Current sheet structure and kinetic properties of plasma flows during a near-earth magnetic reconnection under the presence of a guide field

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH: SPACE PHYSICS, VOL. 118, , doi: /jgra.50310, 2013 Current sheet structure and kinetic properties of plasma flows during a near-earth magnetic reconnection under the presence of a guide field E. E. Grigorenko, 1 H. V. Malova, 2 A. V. Artemyev, 1 O. V. Mingalev, 3 E. A. Kronberg, 4 R. Koleva, 5 P. W. Daly, 4 J. B. Cao, 6 J.-A. Sauvaud, 7 C. J. Owen, 8 and L. M. Zelenyi 1 Received 28 November 2012; revised 23 April 2013; accepted 2 May 2013; published 21 June [1] Fortunate positioning of Cluster and TC-1 in the plasma sheet (PS) of the Earth s magnetotail has allowed studies of the current sheet (CS) structure and particle dynamics in mesoscale and microscale in both sides of the near-earth reconnection, which took place between 03:42 and 03:55 UT on 22 September The distinctive feature of this event was the presence of a strong negative B Y field forming a bell-like spatial profile with the maximum absolute value near the neutral plane. The magnitude of this B Y field was almost two times larger than the interplanetary magnetic field (IMF) and therefore could not be explained solely by the IMF penetration into the magnetotail. We propose a possible intrinsic mechanism of the B Y field enhancement near the neutral plane based on peculiarities of the nonadiabatic ion interaction with the thin CS. An analysis of test particle trajectories shows that in the presence of a guide field with the bell-like spatial profile, a pronounced north-south asymmetry appears in the refraction/reflection properties of nonadiabatic ions from the CS. In a region tailward of the reconnection (B Z < 0), this asymmetry results in an increase of the density of the kev ions ejected into the northern PS and moving tailward. These ions can carry the tailward current which may be responsible for the strong negative B Y near the neutral plane, i.e., self-consistent enhancement of a B Y field could occur near the neutral plane. Citation: Grigorenko, E. E., et al. (2013), Current sheet structure and kinetic properties of plasma flows during a near-earth magnetic reconnection under the presence of a guide field, J. Geophys. Res. Space Physics, 118, , doi: /jgra Introduction [2] The magnetotail current sheet (CS), which separates the northern and southern lobes, plays a key role in the magnetosphere dynamics [e.g., Baker et al., 1996; Sergeev et al., 2012]. Various processes of energy dissipation and 1 Space Research Institute, Russian Academy of Sciences, Moscow, Russia. 2 Skobeltsyn Institute of Nuclear Physics, Lomonosov Moscow State University, Moscow, Russia. 3 Polar Geophysical Institute, Kola Scientific Center, Russian Academy of Sciences, Apatity, Russia. 4 Max Planck Institute for Solar System Research, Katlenburg-Lindau, Germany. 5 Space Research and Technologies Institute, Bulgarian Academy of Sciences, Sofia, Bulgaria. 6 Space Science Institute, School of Astronautics, Beihang University, Beijing, China. 7 Institute de Recherche en Astrophysique et Planétologie, Toulouse, France. 8 Mullard Space Science Laboratory, University College London, London, UK. Corresponding author: E. E. Grigorenko, Space Research Institute, Russian Academy of Sciences, Profsoyuznaya str., 84/32, Moscow , Russia. (elenagrigorenko2003@yahoo.com) American Geophysical Union. All Rights Reserved /13/ /jgra conversion, occurring in the CS at different spatial and temporal scales, may affect the large-scale plasma and energy transport and result in transformation of the magnetotail configuration [e.g., Baumjohann et al., 2007]. Numerous in situ observations performed by Geotail, Cluster, and THEMIS spacecraft have reported different features of magnetic field topology in the magnetotail CS such as flapping, flattening, tilting, waving, twisting, and bifurcation [e.g., Asano et al., 2004; Zhang et al., 2002, 2005; Sergeev et al., 2003; Runov et al., 2003a, 2005; Alexeev et al., 2005; Nakamura et al., 2006; Shen et al., 2008a, 2008b; Petrukovich 2011; Rong et al., 2011]. Usually, the magnetic field lines, crossing the magnetotail equatorial plane, have a normal component B Z due to the Earth s dipole field and a smaller component along the dawn-dusk direction, B Y. Sometimes, however, the B Y component in the magnetotail CS becomes significant [e.g., Tsurutani et al., 1984; ieroset et al., 2001; Nakamura et al., 2006, 2008; Shen et al., 2008a; Rong et al., 2012; Wang et al., 2012]. A strong B Y field can shear the magnetic field of the CS, making the CS thin and causing a cross-tail field-aligned current [e.g., Nakamura et al., 2008; Shen et al., 2008a, Wang et al., 2012], influencing the adiabaticity and orbits of charged particles in the CS [e.g., Büchner and Zelenyi, 1991; Zhu and Parks, 1993; Kaufmann et al., 1994; Shen et al., 2008a; Malova et al., 2012]. A strong B Y field also might be important for the reconnection process [e.g., Ricci et al., 2004; Pritchett and Coroniti, 2004; 3265

