Construction of magnetic reconnection in the near Earth magnetotail with Geotail

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116,, doi: /2010ja016283, 2011 Construction of magnetic reconnection in the near Earth magnetotail with Geotail T. Nagai, 1 I. Shinohara, 2 M. Fujimoto, 2 A. Matsuoka, 2 Y. Saito, 2 and T. Mukai 2 Received 12 November 2010; revised 6 January 2011; accepted 10 February 2011; published 20 April [1] The Geotail spacecraft made in situ observations of magnetic reconnection on 15 May 2003 in the near Earth magnetotail at a radial distance of 28 R E when a moderate substorm started on the ground. For this event, the intense cross tail electron current layer was detected in association with the simultaneous plasma flow and magnetic field reversals, and the scale length along the GSM x axis was obtained. This observation enables us to deduce scales for the basic structure of magnetic reconnection in the near Earth magnetotail. In the center of the electron current layer (a possible X line), the speed of the dawnward electron flow carrying cross tail current exceeds the maximum of the electron outflow speed (earthward/tailward), and it is highly super Alfvénic. The full extent of this central intense cross tail electron current layer is approximately 1 li (ion inertial length) in the x direction, which corresponds to 0.2 R E in the magnetotail. Electron outflow speed reaches its maximum, which is also super Alfvénic, at distances of less than 1 li from the X line, and ion outflow speed reaches its maximum farther away from the center. Electron outflow speed decreases in the downstream region, and it becomes the same as the ion speed at distances of 4 li. The full extent of the ion electron decoupling region is 8 li in the x direction, which corresponds to 1.5 R E in the magnetotail, and the outer region belongs to the MHD regime. Inside the ion electron decoupling region, ions are accelerated up to 10 kev during inflow processes and further accelerated beyond 40 kev toward the duskward direction near the center and along the x axis slightly away from the center. These observations of the ion and electron dynamics in the close vicinity of the X line are fairly consistent with results from the two dimensional particle in cell simulation described here and others. The present Geotail results provide the observational basis for the structure of magnetic reconnection in the near Earth magnetotail. Citation: Nagai, T., I. Shinohara, M. Fujimoto, A. Matsuoka, Y. Saito, and T. Mukai (2011), Construction of magnetic reconnection in the near Earth magnetotail with Geotail, J. Geophys. Res., 116,, doi: /2010ja Introduction [2] Magnetic reconnection occurring in the near Earth magnetotail in association with an onset of substorms is ideal for exploring essential nature of magnetic reconnection. This type of magnetic reconnection can be classified as antiparallel reconnection in the symmetrical system with no initial guide (out of plane) fields. The Alfvén velocity is high (usually >2000 km s 1 ) so that outflow plasmas generated by magnetic reconnection are easily discriminated from other environmental plasmas. Inflowing plasmas are readily identified. Acceleration and heating of electrons are evident. It is not clear at the present stage, however, whether magnetic reconnection in the near Earth magnetotail is 1 Earth and Planetary Sciences, Tokyo Institute of Technology, Tokyo, Japan. 2 Institute of Space and Astronautical Science, JAXA, Sagamihara, Japan. Copyright 2011 by the American Geophysical Union /11/2010JA externally driven or not. First, we briefly review our current observational knowledge on magnetic reconnection in the near Earth magnetotail. A concise review on observations of magnetic reconnection in other regions of the magnetosphere is given by Paschmann [2008]. Second, we clarify what has not been fully investigated with available spacecraft observations. Finally, we present our main targets of this study, in which the nature of magnetic reconnection in the near Earth magnetotail is examined with in situ observations by the spacecraft Geotail Review of Present Observations [3] In situ observations of magnetic reconnection in the near Earth magnetotail have been carried out extensively by the spacecraft Geotail and the four spacecraft Cluster. Magnetic reconnection in association with substorm onsets takes place mostly at radial distances of R E [e.g., Nagai et al., 1998a], and its location is controlled by the solar wind electric field [Nagai et al., 2005; Nagai, 2006]. Under strong solar wind electric fields, magnetic reconnection tends 1of18

