onset, suggesting that magnetic flux was removed from the midtail region while the IMF B Z was between 0 and 1 nt, indicating a very weak solar wind driver. Furthermore, the absolute B X values increased gradually in the midtail at P1(THB) but not at P2(THC) (Figure 3f), indicating that the current sheet was still thinning at the location of P1(THB). At P2(THC), B Z changed sign from negative to positive, suggesting that the possible reconnection region changed from the Earthward to the tailward side of the spacecraft. Evolution of the possible reconnection region will be discussed in more detail in section 3. In the near Earth tail region, on the other hand, the total pressure gradually increased mainly at P5(THA) and slightly increased also at P3 4(THD E). The local current density decreased, and the thickness of the plasma sheet increased (Figure 4a). Progressive reconfiguration toward a thicker plasma sheet with dipolar configuration was particularly prominent at the innermost spacecraft, P5(THA), where B Z increased the most. Although the assumption of a 1D Harris current sheet might not accurately describe the current sheet configuration after dipolarization due to these increases in B Z and the growing difference in the total pressure among the spacecraft, it is still valid to argue that the plasma sheet does thicken. Flux transport increased further during this interval, but only in the midtail P2(THC) (Figure 3j). These observations suggest that the magnetic flux transported from the midtail accumulated in the near Earth tail and inner magnetosphere region during this interval. [17] The second onset (at 0243 UT) was associated with a decrease in pressure (Figure 3d) and an increase in B Z at P2(THC) and at P1(THB). Both P2(THC) and P1(THB) detected a decrease in the absolute value of B X, most likely due to plasma sheet expansion. The inner THEMIS probes measured further plasma sheet thickening and dipolarization. During this second onset, GOES 10 and 12 also observed a brief decrease in B H and an increase in B D (GOES 10 preceded GOES 12 by about 5 min). Such signatures have been interpreted as the westward propagation of the duskside of the current wedge located tailward of the spacecraft [Nagai, 1982]. [18] Ionospheric equivalent currents and aurora for selected times during the two substorms are shown in Figure 5. The ionospheric equivalent currents are obtained using the Spherical Elementary Current Systems (SECS) technique [Amm, 1997; Amm and Viljanen, 1999]. To determine these equivalent ionospheric currents, ground magnetometer data 6of19

7 Figure 5. The ionospheric equivalent current vectors and all sky auroral images for selected times. The ionospheric equivalent currents are shown in the geographic coordinates, while the all sky images are in geomagnetic coordinate system. The foot points of the THEMIS spacecraft shown to the left use the same color scheme as in Figures 1 3. The asterisks in the all sky images and the dots in the equivalent current plots mark the foot points obtained from the adapted magnetospheric model, and the squares in the all sky images indicate the foot points obtained from the T89 model. The area covered by the all sky image plots is given in the top left equivalent current plot. 7 of 19

8 from the THEMIS, CANMOS, CARISMA, GIMA, and MACCS arrays in Northern America and from the Greenland array are used. Mosaic images obtained from THEMIS GBO all sky cameras at RANK, SNKQ, and KUUJ are shown at the right side of each panel. Magnetic foot points from THEMIS and the equivalent current vectors are also shown in the auroral plots. The foot points marked with asterisks (and colored dots in the equivalent current plots) are obtained from the adapted time dependent model [Kubyshkina et al., 2009]; those marked with diamonds are calculated from the Tsyganenko T89 magnetic field model for comparison. The main difference between these foot point positions for this event is the latitude. T89 predicted that the foot point of the spacecraft would be located 1 2 degrees higher in latitude for all probes but P1(THB). As previously discussed, the Kubyshkina et al. adapted modeling is a time dependent model that takes into account changes in spacecraft fields and input parameters. Hence, adapted time dependent modeling is believed to provide better estimates of foot point location. [19] A very localized auroral activation between KUUJ and SNKQ began at 02:16:12 along the faint aurora at 68 latitude, westward and northward of the foot point of P4(THE) (see 0217 UT panel). This auroral activation did not expand until about 0220 UT when a surge like structure (indicated by an arrow in the 0220 UT