2 Figure 1. Sketch of a possible magnetic topology and accelerated plasma flows simultaneously observed by TC-1 and Cluster spacecraft in the near-earth tail. A magenta line shows a schematic representation of the Cluster barycenter trajectory. Mozer et al., 2008; Pritchett and Mozer, 2009]. Thus, it is important to know the spatial distribution of the B Y field across the CS as well as the characteristic scale of the CS region where this component dominates. This knowledge is tightly connected with the understanding of mechanisms, which are responsible for the strong B Y field generation in the magnetotail CS. [3] The B Y component in the magnetotail was suggested to have a uniform distribution across the CS in many previous studies [e.g., Büchner et al., 1991; Zhu and Parks, 1993; Pritchett and Mozer, 2009]. Such a B Y field may exist in the magnetotail due to the direct penetration of the interplanetary magnetic field (IMF). Indeed, many previous observations reported a good correlation of the CS B Y with the IMF B Y [e.g., Fairfield, 1979; Lui, 1984; Sergeev 1987; Kaymaz et al., 1994, Shen et al., 2008a]. Nakamura et al. [2008] also reported a good correlation of the CS B Y and the IMF B Y but pointed out that the pronounced variations of the CS B Y may be caused by the internal CS dynamics. [4] Rong et al. [2012] studied the spatial profiles of the strong B Y field in the magnetotail CS on the basis of Cluster observations. They statistically demonstrated that around the midnight meridian, the strength of the B Y component is usually enhanced at the center of the CS relative to that in the CS boundaries, so that the spatial profiles of the B Y component along the north-south direction have a belllike shape. [5] However, the origin of a strong B Y field in the magnetotail CS is still unclear. Petrukovich [2011] demonstrated that the origin of a strong B Y in the magnetotail could be caused by internal processes taking place especially during substorms. Rong et al. [2012] also showed that in some cases, a B Y field in the CS was stronger or even had a sign opposite to the sign of the IMF B Y. The authors suggested that internal CS processes could be responsible for the observed B Y variations across the CS. [6] The understanding of the internal CS dynamics under the presence of a B Y field is therefore an important subject for study. In a thin CS with a thickness comparable to the ion gyroradius, ions become nonadiabatic and experience a chaotic motion, while electrons are still magnetized. This results in the decoupled motion of ions and electrons [e.g., Nagai et al., 2001, 2003; Asano et al., 2004, Aunai et al., 2011] and a specific current density distribution [Nakamura et al., 2006; Henderson et al., 2006]. The presence of a strong B Y field influences the nonadiabatic particle dynamics. Malova et al. [2012] studied the self-consistent configuration of the thin CS under the presence of a global (almost constant) B Y field in the magnetotail without spatial variation of B Y across the CS. The authors showed that the presence of such B Y field influences the particle scattering in the thin CS and leads to the appearance of an asymmetric CS. The existence of the asymmetry in the current density distribution in the thin CS under the presence of a guide B Y field was also confirmed by Cluster observations earthward of the reconnection region by Wang et al. [2012]. [7] In this paper we report the Cluster observations of a strong B Y field with the bell-like spatial distribution across a thin CS during an interval of energy dissipation in the near-earth magnetotail (possibly magnetic reconnection), which occurred between 03:42 and 03:55 UT on 22 September In contrast to the previous studies, these observations were made tailward of a near-earth reconnection region. On the basis of multipoint observations, we study the CS structure and kinetic features of particle dynamics and show the existence of a closed plasmoid-like magnetic configuration tailward of a possible X line (Figure 1). We reveal the presence