2 to take place closer to the Earth. Indeed, many studies using Cluster observations (the apogee of 19 R E ) have been made for magnetic reconnection during large substorms under highly disturbed conditions. [4] Nagai et al. [1998a] report various characteristics of magnetic reconnection on the basis of ion and electron velocity distribution functions. Ions show high speed outflows perpendicular to the local magnetic field near the equatorial plane, and the outflows consist of hot ions having counterstreaming structure. The counterstreaming structure is made by ions coming from the northern and southern tail lobes. Counterstreaming ions are also found near the current sheet in the Cluster observations and is considered to be important constitution of magnetic reconnection [Wygant et al., 2005]. The counterstreaming structure can be also observed in plasmoids [Hoshino et al., 1998], a leading edge of tailward flows [Nagai et al., 1998b], and a leading edge of earthward flows not related with substorm onsets [Nagai et al., 2002]. Since these counterstreaming ions usually exist with a strong north south magnetic field, existence of the counterstreaming structure may not be a unique signature in the immediate vicinity of magnetic reconnection and the counterstreaming structure of ions can form in the flux piled up region (a head of high speed outflows). Coexistence of inflowing ions and outflowing ions is commonly observed off the equatorial plane. Inflowing ions are not cold like tail lobe ions and have high speed, so that these ions are already accelerated and heated in the inflowing process. The most evident signature in the immediate vicinity of magnetic reconnection is strong heating and acceleration of electrons. This electron signature is commonly observed by Geotail [e.g., Shinohara et al., 1998; Asano et al., 2004a] and Cluster [e.g., Nakamura et al., 2006a; Borg et al., 2007; Asano et al., 2008]. Fluxes of low energy electrons are reduced and fluxes of high energy electrons are enhanced, so that velocity distribution functions have a flat top shape. Electrons show an isotropic flat top distribution off the equatorial plane as well as near the equatorial plane. However, field aligned nature (high energy electrons flowing out) becomes evident in the outer layer. One outstanding finding is that medium energy (less than 5 kev) electrons flowing into the reconnection site exist in the outermost layer. It is suggested by Nagai et al. [1998a] that these electrons are current carriers in the Hall current loops originally proposed by Sonnerup [1979] [see also Fujimoto et al., 1997]. These basic ion and electron distributions are identified commonly in data of subsequent studies [e.g., Asano et al., 2004a, 2008; Wilber et al., 2004; Nakamura et al., 2006a, 2008; Retinò et al., 2008; Runov et al., 2008; Zhou et al., 2009]. [5] Topics on the Hall physics [Sonnerup, 1979] have been widely studied, and the existence of the Hall current system has been well established observationally for magnetic reconnection in the near Earth magnetotail [Nagai et al., 2001, 2003]. Electrons show basically bidirectional field aligned distributions near the separatrix layer [see also Hoshino et al., 2001]. The above mentioned mediumenergy electrons flowing into the reconnection site exist indeed in four quadrants of the x z meridian plane in the magnetospheric coordinates. These inflowing mediumenergy electrons are identified as carriers of the Hall current. The Hall quadruple magnetic field structure is easily identified. Although these observations indicate mainly fieldaligned current parts of the Hall current system, ion electron velocity difference near the equatorial plane shows existence of the Hall term of the generalized Ohm s law, and the Hall electric field toward the equatorial plane is observed near the separatrix layer. [6] Using the ability of four spacecraft observations, Cluster has clarified various important characteristics of magnetic reconnection. Runov et al. [2003] show the X line structure of the magnetic field in association of flow reversal from tailward to earthward by calculating field line curvature vectors in the 1 October 2001 substorm event near the peak of a storm (Dst = 150 nt). This event is widely used in subsequent studies [e.g., Wilber et al., 2004; Cattell et al., 2005; Kistler et al., 2005; Wygant et al., 2005; Imada et al., 2007; Chen et al., 2008, 2009; Deng et al., 2009;Egedal et al., 2010]. The magnetic null geometry is also studied [e.g., Xiao et al., 2006; He et al., 2008a, 2008b]. The intensity of field aligned Hall currents is calculated using magnetic field measurements with the curlometer technique and using direct electron measurements [e.g., Alexeev et al., 2005; Nakamura et al., 2006a, 2008]. The intensity up to 60 na m 2 is obtained. The value obtained by Cluster appears to be extremely higher than the value of 6 13 na m 2 obtained on the basis of ion and electron measurements by Geotail [Nagai et al., 2001]. It should be noted that the Cluster events are generally obtained for large substorms in highly disturbed periods. The Hall quadruple magnetic field structure is easily identified near the reconnection site [e.g., Runov et al., 2003; Borg et al., 2005, 2007; Nakamura et al., 2006a; Eastwood et al., 2007, 2010a, 2010b; Laitinen et al., 2007; Østgaard et al., 2009]. The large Hall electric field up to 50 mv m 1 is also identified [e.g., Borg et al., 2005; Wygant et al., 2005; Henderson et al., 2006; Eastwood et al., 2007]. Again, the value obtained by Cluster is extremely higher than the value of 10 mv m 1 obtained by Geotail [Nagai et al., 2001]. The electric field from electron pressure divergence (less than 1 mv m 1, opposite to the Hall electric field) is also obtained [Henderson et al., 2006]. The field aligned electron distributions have been studied extensively as a signature of electron behavior in the vicinity of the magnetic reconnection site [e.g., Egedal et al., 2005, 2010; Chen et al., 2008, 2009; Wang et al., 2010]. The ability of the four spacecraft observations makes a great advantage in estimating the current sheet thickness. The widely accepted view from the Cluster observations is that the full thickness of the reconnection site (in the z direction in the magnetospheric coordinates) is order of ion inertial lengths [e.g., Nakamura et al., 2006a, 2006b; Xiao et al., 2007; Runov et al., 2008] Unsolved Issues [7] The electron diffusion region, where electrons are unmagnetized and the magnetic field lines are reconnected, is expected to be located in the center of the reconnection site. It is advocated by Scudder and Daughton [2008] and Scudder et al. [2008] that electron agyrotropy is the most unambiguous indicator of the electron diffusion region. Present electron measurements do not have ability of measuring electron agyrotropy adequately. Chen et al. [2008, 2009] show existence of electron beams along the electric field, however, full distribution functions are not presented. 2of18