panel) began to form between SNKQ and KUUJ. This surge developed westward, reaching the zenith meridian of SNKQ (left arrow in 0223 UT panel). East of that surge, around the local time of the foot point of P4(THE) and P3(THD), two slanted, north south aligned auroral signatures developed (right arrow). The region of the north south aligned aurora then intensified and expanded locally poleward (shown in 0225 UT panel). By 0227 UT the westward expansion stopped and began to dim (not shown). A localized westward electrojet region centered at around 69 geomagnetic latitude, east of the surge and north southaligned aurora, developed by 0223 UT. [20] The second onset began with intensification of an auroral arc at around 70 to the west of the SNKQ all sky imager at 02:42:42 UT. The intensification became visible within the field of view of KUUJ, expanding northward as well as westward (arrow in the 0243 UT panel), until it also became visible at RANK. As the expanding aurora s poleward edge moved out of KUUJ s field of view, diffuse aurora mixed with fine structured, north south filamentary aurora developed equatorward (arrow in the 0247 UT panel) where the foot points of P4(THE) and P3(THD) were located. The westward electrojet for the second onset was stronger and distributed over a wider region than for the previous onset. It was also centered at a very high latitude region ( 75 ) with its eastern edge most likely located east of the Greenland stations. 3. Observation in the Midtail and the Flow Braking Region [21] As described in section 2, P2(THC), which was located in the southern hemisphere in the outer edge of the plasma sheet, observed plasma sheet thinning and expansion associated with the two auroral onsets. Figures 6a 6d show magnetic field and ion velocity; the latter is a merged product between the two particle instruments, ESA and SST. The Y component of the V B electric field, cumulative flux transport, ion energy spectra from ESA and SST, and electron spectra from ESA are shown in Figures 6e 6h. Figures 6i 6k are the electron pitch angle spectra from selected times, indicated by the solid bars at the base of Figure 6h. Note that the resolutions in the bottom plots of Figures 6j and 6k are different from that in the bottom plot of Figure 6i due to different modes of operation for the instrument, i.e., burst mode and reduced mode [see McFadden et al., 2008]. [22] Both ion and electron energy decreased before 0220 UT during current sheet thinning. The spacecraft remained in the outer plasma sheet during the 0220 UT onset. B Z turned negative just before onset (Figure 6b); E Y turned positive (Figure 6e); and cumulative flux transport (Figure 6f), which is associated with transient enhancement in V Z and B Z, started to increase from 02:20:30 UT (indicated by dashed line). Such E Y and V Z enhancements have been interpreted as evidence of X line activation [Angelopoulos et al., 2009]. After the onset B Z stayed mainly negative; B Y increased; and the flow direction became more persistently tailward and mostly parallel to the field. [23] At 02:30:00 UT another increase in the flux transport occurred along with an enhancement in V Z, B Z, V X, and B Y (indicated by dashed line), again suggesting activation of an X line Earthward of the spacecraft. The slight difference between the electron energy flux for pitch angle 0 and that for 180 (Figure 6i) supports this interpretation. After 0236 UT B Z turned mainly positive. [24] A third increase in the flux transport, which took place at 02:38:20 UT, was associated with an increase in V Z, B Z and V X, suggesting that this event is due to the X line tailward of the spacecraft. At 0243 UT when V Z and B Z increased, the first signatures of an Earthward electron beam with an energy of a few kev were detected when the lower energy (500 ev) electrons were restricted in the tailward direction (Figure 6j). This suggests that the spacecraft was near the region of the open lobe field line, where low energy electrons come exclusively from the Earth, and very close to the X line at the tailward side of the spacecraft, where Earthward directed energetic beams could be produced. Following this observation, the spacecraft shortly moved into the lobe and then reentered deeply into the plasma sheet, where it measured Earthward fast flows associated with the final increase in the flux transport rate at about 02:44 UT (indicated with the dashed vertical line). During this flux transport increase, more energetic Earthward electron beams (10 kev) were observed together with tailward streaming lower energy beams (Figure 6k). These electron beam properties have been observed in many previous reconnection studies [e.g., Fujimoto et al., 2001; Nagai et al., 2003; Angelopoulos et al., 2008]. [25] The flux transport profile and the particle and magnetic field signatures at P2(THC) suggest that four main reconnection events took place during the interval. The first two enhancements took place due to reconnection Earthward of the spacecraft; the latter two took place tailward of the spacecraft. Thus, a transient tailward retreat of the reconnection region took place with these multiple activations. [26] Figure 7 summarizes the field and plasma signatures of the dipolarization events observed at P3(THD) and P4(THE). P4(THE) was located 1.2 R E Earthward, 0.5 R E duskward, 8of19