of a nonadiabatic population of kev ions and demonstrate that this population contributed to the negative parallel current observed near the neutral plane. On the basis of the kinetic analysis of test particle trajectories in the magnetic configuration with a bell-shaped B Y field observed tailward of the reconnection region, we demonstrated a north-south asymmetry in the refraction/reflection properties of nonadiabatic ions from the CS. This peculiarity of nonadiabatic ion dynamics may be responsible for the production of ion current, which, in turn, could form a closed current loop in the PS tailward of the reconnection region and contribute to the significant enhancement of the B Y field near the neutral plane. 2. An Overview of the Event [8] Figure 2 presents an overview plot of the interplanetary media, geomagnetic conditions, and magnetotail state at TC- 1 and Cluster locations between 03 and 04 UT on 22 September IMF and solar wind (SW) parameters were measured by the WIND spacecraft at [21, 88, 23] R E (GSM) and were consistent with the corresponding shifted profiles obtained by ACE and SOHO in the upstream SW (not shown). Geomagnetic indices (AL) were retrieved from the World Data Center for Geomagnetism in Kyoto ( wdc.kugi.kyoto-u.ac.jp). [9] During this 1 h interval, solar wind pressure was low and stable. The Z component of the IMF was mostly northward. A negative value of B Y was observed during the entire interval; however, its value was almost two times smaller than the value of the negative B Y variations detected in the magnetotail plasma sheet (PS) between 03:00 and 03:10 UT (not analyzed in this paper) and between 03:42 and 03:55 UT (the interval of interest). This indicates that the strong negative B Y could not be caused solely by the penetration of the IMF B Y into the magnetosphere, but represents an own mode generated by the internal CS dynamics. 3266

3 [11] Between 03:00 and 04:00 UT, the TC-1 spacecraft was located in the Southern Hemisphere and moved tailward. The positions of TC-1 and the barycenter of the Cluster tetrahedron for this period are shown in the top row of Figure 3. Unless noted otherwise, we use GSM coordinates in the subsequent discussions. The separation between TC-1 and Cluster in the dawn-dusk direction was less than 1 R E during the entire interval. This ensures that the phenomena observed at both locations have a common source. [12] Between 03:42 and 03:55 UT, TC-1 observed a burst of earthward moving plasma flow together with two pulses of positive B Z. Within this interval, Cluster-1 and Cluster-3 (not shown) registered a burst of a tailward moving plasma flow along with a negative B Z excursion. This feature may indicate the activation of an acceleration source (possible X line) between Cluster and TC-1 locations, at 18 R E < X < 10 R E. [13] During this event, TC-1 stayed in the southern PS and did not approach close to the neutral sheet. The Cluster spacecraft crossed the neutral plane several times. The details of these crossings and the observed CS structure are described in the next section. Figure 2. An overview plot of the interplanetary media, geomagnetic conditions and magnetotail state at TC-1 and Cluster spacecraft locations between 03 and 04 UT on 22 September (first to third panels) Solar wind pressure and three GSM components of the IMF measured by WIND spacecraft and the time profile of AL index; (fourth to sixth panels) three GSM components of the magnetic field, ion density, and three components of ion velocity observed by TC-1; and (seventh to ninth panels) three components of the magnetic field, ion density and velocity observed by Cluster-1. [10] This period was geomagnetically very quiet. The absolute value of the AL index was less than 50 nt during the entire interval. No geomagnetic perturbations were detected by ground stations located in the vicinity of the TC-1 projection into the ionosphere (not shown). 