3 This is partly caused by the fact that high time resolution electron data are sorted in pitch angles in Cluster observations. An intense dawn dusk current, which is carried by dawnward moving electrons, should exist in the electron diffusion region. It is known that the thin intense current layer forms in two dimensional particle in cell simulations for magnetic reconnection [e.g., Hesse et al., 1999, 2001; Pritchett, 2001; Shay et al., 2001]. Recently, elongation of this current layer and its structure are studied in numerical simulations [Fujimoto, 2006; Daughton et al., 2006; Karimabadi et al., 2007; Shay et al., 2007]. The intense cross tail current layer can be a good indicator of the electron diffusion region in current available data obtained in the near Earth magnetotail. However, the existence of the intense out of plane current layer in the center of the reconnection site and its spatial extent had not been explored in current spacecraft observations yet Main Targets of This Paper [8] Geotail has provided a large number of observation data in the near Earth magnetotail since In the survey using the data obtained in , there are 34 reconnection events in the near Earth plasma sheet in association with ground substorm onsets [Nagai et al., 2005]. However, there are only a few events in which flow reversal from tailward to earthward takes place well inside the plasma sheet. Among these events, the 15 May 2003 event only shows a high speed dawnward electron flow with speed of >6000 km s 1. This speed is super Alfvénic. In other events, speed of the dawnward electron flow is enhanced during the flow reversal, however, it reaches only up to 2000 km s 1 or the magnetic field is highly variable during one plasma sampling. The speed up to 1000 km s 1 can be found in the events in which only MHD outflows (electron outflow velocity is equal to ion outflow velocity) are observed. Hence, the speed of 2000 km s 1 is not significantly large. Observation data after 2004 (through the end of 2009) are also examined. However, Geotail was usually located at high latitudes in the magnetotail, so that Geotail seldom observed magnetic reconnection in [9] In this paper, we study ion and electron dynamics of magnetic reconnection in the near Earth magnetotail using Geotail observations of the 15 May 2003 event. In this event, we can identify the intense electron cross tail current layer, which indicates the center of the reconnection site. MHD flows provide a speed of tailward retreat of the reconnection site, and we deduce characteristic scales for the structure of magnetic reconnection along the GSM x axis. Plasma wave observations were not made with adequate modes in this event and not discussed here. Although the ability of field and particle instruments on Geotail is limited and this is a single spacecraft observation, the present Geotail observations give unprecedented observational insight for antiparallel reconnection in the symmetrical system with no initial guide (out of plane) fields. 2. Geotail Observations 2.1. Overview of the 15 May 2003 Event [10] Geotail moved from the northern hemisphere to the southern hemisphere in the premidnight magnetotail on 15 May When Geotail approached the neutral sheet, it made in situ observations of magnetic reconnection starting at 1051 UT. A moderate substorm was recorded on the ground. The interplanetary magnetic field (IMF) B z turned southward at 1034 UT (time delay of 38 min from the ACE position is taken into consideration) and continued to be southward until 1210 UT. AL index gradually developed from the background level at 1043 UT, indicating the growth phase. A sharp drop of AL, indicating an onset of the expansion phase, was seen near 1055 UT. It is known that the ground onset timing is delayed a few minutes relative to the onset of magnetic reconnection in the premidnight tail [e.g., Nagai et al., 1998a; Ohtani et al., 1999; Angelopoulos et al., 2008]. The AL index became 300 nt at 1102 UT. The interplanetary electric field prior to the onset was approximately 2 mv m 1. The Geotail position at 1050 UT was (X GSM, Y GSM, Z GSM )=( 27.8, +3.4, +3.5 R E ). Hence, this was a normal substorm in which magnetic reconnection could be observable at the Geotail position [Nagaietal., 1998a, 2005; Nagai, 2006]. Furthermore, since this event was under moderate disturbed periods without any significant Dst development, effects of O + can be neglected for this event [cf. Kistler et al., 2006]. [11] The magnetic field and ion moment plot for UT on 15 May 2003 is available at isas.jaxa.jp/stp/cef/cef.cgi and the energy time spectrograms of ion counts and electron counts for UT are seen at Geotail traveled toward the midnight meridian near the northern tail lobe plasma sheet boundary until 1050 UT near X GSM = 28 R E. The x component of the magnetic field B x is 20 nt and there is no significant out of plane field B y. The tail lobe field is estimated to be approximately 20 nt on the basis of the total (magnetic field plus plasma) pressure balance. Geotail approached the neutral sheet and observed an onset of tailward plasma flow with southward B z at 1051 UT. The plasma flow made a reversal to earthward and continued until 1110 UT. After this, although Geotail stayed inside the plasma sheet, it did not observe any fast flow activity any more. Intense earthward flows started near 1136 UT in association with another substorm onset, and then Geotail entered the southern tail lobe. We concentrate on the time period from 1053 to 1058 UT when Geotail made in situ observations of magnetic reconnection, and we identify the time interval in which ion electron decoupling is observed. The representative Alfvén velocity for this magnetic reconnection event is estimated to be 2200 km s 1, when we adopt the magnetic field intensity of 20 nt and plasma number density of 0.04 cm 3 (the number density in the plasma sheet as discussed in 2.2) Identification of the Ion Electron Decoupling Time Interval [12] Figure 1 shows magnetic field and plasma moment data for the period from 1053 to 1058 UT on 15 May 2003 from Geotail. The magnetic field data from the magnetic field experiment MGF [Kokubun et al., 1994] are shown at 1/16 s time resolution (16 vectors s 1 ) in the GSM coordinate system. Electron flow velocity V e (ion flow velocity V i ) is obtained from moment calculations on the basis of observed velocity distribution functions from the lowenergy particle experiment LEP [Mukai et al., 1994]. One complete velocity distribution function is obtained for each 3of18

4 Figure 1. Magnetic field and plasma data from Geotail for the period from 1053 to 1058 UT on 15 May (a d) The magnetic field B x, B y, B z, and B t with time resolution of 1/16 s in the GSM coordinate system. (e g) Electron flow velocity V ex, V ey, and V ez (thick lines) and ion flow velocity V ix, V iy, and V iz (thin lines) with time resolution of 12 s. (h and i) Electron flow velocity perpendicular to the local magnetic field V e?x and V e?y (thick lines) and ion flow velocity perpendicular to the local magnetic field V i?x and V i?y (thin lines). (j) Ion number density. The time interval of the ion electron decoupling region from 1055:07 to 1056:44 UT is indicated by vertical dashed lines. 12 s plasma sampling time, so that plasma data are shown at 12 s time resolution. Possible photoelectron contaminations are excluded for V e calculations. V ex, V ey, and V ez (V ix, V iy, and V iz ) are electron (ion) velocity moment data presented in the GSM coordinate system. V e? (V i? ) is calculated as the component of V e (V i ) perpendicular to the local magnetic field (averaged over the 12 s plasma sampling time). V e?x and V e?y (V i?x and V i?y ) are presented. Each electron and 4of18

5 Figure 2. Ion and electron energy time spectrograms from LEP for the period from 1053 to 1058 UT on 15 May Tailward, duskward, earthward, and dawnward ions and electrons (counts/sample) are color coded according to the logarithmic color bar at the right hand side. The time interval of the ionelectron decoupling region from 1055:07 to 1056:44 UT is indicated by vertical red dashed lines. ion moment data point is plotted at the time of the center of its 12 s sampling period. [13] B x frequently makes zero crossing, indicating that Geotail is near the neutral sheet. High frequency variations appear for the period from 1055:10 to 1056:45 UT in all components. The high frequency variations are also seen in data from the search coil magnetometer (frequency range of Hz) on Geotail in the same interval (not shown here). During the period from 1055:10 to 1055:40 UT (the average total magnetic field is 3.3 nt), B y is almost zero when B x is near zero. This indicates that there is no significant guide field in this event. [14] Gross characteristics of ion and electron behaviors are easily recognized in ion and electron energy time spectrograms for the period from 1053 to 1058 UT in Figure 2. Ions show tailward flow for the period until 1055:44 UT and then they show earthward flow. Near this flow reversal from tailward to earthward, ions have energies higher than 2 kev, and high energy ions show duskward flow while mediumenergy ions show dawnward flow. These are typical for a flow reversal during magnetic reconnection. Electron acceleration which produces dominance of high energy (>1 kev) electron fluxes is seen in the time interval from 1054:18 to 1057:07 UT. This is a characteristic signature of magnetic 5of18