9 Figure 6 9of19

10 Figure 7. Field and plasma observations from P3 (THD in the top row) and P4 (THE in the bottom row) between (left) 0221 and 0227 UT and (right) 0243 and 0249 UT. Shown in each quadrant are three components of the magnetic field (first panel), three components of the E B velocity (second panel), the pressure (third panel), and the electron energy spectra (fourth panel) from P3(THD) and P4(THE). Figure 6. Fields and plasma data observed by P2(THC) between 0215 and 0250 UT. (a) B X, (b) B Y and B Z components of the magnetic field, (c) three components of the ion velocity, (d) V X and V Z components of the ion velocity perpendicular to the magnetic field, (e) E Y calculated from (V B) Y, (f) cumulative flux transport, (g) ion energy spectra from ESA and SST, and (h) electron energy spectra from ESA. The vertical solid lines show the ground onset times at 0220 and 0243 UT, and the vertical dashed lines show the activation times of the reconnection. (i k) The electron pitch angle profiles for selected times, which are marked with a thick vertical bar at the bottom of Figure 6h. For Figures 6i 6k, the upper parts show the energy spectra for different pitch angles; the lower parts show the phase space density profile for parallel (0 ), antiparallel (180 ), and perpendicular (90 ) directions to the field lines. 10 of 19

11 Figure 8. Changes in the equatorial flow pattern observed by (top) P3(THD) and (bottom) P4(THE) between (left) 0221 and 0226 UT and (right) 0244 and 0249 UT. Shown in each quadrant are the B Z component (first panel), the V X component (second panel), the flow vectors (third panel), azimuthal angle of the flow (fourth panel) from P3(THD) and P4(THE) in GSM. The flow data shown here are the E B drift velocity. Four Hz data are plotted for the B Z, V X, and azimuthal angle panels and every 30th point (8.5 s) is given for the flow vectors. The vertical lines indicate time of determination of the dipolarization front direction listed in Table 1. The crosses in the V X panels indicate the times for which the flow velocity vector is also plotted in Figure 9 as a representative value for the fast flow. The crosses in the azimuth angle plot indicate times when the horizontal flow speed exceeded 200 km/s. and 0.5 R E northward of P3(THD). In Figure 7 as well as Figures 8 10, we used E B/B 2 velocity in GSM coordinates obtained from the EFI measurements. The two spin plane components were used for determining the spin axis component by assuming E B = 0. To check for consistency, we also compared the E B/B 2 velocity with the velocity from the ion moment obtained by combining EFT and SST data. Since we are interested in the flux transport process and therefore in the perpendicular component of the flow, we used the flow velocity obtained from the EFI measurement, which has the advantage of high temporal resolution, for further detailed analysis below. [27] Associated with the 0220 UT onset, a sharp enhancement in B Z accompanied by a subsequent Earthward fast flow disturbance was observed at both spacecraft starting at 02:21:59 UT for P3(THD) and at 02:22:17 UT for P4(THE) (indicated by the vertical lines in Figure 7 (left)). The Earthward flow and the dawn dusk electric field (not shown) were about 3 times larger for P3(THD), indicating that a flowbraking process was taking place. The dipolarization was preceded by an increase in the plasma pressure due to a density enhancement (not shown). The plasma was therefore compressed at the head of the dipolarization front. For both spacecraft, enhancements in the B Z component and fast flows 11 of 19

12 Figure 9. Dipolarization front obtained from variance analysis and the direction of the fast flow are shown for (a, d) the 0222 UT event, (b, e) the 0244 UT event, and (c, f) the 0246 UT event for P3(THD) and P4(THE). The dotted lines show the dipolarization front; the thin arrows perpendicular to the dotted lines show the estimated velocity of the front; and the thick arrows show the direction of the fast flow in the GSM X Y plane (Figures 9a 9c) and in X Z plane (Figures 9d 9f). were associated with an increase in B X in the front, also suggesting enhancement in the local tail current density (or thinning of the current sheet). In the disturbed fast flows after 0223 UT, however, B Z enhancements were associated with adecreaseinb X, and the plasma sheet magnetic field configuration became more dipolar (B Z > B X ), more apparent at P4(THE). [28] The dipolarization for the 0243 UT onset shown in the right column of Figure 7 looks quite different from that for the 0220 UT onset shown in the left column of Figure 7. The magnetic field and plasma signatures were also different between the P3(THD) and P4(THE) spacecraft. The first enhancement in B Z was associated with predominantly dawnward flow at P3(THD) starting at 02:44:16 UT and P4(THE) at 02:44:47 UT (indicated by the first vertical line in Figure 7 (left)). Dipolarization fronts associated with fast Earthward flow were also observed starting at 02:45:42 UT in P3(THD) and 02:46:14 UT at P4(THE) (indicated by the second vertical line). The flow direction at P3(THD) turned tailward and again Earthward within minutes, suggesting highly structured flows. Such frequent flow reversals were not detected by P4(THE), but the flow had multiple peaks. While the changes in the pressure associated with the dipolarization in P4(THE) were similar to the dipolarization front observed in the 0220 UT event, changes in the pressure were less clear at P3(THD). [29] When the dipolarization fronts were encountered, the electron energy was enhanced for most events. But this was not the case for the first dipolarization event observed by P3(THD). This first dipolarization event was preceded by a decrease in B X, which suggests that the spacecraft may have encountered the hotter central plasma sheet population before the dipolarization front. The observed abrupt electron energy enhancements in the other dipolarization fronts, on the other hand, are consistent with the idea that these dipolarization fronts are related to the injection process [Sergeev et al., 2009]. There are also cases, however, in which the electron energy changes at a boundary not particularly relevant to an Earthward fast flow, such as the first front observed by both spacecraft for the 0243 UT event and later during the second interval. [30] The evolution of the flow pattern near P4(THE) and P3(THD) (i.e., a scale size of about 1 R E ) and the shear structure around the dipolarization front can be inferred by examining the flow vectors and the changes in the azimuthal flow angle, V = arctan(v Y /V X ), as shown in Figure 8. While the azimuthal angle is plotted using the 4 Hz field data, the flow vectors are shown only for every 30th data point (8.5 s) to more clearly display the data. In the azimuthal angle panel, times when the equatorial flow speed exceeds 200 km/s are marked with a cross. For the 0220 UT onset (Figure 8, left), flow evolution sequences were similar at the two spacecraft, except that the flow speed was larger and the azimuthal component stronger for P3(THD) than for P4(THE). We believe that the left column of Figure 8 shows that initially the fast flow was directed Earthward and duskward but then followed by another fast flow stream that was directed mainly Earthward and dawnward. The flow pattern after 0243 UT 12 of 19

13 Figure 10. (left) The magnetic field vector (first panel), the velocity vector determined from the electric field drift (second panel), the electric field (third panel), the velocity vector direction (fourth panel), and the magnetic field direction (fifth panel) associated with the 0222 UT dipolarization front observed by P3(THD) plotted in maximum variance coordinates of the electric field. The l, m, n indicate the minimum, intermediate, and maximum variance direction of the electric field. (top right) Their vectors in GSM coordinates are given in the upper right corner. (bottom right) Flow and magnetic field disturbance relative to the dipolarization front (solid line) is illustrated. onset, on the other hand, was significantly different for the two spacecraft. The first dipolarization was associated with dawnward flow at both spacecraft. The flows at P3(THD) then changed to duskward, while at P4(THE), they stayed dawnward. Although the maximum flow was directed Earthward and slightly duskward for both spacecraft, no coherent pattern between the two can be identified. [31] In addition to the evolution of the flow pattern, the angle, V, decreases ahead of all the dipolarization fronts. (See the third and sixth panels of Figure 8.) This suggests clockwise rotation of the flow