6 Figure 3. Magnetic field and spacecraft potential data for the period from 1053 to 1058 UT on 15 May (a and b) Magnetic field B t and B z with time resolution of 1/16 s. (c) Spacecraft potential (inversely plotted, indicative of electron number density) with time resolution of 3 s from EFD. (d) Ion number density from LEP. The time interval of the ion electron decoupling region from 1055:07 to 1056:44 UT is indicated by vertical dashed lines. reconnection [e.g., Nagaietal., 1998a, 2001; Shinohara et al., 1998; Asano et al., 2004a, 2008; Nakamura et al., 2006a]. In association with the flow reversal seen in ions, average energy of electrons drops at 1055:44 UT and duskward flowing electrons are largely diminished, indicating a significant dawnward motion of electrons. [15] Detailed characteristics of ion and electron behaviors are examined with flow velocities presented in Figure 1. V ix and V ex start to develop coherently after 1053 UT. Significant differences between V ix and V ex are seen after 1055 UT. Furthermore, V iy are always positive (duskward), while V ey is generally negative (dawnward). V ey becomes largely negative (dawnward) after 1055 UT. In the examined time period, ions may be decoupled to the magnetic field and electrons are probably coupled to the magnetic field. In this context, V i? may not present ion motions correctly, however, difference between V i? and V e? is a good indicator for the ion electron decoupling. [16] The flow reversal and the simultaneous enhancement of the dawnward electron flow are evident in Figures 1h and 1i. Both ions and electrons show tailward flows (see V i?x and V e?x ) with southward B z until 1055:44 UT. V e? is almost the same as V i? in the first part. This means that the plasma flow is a convection motion in MHD. V e? differs significantly from V i? in the last 36 s interval from 1055:07 UT (three plasma samplings). The magnetic field is southward (B z < 0) until 1055: UT and then it becomes northward (B z > 0) from 1055: UT. One plasma sampling (12 s) starts at 1055: UT. Ions and electrons show earthward flows (see V i?x and V e?x ) with northward B z in the period from 1055:44 to 1057:21 UT. The x component of V e? is significantly larger than that of V i? after 1055:44 UT. It is also noted that V e?x is larger than V ix in the same period. V e?x becomes the same as V i?x at 1057:09 UT, however, V e?x after 1056:45 UT is comparable to that at 1057:09 UT. Furthermore, the magnetic field behavior after 1056:45 UT is different from that before 1056:45 UT. Hence, the time interval in which V e?x is significantly higher than V i?x is identified as the first 60 s interval from 1055:44 UT (five plasma samplings). After 1056:45 UT, V e?x is almost the same as V i?x (and V ix ), indicating that the plasma flow is a convection motion in MHD, and the earthward flows continue until 1110 UT. [17] During this time interval starting at 1055:07 UT, electrons show high speed dawnward motion (V e?y < 1000 km s 1 ). This electron behavior is a significant contrast to ion behavior, since V i?y is small around zero or duskward (positive). Furthermore, the magnitude of V e?y exceeds 3000 km s 1 at 1055:31 UT and 1055:44 UT, just before and just after the turning of B z from southward to northward. Ion number density is very low (near 0.02 cm 3 ) in this time interval. Ion number density is higher than 0.04 cm 3 prior to and after this time interval. It is important to note that this lowest ion number density period is observed near the neutral sheet. During the first 36 s period of this time interval, B z is only 2.0 nt except for a short interval. This value is significantly smaller than B z value of 5.1 nt just prior to this period, when ions and electrons show the same tailward speed (V e?x is close to V i?x ). B z is 5.0 nt even when B x becomes zero near 1054:20 UT. During the last 60 s period of this time interval, B z is only +2.2 nt. Hence, the north south component of the magnetic field B z is small in this time interval. [18] In summary, the ion electron decoupling time interval is from 1055:07 to 1056:44 UT. The first 36 s (three 12 s plasma samplings) time interval corresponds to the region of tailward outflow, and the last 60 s (five 12 s plasma samplings) time interval corresponds to the region of earthward outflow Spacecraft Potential Measurements [19] Figure 3 shows spacecraft potential data for the period from 1053 to 1058 UT. Spacecraft potential measurements are made by the electric field detector EFD [Tsuruda et al., 1974]. The spacecraft potential is higher than 20 V for the time interval from 1055:11 to 1056:45 UT. This time interval corresponds to the low density interval identified with ion number density from LEP. The highest 6of18

7 Figure 4. Electron distribution functions for the 12 s sampling starting from (a) 1055:44 UT and (b) 1055:56 UT. The horizontal axis (V k ) is aligned to the average magnetic field, and the vertical axis is perpendicular to the magnetic field, such that this plane is almost parallel to the GSM x y plane. The left hand direction (V k ) is earthward, and the downward direction (V?1 ) points duskward. Electron phase space densities from to s 3 m 6 are color coded according to the logarithmic color bar. spacecraft potential is observed at 1055:47 UT, just after the B z turning. Hence, the plasma density becomes the lowest at this timing in the data period, but the plasma density seems to be higher than that in the tail lobes. The spacecraft potential exceeds 60 V in the tail lobe period after 1130 UT Electron Velocity Distribution Functions [20] Figure 4a shows electron velocity distributions at 1055:44 UT, when V ey is largely negative (dawnward) just after the northward turning of B z (see also the electron energy time spectrograms in Figure 2). The distribution is color coded and electron counts are less than sensor s onecount level in white areas. One three dimensional velocity distribution function is taken over 12 s (1055:44 UT is the start time of 12 s sampling). The average magnetic field over the 12 s sampling is (B x, B y, B z ) = (6.06, 4.13, 1.33 nt) in GSM. The magnetic field variations are less than 22 in the x y plane (see Figure 5d) and less than 15 in the x z plane during this sample. The distribution function is presented in the V k V?1 plane that includes the averaged magnetic field Figure 5. Energetic (>8.47 kev) electron angular distributions averaged over 12 s intervals for the period from 1055:07 to 1056:44 UT. Count rates (counts/sample) for 16 sectors are presented in the GSM x y plane. The 1/16 s magnetic field vectors are also plotted. The long line in each plot is the direction of the average magnetic field during one sampling. 7of18