disturbance viewed from north around the Earthward moving front. Such flow shear is expected to create an anticlockwise magnetic perturbation that corresponds to an upward field aligned current [Sergeev et al., 1996]. [32] The orientations of the dipolarization fronts marked with vertical lines in Figure 8 are examined based on minimum (maximum) variance analysis of the magnetic and electric fields. The required velocity of the boundary for the latter analysis was obtained through Minimum Faraday Residue (MFR) analysis [see Sonnerup and Scheible, 1998; Sonnerup et al., 2008; and references therein] (for details of these analysis methods). Table 1 summarizes the dipolarization front characteristics. In Figure 9 orientation of the dipolarization front and its normal velocity vectors are plotted with dotted lines and thin vectors, respectively. The thick arrow shows the maximum speed flow vector around the dipolarization front. The normal direction shown here is obtained from the maximum variance of the electric field because for the majority of the cases, we obtained a larger ratio of the first two eigenvalues, l E1 /l E2 (maximum to intermediate), calculated from variance analysis of the electric field, than the ratio, l B2 /l B3 (intermediate to minimum), which we calculated from the variance analysis of the magnetic field. In the following we use the boundary coordinates obtained from the electric field, in which n is the normal direction to the boundary, l is the minimum variance of the electric field direction, and m is the intermediate variance direction. [33] As shown in Table 1, we obtained a reasonable DeHofmann Teller velocity, which means that its component along the normal direction, V HT n, has a value comparable to the normal velocity obtained from MFR method, U n,as well as to the average plasma normal velocity, hv n i, for only the first dipolarization event for both spacecraft. For the other events, no clear DeHofmann Teller velocity existed, or the normal velocity obtained from the variance analysis differed significantly from the DeHofmann Teller velocity. Using 13 of 19

14 Table 1. Orientation and Motion of the Dipolarization Front Obtained From Maximum Variance Analysis of the Electric Field Spacecraft Time n X n Y n Z l E1 /l E2 (km/s) U n hv n i (km/s) hb n i (nt) V HT n (km/s) THD 02:21:55 02:22: (620) THE 02:22:15 02:22: (800) THD 02:44:10 02:44: (220) THE 02:44:40 02:45: a 220 (.. a ) THD 02:45:38 02:45: a 1320 (.. a ) THE 02:46:12 02:46: a 1780(.. a ) a V HT failed to be determined. b Obtained using dipolarization time scale and U n or normal component of V HT (in parentheses). d b (km) the obtained velocity of the dipolarization front and the time scale of the dipolarization (B Z enhancement), we estimate that the boundary scale size varies between 220 and 1800 km. The ion inertial length is about 300 km in the first event and 600 km in the second event, which means that these boundaries are between a fraction and 3 ion inertial lengths, as expected in thin dipolarization fronts [Nakamura et al., 2002; Runov et al., 2009]. It should be noted that the first and the third dipolarization fronts are consistent with fronts created by Earthward flow (see also Figures 9a, 9c, 9d, and 9f), whereas the fast flow of the second event is nearly parallel to the dipolarization front for both spacecraft (see also Figures 9b and 9e). The normal component of the field is relatively small in most cases, indicating that these fronts are tangential discontinuities. [34] The field and shear flow pattern near the dipolarization front at 0222 UT observed by P3(THD) is given in Figure 10. The average magnetic field magnitude normal to the boundary is about 4 nt, about 30% of the horizontal component. The speed of the boundary obtained using the MFR method, U n = 109 km/s, is comparable to the normal component of the DeHofmann Teller velocity, 138 km/s. The normal component of the flow velocity, V n, was relatively stable compared to the other components and was on average 201 km/s, slightly higher than but still comparable to the boundary speed. Hence we believe that this boundary can be regarded as a stable structure at least over the time the boundary and its surrounding structure take to pass over the spacecraft, i.e., 20 s. The flow eventually turned mainly along the normal direction. Such a flow rotation has also been observed behind dipolarization fronts [Nakamura et al., 2002], indicating that the flow shear structure is created by high speed flow interacting with the ambient field. Clear shear structures observed in the intermediate component of the field and the flow, V m and B m, can be seen in the vector plots. The