8 Figure 6. (a) Tailward fluxes of high energy electrons (thick lines) and high energy ions (thin lines) for the period from 1053 to 1058 UT. (b) Earthward fluxes of high energy electrons (thick lines) and highenergy ions (thin lines) for the period from 1053 to 1058 UT. Unit is particles/(cm 2 s sr kev). vector and one vector in the GSM x y plane. The positive V k (left hand) direction is earthward and positive V?1 (bottom) direction is duskward. In this case, this plane is almost parallel to the GSM x y plane. Furthermore, the rotation angle from the GSM to GSE coordinates is only 3 for this interval, so that this plane is approximately perpendicular to the spacecraft spin. Since one central (equatorial) electron sensor sweeps for all azimuthal angles during the 3 s spacecraft spin [Mukai et al., 1994], observed dawn dusk asymmetry in the electron distribution is highly reliable. The dawnward excess is seen in all energies (V?1 < 0, the upper half of distribution). A bump is seen for high energy electrons in the earthward dawnward direction. It is also noted that medium energy (less than 5 kev) electrons are basically field aligned and have a significant excess in the tailward part (V k < 0, the right hand half of the distribution function). [21] Figure 4b shows electron velocity distributions at 1055:56 UT in the V k V?1 plane. This plane is also almost parallel to the GSM x y plane. Electrons show rather an isotropic flat top distribution, however, high energy (>5 kev) earthward flowing electrons are evident (V k > 0, the left hand half of the distribution). Although, the dawnward excess (V?1 < 0, the upper half of the distribution) is less evident, a slight excess in the dawnward side produces a largely negative V ey. A field aligned feature and a significant excess in the tailward part (V k < 0, the right hand half of the distribution function) are evident for medium energy (less than 5 kev) electrons. These medium energy tailward electrons, which are seen until 1056:45 UT, are Hall electrons flowing into the reconnection site, as described in many previous studies [e.g., Nagai et al., 1998a, 2001, 2003; Asano et al., 2004a]. Since Geotail is in the region of earthward outflow in the northern hemisphere (B x > 0), the outer Hall current should be earthward and the expected B y deflection is positive [e.g., Sonnerup, 1979]. Indeed, relatively large positive B y variations are observed for this period, as seen in Figure Electron Dynamics [22] Figure 5 presents changes in electron anisotropy for the period from 1055:07 to 1056:32 UT. Counts only for electrons with energies of higher than 8.47 kev are projected in the x y plane with the magnetic field vectors (at 1/16 s time resolution). Since electrons with energies of less than 5 kev are dominated by the Hall electrons, these electrons are excluded here. Electron counts until 1055:31 UT are taken in the region of tailward outflow, and high energy electrons exist mostly in the tailward half. For the period from 1055:07 and 1055:31 UT when the large V e?y is observed, excesses in the dawn side are found. Electron counts after 1055:44 UT are taken in the region of earthward outflow, and electrons exist mostly in the earthward half. Excesses in the dawn side are also seen for the period from 1055:44 to 1056:32 UT. Hence, the reversal from tailward flowing electrons to earthward flowing electrons is associated with the B z turning at 1055:44 UT, and accelerated electrons show a dawnward motion. [23] Figure 6 shows tailward and earthward fluxes of electrons for the period from 1053 to 1058 UT. Electrons in the energy range kev have counts above the background level and counts for electrons with energies of higher than kev are in the background level in LEP observations, so that the electron fluxes from the kev channel are plotted. The highest tailward electron flux is recorded at 1055:19 UT and the highest earthward electron flux is recorded at 1055:56 UT. The increase in the earthward electron flux is easily identified by comparison between the distribution at 1055:44 UT (Figure 4a) and that at 1055:56 UT (Figure 4b). 8of18

9 Figure 7. (a and b) Ion distribution functions at 1055:31 and 1055:44 UT in the V k V?1 plane. The lefthand direction (V k ) points earthward, and the downward direction (V?1 ) points duskward. (c and d) Ion distribution functions at 1055:31 and 1055:44 UT in the GSM V y V z plane. Ion phase space densities from to s 3 m 6 are color coded according to the logarithmic color bar Ion Dynamics [24] Figure 6 also shows tailward and earthward ion fluxes. Here, the ion fluxes from the highest energy channel kev of LEP are adopted. The highest tailward ion flux is recorded at 1055:07 UT, while the highest earthward ion flux is recorded at 1056:08 UT. Ions with energies higher than 40 kev are measured by the energetic particle and ion composition instrument EPIC on Geotail [Williams et al., 1994]. Tailward flux of ions with energies higher than 61.5 kev has a flux peak near 1055:07 UT and earthward flux of high energy ions has a flux peak near 1056:08 UT (time resolution of EPIC is 3 s). Hence, the LEP results indicate the behavior of higher energy ions. [25] Figures 7a and 7b show ion distribution functions at 1055:31 (B z < 0) and 1055:44 (B z > 0) UT in the V k V?1 plane containing the magnetic field vector and one vector in the GSM x y plane. The positive V k (left hand) direction is earthward and the positive V?1 (bottom) direction is duskward. Ions with energies of higher than 10 kev (speed of higher than 1400 km s 1 ) show a tailward and duskward motion at 1055:31 UT, while ions show an earthward and duskward motion at 1055:44 UT. This indicates that the ion flow direction changes in association with the B z turning. Furthermore, it is important to note that ions show neither simple field aligned motion nor convection motion. When ions have convection motion, an ion population should have the same V? velocity. Higher energy ions are located successively toward the dusk direction, far from the magnetic field, indicating duskward acceleration. During this period, ion populations are well organized in the velocity space, so that it is unlikely that different regimes are mixed during one 12 s plasma sampling. [26] Figures 7c and 7d show ion distribution functions at 1055:31 and 1055:44 UT in the GSE y z plane (almost in the GSM y z plane). This plane is chosen for showing counterstreaming structure for ions, since the plane perpendicular to the magnetic field does not include counterstreaming ions simultaneously. Wygant et al. [2005] also show counterstreaming ions near the current sheet in the GSE y z plane in the Cluster event. LEP on Geotail does not observe ions in the direction of the z axis (shaded regions in Figures 7c and 7d). The distribution at 1055:31 UT is taken close to the equatorial plane (the average B t is 3.23 nt and B z < 0), and both the northward (V z > 0) and southward 9of18

10 Figure 8. Velocities of ions from one energy channel are calculated for 1055:07 and 1056:08 UT. Number indicates the central energy (kev) of each energy channel. Velocities are presented in the GSM x y plane. The average magnetic field vectors are also plotted. (V z < 0) components of ions are almost equally heated and have a high speed (1400 km s 1 ). The distribution at 1055:44 UT is taken just above the equatorial plane (the average B t is 7.45 nt and B z > 0). The southward (V z <0) component of ions is easily found, and the northward (V z >0) component of ions exists near the z axis. The counterstreaming features are seen much easily close to the flow reversal. It is also noted that ions with energies higher than 10 kev exist in the duskward sector. [27] As seen in Figure 7, ions with different energies move to different directions in the three dimensional space. In order to clarify this ion behavior, we examine a direction of motion for ions in the certain energy range. A velocity vector of ions in a limited energy range can be calculated with ion counts from one energy channel of LEP. Using these velocity vectors, we can obtain direction of motion for ions in a specific energy range. Figure 8 shows velocity directions in the GSM x y plane for the period from 1055:07 to 1056:08 UT. The average magnetic field vector is also plotted. For ions with energies of less than 10 kev, data points are located in the area of V y < 0 near V x = 0. These ions show counterstreaming structure nearly along the z axis, as seen in Figures 7c and 7d. Although 10 kev ions have a speed of 1400 km s 1, these ions do not show any significant motion along the x axis and make motions mostly inside the y z plane near V x =0.V y < 0 means that ions are transported dawnward. It is known that inflowing ions show a dawnward motion due to the Hall electric field (toward the equatorial plane) [e.g., Nagai et al., 1998a, 2001]. [28] Although ions with energies of <10 kev have two components (the counterstreaming structure), ions with energies of >10 kev show one hot component. Data points for ions with higher than 10 kev at 1055:31 and 1055:44 UT are plotted toward the positive y direction. This is evident in the distribution function in Figures 7a and 7b (note that the distribution functions are presented in the V k V?1 plane) and high energy ions appear in the duskward part in Figures 7c and 7d. Data points for these ions before 1055:19 (after 1056:56 UT) are plotted mostly toward the negative (positive) x direction. Hence, high energy ions are accelerated mainly duskward near the center of the reconnection site, while high energy ions are accelerated mainly earthward and tailward slightly far from the center. 3. Scaling of Magnetic Reconnection on the Basis of the 15 May 2003 Event [29] It is important to get spatial scale for the structure of magnetic reconnection observed by Geotail. Here, we adopt a two dimensional picture in the meridian plane (the x z plane in GSM). We have no information in the vertical z direction. The scale in the GSM x direction can be estimated by tailward retreat speed of the reconnection site. The MHD tailward flow speed before 1055:07 UT is approximately 650 km s 1, and the MHD earthward flow speed after 1056:45 UT is approximately 450 km s 1. This difference is attributable to the tailward retreat speed of 100 km s 1. [30] Although this estimation appears to be rough, the obtained retreat speed of 100 km s 1 can be justified with various observations. Near the center of the reconnection site, inflow speed should be the same in the region of earthward flow and in the region of tailward flow for the symmetrical system. In the 15 May 2003 event, there is a significant asymmetry for the velocity along the x direction for inflowing ions (see Figure 8). At 1055:31 UT (in the region of tailward flow), V?x of low energy (<10 kev) inflow ions is approximately 269 km s 1, while at 1055:44 UT (in the region of earthward flow), V?x of inflow ions is approximately 64 km s 1. This difference can be attributed to the tailward retreat speed of 100 km s 1 of the X line. In the flow reversal events from Geotail observations [Nagai et al., 2005; see also Asano et al., 2004a], speed of tailward MHD flows does not differ significantly from that of subsequent earthward MHD flows. The 10 December 1996 event at radial distances of 26 R E studied by Nagai et al. [2001] 10 of 18

11 Figure 9. Structure of magnetic reconnection deduced from Geotail observations. (a) The north south magnetic field B z. (b) Electron flow velocity (thick line) V e?x and ion flow velocity (thin line) V i?x. (c) Electron flow velocity V e?y and ion flow velocity V i?y. The data are presented only for the 4 min interval from 1053:44 to 1057:44 UT in the reverse direction (time proceeds from right to left, indicating the geometry of magnetic reconnection). (d) The supposed two dimensional X line geometry. ECL is the intense cross tail electron current layer. (e) Positions of electron (dot) and ion (circle) flow velocity peaks. (f) Positions of electron (dot) and ion (circle) flux peaks. shows earthward flows, tailward flows, and then earthward flows, and speed of the dawnward electron motion is enhanced for each flow reversal up to 1500 km s 1. This example indicates that the reconnection site does not necessarily move straightway at high speed [see also Eastwood et al., 2005]. There are several events in which only tailward flows are observed like the 27 January 1996 event at radial distances of R E [Nagai et al., 1998a, 2001]. These results suggest that the reconnection site does not move tailward at speed of much higher than 100 km s 1 inside 30 R E. Cluster has provided tailward retreat speed for some events. Baker et al. [2002] report the speed of 100 km s 1 for the 27 August 2001 event. Imada et al. [2007] get the speed of 100 km s 1 for the 1 October 2001 event. On the basis of numerical simulations, Oka et al. [2008] show that the retreat speed is nearly 0.1 times the Alfvén velocity and that reconnection rate does not change significantly during the retreat. Hence, the retreat speed of 100 km s 1 is a reasonable value at radial distances of less than 30 R E. [31] In this discussion, we adopt the tailward retreat speed of 100 km s 1 and discuss the extent of the current layer and distances from the center of the reconnection site in the GSM x direction. One ion inertial length (1 li) is 1200 km with ion number density of 0.04 cm 3. Conveniently, the scale size covered by one plasma sampling (12 s) corresponds to 1 li. There is a possibility that Geotail can resolve the structure of magnetic reconnection in this event in which ion inertial length is relatively long (low plasma density). The Alfvén velocity is 2200 km s 1, as described earlier. Various key signatures with their scale sizes are schematically presented in Figure 9. [32] V ey shows a large value ( 6000 km s 1 ) only at 1055:44 UT, although V e?y exceeds 3000 km s 1 at 1055:33 and 1055:44 UT (Figure 1). Furthermore, the magnitude of V e?x (V ex ) at 1055:33 UT is significantly higher than that at 1055:44 UT. Low energy electrons are seen only at 1055:44 UT, and the large depletion of the duskward moving electrons is seen also only at 1055:44 UT (see the energy time spectrograms of Figure 2). The extent of the central intense current layer is estimated to be approximately 1200 km (1 li) or less than 1 li. It is reasonable to locate this current layer at the center of the reconnection site. The current density is approximately 10 na m 2 when we use 11 of 18