observed flow shear and magnetic perturbation pattern are illustrated Figure 10 (right) in GSM. The rotation shown in the l, m, n coordinate corresponds to a clear magnetic (flow) shear pattern in a clockwise (anticlockwise) direction observed from north and anticlockwise (clockwise) observed from the tail, indicating a structure associated with a current flowing out from the ionosphere. This is consistent with the duskside edge of a BBF [Sergeev et al., 1996]. [35] The foot point P3(THD) is predicted to be northeast (using the T89 model) or slightly southeast (using the adaptive model) of northwest to southeast directed slant auroral structures (see 0223 UT panel in Figure 5). The foot point of P4(THE), on the other hand, maps with the region of auroral precipitation. The conjugate location of auroral precipitation has been shown to correspond to the duskside edge of the BBF [Nakamura et al., 2001]. Hence, the tendency of the aurora to be duskside rather than dawnside of the foot points in this event indicates that the aurora relevant to the BBF observations at P3 4(THD E) corresponds to these slanted N S auroral structures. [36] We can obtain the spatial scale of the dipolarization front using the speed of the boundary, km/s, and the temporal scale of the dipolarization front, 4.5 s. The estimated spatial scale of the dipolarization front was between 490 and 620 km, which is 1 2 ion inertial scale lengths for this event. The B l rotation suggests that this front is associated with a strong current j m 70 na/m 2. The observed shear flow region lasted for 20 s, which then corresponded to a shear about 3200 km wide. From the B m disturbance, we can also estimate the possible field aligned current density, which was determined to be 10 na/m 2 on average. The length of the dipolarization boundary then would be about 3800 km, if this magnetic shear produces a field aligned current intensity of 0.1 MA, which corresponds to the value estimated from flow vortices at 11 R E in a different THEMIS event analyzed by Keiling et al. [2009] that was associated with an auroral spiral. 4. Discussion [37] In this study we examined the ground signatures, midtail, and near Earth plasma sheet during a double onset substorm when a weakly southward IMF B Z gradually turned northward (+4 nt). The center of the electrojet and the aurora were located at higher latitudes than normal (above 71 in geomagnetic latitude) beyond the standard AE stations, which detected only a small AL decrease of 250 nt. An intense westward electrojet with a peak magnitude of 500 nt was observed at the high latitude station Iqaluit during the second substorm. The higher latitude distribution of the disturbance recorded by magnetometer arrays and auroral images has been identified as characteristic of northward IMF B Z substorms [Akasofu et al., 1973]. Interestingly, even during such high latitude substorms, when the IMF input was relatively small, active plasma sheet signatures (fast flows) have been reported [Petrukovich et al., 2000]. It was suggested that during such small substorms, magnetotail coherency in the inner and midtail regions across the tail is lost [Petrukovich et al., 2000]. The double onset substorm event discussed in this study also shows typical signatures of a substorm within the plasma sheet, including clear dipolarization, fast flows, current wedge features at geosynchronous orbit, and midtail reconnection signatures despite the contracted oval. A particularly interesting feature of this event is the apparent dif- 14 of 19

15 Figure 11. Sketch of the magnetotail configuration and possible location of the reconnection in the thin current sheet based on the THEMIS spacecraft observed around the 0220 UT onset (first panel), the 0230 and the 0238 UT activation events (second and third panels), and the 0243 UT onset (fourth panel). ference in the strength and the location of the two onsets, which is rather unexpected, given the relationships between weak and strong substorms examined statistically in the study by Miyashita et al. [2004] or the relationship between growth phase, pseudobreakup, and the major onset [Ohtani et al., 1993; Nakamura et al., 1994]. In other words, the second onset, which took place during a weaker IMF B Z period, was more disturbed in the ionosphere as well as in the magnetosphere than the first onset, and the location of the second onset was determined to be further tailward in the magnetosphere and poleward in the ionosphere than the first onset. [38] We suggest that these seemingly inconsistent features can be explained by the reconfiguration of the current sheet that took place between these two onsets (illustrated in 15 of 19