12 electron flow speed of 3000 km s 1 and number density of 0.02 cm 3. [33] The dawnward electron current layer is seen for the 96 s period from 1055:07 to 1055:45 UT, so that its extent is estimated to be 10,000 km. This is approximately 8 li(1.5r E ). For this 96 s period, electron outflow velocity exceeds ion outflow velocity, and the normal component of the magnetic field B z is small (magnitude of 2 nt). Hence, the ion electron decoupling region extends over the scale of 8 li. Nagai et al. [2005] identify 209 tailward flow events (with negative B z ) in association with substorm onsets mostly at radial distances of R E (Y GSM = 5 to+15r E )in Only 34 events are associated with highly accelerated electrons, which are a good indicator for the ion electron decoupling region [e.g., Nagai et al., 1998a, 2001]. If the ion electron decoupling region extends over 10 R E (50 li), spacecraft could observe almost always some signatures of the ion electron decoupling at X GSM = 20 to 30 R E. The occurrence probability of 16% (34 out of 209) suggests the length of the ion electron decoupling region of 1.6 R E in the x direction. [34] Inside the ion electron decoupling region, electron outflow speed reaches its highest speed at distances of 1 2 li from the center. Ion outflow speed reaches its highest speed at distances of 3 li from the center, further away from the center. The electron outflow speed is reduced and becomes equal to the ion outflow velocity at distances of 4 5 li from the center. Electrons can be efficiently accelerated near the center. The highest fluxes of tailward and earthward high energy electrons appear at distances of 2 li from the center. Probably, this is caused by acceleration of the reconnection electric field (the positive y direction), since high energy electrons have large dawnward velocity (Figure 5). There is a possibility that part of electron current is caused by upstream electron pressure anisotropy, as suggested by Le et al. [2010]. Energies of these electrons are only up to 13 kev, although LEP does not have good capability of observing suprathermal electrons. There is no clear signature for magnetic islands inside the ion electron decoupling region of this event. [35] Outside the ion electron decoupling region, the structure becomes MHD. Just outside of the tailward region of the ion electron decoupling region, the magnetic field B z is 5 nt in the region of tailward outflow. The magnitude of B z is 2 nt inside the ion electron decoupling region. Fluxes are piled up in the MHD region. As seen in Figure 1, there are irregular variations in the magnetic field, a positive spike in negative B z near 1055:07 UT and a negative spike in positive B z near 1055:45 UT, just near the boundary of the ion electron decoupling region. Hence, the boundary may not be a simple transition layer. 4. Comparison to Simulation Results [36] It is difficult to judge whether observations by a single spacecraft cover various key ingredients of magnetic reconnection in right places. Results from a number of twodimensional particle in cell simulations for magnetic reconnection have been reported and a widely accepted view for nature of magnetic reconnection has emerged [e.g., Birn and Priest, 2007]. Since it is practically difficult to evaluate the present observational results by Geotail consistently with the published results, we make here a 2 D particle in cell simulation of magnetic reconnection with specific parameters which are taken to simulate the observed reconnection event as possible. The present simulation is carried out basically in the same manner with our previous study [Nagai et al., 2003], and the results are essentially consistent with the previous results. [37] The initial magnetic field is given by the Harris sheet B x (z) =B 0 tanh(z/d), where B 0 is the asymptotic magnetic field and D is the current sheet half thickness. A plasma inside the current sheet has number density n CS and ion to electron temperature ratio T i,cs /T e,cs = 5. In addition to the current sheet component, a uniform background plasma with number density of n BK = n CS is distributed. Temperatures for the background plasma are T i,bk = T e,bk = T e,cs.in order to trigger magnetic reconnection quickly, a magnetic island shaped perturbation is added inside the Harris current sheet. The perturbed magnetic flux function is given by y(x, z) = y 0 sin(2px/l x ) cos(2pz/l z ). This makes the magnetic field perturbation ~B(x, z) =^e y ry(x, z) tobe superposed on the Harris equilibrium. The simulation box size is [ L x /2, +L x /2] [ L z /2, +L z /2] where L x =48Dand L z =24D. Periodic boundary conditions are imposed in the x direction while conducting walls are set at the z boundaries. In the present simulation, following parameters are used; ion to electron mass ratio m i /m e = 400, frequency ratio w pe /W e =4, pinitial ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi current sheet thickness D = 0.5 l i, where w pe pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4n CS e 2 =m e, W e eb 0 /m e c, and l i c/w pi = c/ 4n CS e 2 =m i are the electron plasma frequency, the electron gyro frequency, and the ion inertia length, respectively. The spatial grid size D is equal to the Debye length of the background plasma, and it corresponds to l i = 200 D. The number of simulation grids is and particles for each species are loaded into the simulation domain, so that the spatial resolution of the magnetic diffusion region is quite higher than previous studies. After the trigger of magnetic reconnection, reconnection outflow increases up to V ix 0.3 V A at ptw ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi i = 35 where V A is the Alfvén speed defined by B 0 / 4m i n CS. Because of the periodic boundary condition in the x direction, this timing is the peak of magnetic reconnection (the final time when the boundary conditions do not affect evolution of magnetic reconnection). Hereafter, we discuss structure of magnetic reconnection based on the snapshot taken at TW i = 35. [38] Figure 10a shows the out of plane magnetic field component B y in the upper left quadrant of the simulation domain (corresponding to the northern earthward side of the reconnection site in the observations). An X line exists at the bottom right corner (x, z) = (0, 0). An intense positive B y deflection (colored red) is a part of the wellknown Hall quadruple magnetic structure of the magnetic field. Figures 10b 10d show electron and ion flow velocities and the magnetic field along the x axis in the outflow region. V ex has a peak of 2.5 V A at x 0.8 l i (Figure 10b), and it rapidly drops down to 0.5 V A at x 1.6 l i and gradually decreases to 0.2 V A. In contrast to the electron outflow, V ix dose not show any clear peak and has a broad maximum of 0.3 V A near x =4l i in the piled up magnetic field region. The ion and electron outflow speeds become almost the same in further downstream region x >4l i. Hence, the full extent of the ion electron decoupling region in the x direction is 8 l i in this simulation. Figure 10c 12 of 18

13 Figure 10. Structure of magnetic reconnection in the simulation. (a) The Hall magnetic field B y in the northern earthward quadrant of the simulation box, (b) V ex (blue) and V ix (red) along the +x axis at z =0, (c) V ey and V iy along the +x axis, (d) B z along the +x axis, (e) B x along the +z axis at x = 0 and, (f) V ey and V iy along the +z axis. shows the existence of the intense electron current layer near the X line. The outer edge of the intense electron current layer is near the V ex peak at x 0.8 l i, and the electron current layer with the lower V ey extends in both sides up to x 4 l i. The total magnetic field B t is exactly the same as the magnitude of B z on the x axis and the peak of the total magnetic field is seen at x =4l i (Figure 10d). The peak corresponds to the piled up field due to reconnection outflows. The full extent of the minimum B t region is less than 1 l i, and B t is only 10% of the asymptotic magnetic field even at x =1l i. It is important to note that the full extent of the intense electron current layer becomes approximately 1.6 l i even at TW i = 29 and remains to be constant until TW i = 35 in this simulation. [39] Key characteristics of the electron and ion flows in the simulation can be found in the observations. The observed V ex has a sharp peak and maintains a high speed in the region of earthward outflow (see Figure 1). The ratio of the maximum outflow speed V ix /V ex is 0.1 in the simulation, while the observed ratio is The peak V ey reaches 3.5 V A, which is higher than the peak V ex. The ratio of V ey /V ex is 1.4 in the simulation and 1.25 in the observation. Hence, speed of the electron flow carrying the crosstail current in the center of the reconnection site is higher than the peak outflow speed. The full extent of the super Alfvénic V ey flow region is 1.6 l i, and the lower V ey flow region extends in both sides. In the observations, the full extent of the high speed V ey region is approximately 1 l i, 13 of 18

14 Figure 11. Selected electron distribution functions in the V x V y plane (a) at (x, z) = (0.8, 0.2), (b) at (x, z) = (0.4, 0.2), (c) at (x, z) = (0.0, 0.2), (d) at (x, z) = (0.8, 0.0), (e) at (x, z) = (0.4, 0.0), and (f) at (x, z) = (0.0, 0.0) in the simulation. The bottom right distribution (f) corresponds to that at the X line. The thick line indicates a velocity moment vector, and the thin line indicates a magnetic field vector in each plot, see the text. and the electron current layer is identified in the full extent of 8 l i. Hence, the observed plasma flow structure and the spatial scales of the electron current layer are fairly consistent with the simulation results. [40] Figures 10e and 10f show electron and ion flow velocities and the magnetic field along the z axis in the inflow region at x = 0. The total magnetic field becomes 40% of the asymptotic magnetic field at z 0.2 l i. The observed magnetic field of 2.5 nt near 1055 UT is probably 10% of the tail lobe field, indicating that Geotail is very close to the neutral sheet. Even the observed magnetic field of 6 nt near 1056 UT indicates that Geotail is near the neutral sheet. The Hall magnetic field deflection B y starts near (x, z) = (0.6, 0.1) in the simulation (see Figure 10a). It is reasonable that Geotail detected the positive Hall B y field near 1056 UT in the region of earthward outflow. The intense electron current layer with electron flow velocity higher than V A is confined into z < 0.1 l i in the simulation (Figure 10f). Hence, it is reasonable that spacecraft can rarely encounter the intense electron current layer. [41] Figure 11 shows selected electron distribution functions in the V x V y plane from the simulation. In each panel, the number of electrons is displayed by contour map. The direction of the magnetic field is plotted by the thin line and a vector of the velocity moment is presented by thick line. Figure 11f indicates the position at the X line. At z =0 (Figures 11d 11f), the dawnward velocity vectors are evident, while the magnetic field vectors almost vanish since B t is less than 0.1 B 0.Atz = 0.2 (Figures 11a 11c), the positive (earthward) magnetic field vectors are evident, while the velocity vectors are small. The electron population in each position can be characterized by an area colored green to red. The electron population shows an evident shift of its center to the y direction only at the X line (Figure 11f). However, along the x axis (Figures 11d 11f), the electron population shows a bump toward the y direction, which makes a dawnward electron flow. The dawnward bump successively moves toward +x direction (Figures 11f, 11e, and 11d), making an earthward outflow. At z = 0.2 (Figures 11a 11c), the electron population is elongated along the magnetic field direction, which is well known bidirectional streaming features in the reconnection site. As seen in Figure 10a, the magnetic field vector shows a significant positive Hall field B y at (x, z) = (0.8, 0.2). The electron population colored red corresponds to mediumenergy electrons. These electrons show an excess in the V x part, which corresponds to the Hall electrons flowing into the reconnection site. Hence, characteristics of electron distributions in the observations are essentially the same as those near the X line in the simulation. [42] Figure 12 shows selected ion distribution functions along the x axis in the V x V y plane and those in the V y V z 14 of 18

15 Figure 12. Selected ion distribution functions in the V x V y plane in the equatorial plane (a) at (x, z) = (0.8, 0.0), (b) at (x, z) = (0.4, 0.0), and (c) at (0.0, 0.0) in the simulation. Selected ion distribution functions in the V y V z plane in the equatorial plane (d) at (x, z) = (0.8, 0.0), (e) at (x, z) = (0.4, 0.0), and (f) at (x, z)= (0.0, 0.0) in the simulation. plane. The ion population can be characterized by an area colored green to red. It is clearly seen that the center of the ion population is shifted from zero in the V x V y plane (Figures 12a 12c). The center of the ion population is shifted toward the positive y direction, and a tip forms in the +y direction. The tip direction moves toward the +x direction, making an earthward outflow. This is seen in the observations as presented in Figure 8. It is important to note that these high energy ions are not field aligned, since there is only weak magnetic field component along the x axis in the simulation. The reconnection electric field accelerates these demagnetized ions toward the duskward direction in the reconnection site. In the view of the single particle motion, these ions are picked up by the magnetic field lines carried by accelerated outflowing electrons and escape from the acceleration site [e.g., Cowley, 1985]. In the V y V z plane (Figures 12d 12f), counterstreaming structure (two peaks colored red) is evident in the area of V y < 0, indicating that these inflowing ions have a dawnward motion. Higherenergy ions are seen in the +V y part. These are essentially the same as seen in the observations in Figure 7. Hence, the observed ion distribution functions are also consistent with the signatures of the ion distribution functions near the X line. [43] On the basis of the comparisons discussed above, it is concluded that the present observations by Geotail are made in the close vicinity of the X line. The scale sizes for the observed structure are taken in adopting the two dimensional picture in section 3. This procedure is justified by good agreement between the observation results and the simulation ones, however, the three dimension structure of magnetic reconnection in the near Earth magnetotail should be studied observationally. [44] The spatial extent of the ion electron decoupling region in the x direction is only 8 l i in the Geotail observations. The recent studies of magnetic reconnection with the open boundary conditions [e.g., Daughton et al., 2006] show that the full extent of the electron current layer is elongated much longer than that of earlier simulations. The present simulation has a high ion to electron mass ratio (m i /m e = 400), but the simulation box is small with the periodic boundary conditions in the x direction. Although this simulation belongs to the earlier simulations, the relatively short extent of the central intense electron current layer is almost time stationary even when there are no effects of the adopted boundary conditions. [45] Shay et al. [2007] and Karimabadi et al. [2007] propose the multiple scale structure of electron diffusion region. The inner region is the intense out of plane electron current layer where V ey highly exceeds V A. This layer can have a half length of several l i. However, its length is probably shorter for realistic ion electron mass ratio. Shay et al. [2007] predict that the full extent of the inner region is 0.6 l i for the real mass ratio. Hence, the full extent of the 15 of 18