3. Detailed Analysis of Cluster Observations Tailward of the Reconnection Region [14] In this section we analyze the magnetic field data, the ion velocity distribution functions, and the electron pitch angle distributions observed by Cluster tailward from the possible reconnection region. Basically the Cluster spacecraft stayed near the neutral plane during ~7 min allowing a timing of the CS dynamics and the plasma acceleration process. [15] Before analyzing the observed magnetic structure, we estimated and subtracted the background B Y field originating from the tail flaring effect [e.g., Fairfield, 1979; Petrukovich, 2011; Rong et al., 2012]. For Cluster data, the flaring-related component of the B Y field was estimated according to the method used by Rong et al. [2012]. We obtain a flaring component of approximately 0.2B X. This value is also in agreement with the value estimated in the statistical study by Petrukovich [2009] for Y ~4 R E. Since Cluster and TC-1 had close Y locations, we used the same estimation for TC- 1 data. It is worth noting that since both spacecraft were located in the premidnight sector, the subtraction of the flaring-related component does not significantly influence our results Magnetic Topology and CS Structure [16] Between 03:42 and 03:55 UT, the Cluster spacecraft stayed in the central plasma sheet (CPS), forming an almost regular tetrahedron with the largest interspacecraft distance of ~1250 km. The Cluster tetrahedron configuration averaged for this interval is shown in the bottom row of Figure 3. Figure 4 presents a summary plot of Cluster observations during this interval: the three components of the ion velocity measured by the Hot Ion Analyzers (HIA) [Réme et al., 2001] onboard Cluster-1 and Cluster-3, the three components of the magnetic field measured by the fluxgate magnetometer (FGM) [Balogh et al., 2001], the three components of the electric current density, the perpendicular and parallel components of the current density at Cluster barycenter, the estimated error on the electric current density (rb)/(rb) 3267

4 Figure 3. (top row) TC-1 and Cluster tetrahedron locations between 03:42 and 03:55 UT and (bottom row) locations of the four Cluster spacecraft in (XY) GSM,(XZ) GSM, and (YZ) GSM planes. [Dunlop et al., 1988, 2002], and the three components of the magnetic curvature vector curv(b)=(br)b measured at the Cluster barycenter [Runov et al., 2003b]. [17] At the beginning of the interval of interest, the Cluster spacecraft were located in the southern part of the outer PS (B X ~ 15 nt at the barycenter) and the southernmost Cluster-3 detected the smallest B X and the largest B. Then the Cluster quartet started to approach the neutral plane. Simultaneously with the B X reduction to zero value, the negative value of B Y increased. Near the neutral plane in the region between B X = 5 and 0 nt, the negative value of B Y reached 9 nt. [18] The B Z component experienced a sharp positive jump up to 4 nt around 03:44:50 UT registered first by Cluster-1, then by Cluster-2 and Cluster-3, and finally by Cluster-4. This positive B Z perturbation propagated tailward with a V X velocity of approximately 190 km/s, which was determined from the timing analysis, applied to four Cluster spacecraft data (Harvey, 1998). Simultaneously with the B Z jump, the X component of the curv(b) experienced also a positive variation, which indicates the propagation of the tailward edge of a plasmoid by the Cluster spacecraft [e.g., Hones et al., 1984]. After that, the Z component of the magnetic field became negative according to the observations of all Cluster satellites and experienced strong temporal fluctuations db Z /<B Z > ~ 2.0. [19] During the entire interval of interest, the correlation between the magnetic field variations registered by each pair of Cluster spacecraft was high (correlation coefficients 0.76 < Rcc < 0.98), so that we may assume that all spacecraft stayed in the same physical region and the linear gradient and rb estimator technique can be applied [Chanteur, 1998]. For further analysis of the magnetic and current sheet structure, as well as the particle dynamics observed by the Cluster spacecraft during the interval of interest (03:42 03:55 UT), we divide this interval into three subintervals according to the shape of the B Y spatial profiles shown in Figure 5 (left column). [20] During subinterval I (between 03:42 and 03:46:40 UT), the B Y component of the magnetic field had a bell-like spatial profile with a strong negative value detected near the neutral plane (see Figure 5, left column). Outside the neutral plane, the absolute value of B Y decreased and reached zero at the region with B X ~ 15 nt. During this subinterval, the Y component of electric current, J Y, increased up to 10 na/ m 2, but since the B Y field dominated in the CPS and near the neutral plane, this current was mainly field aligned. Indeed, as can be seen from Figure 4, the perpendicular current at the Cluster barycenter was significantly smaller ( 5 na/m 2 ) than the parallel current. [21] Figure 6 shows the distributions of the Y component of the electric current density (Figure 6a) and the parallel current density, J (Figure 6b), calculated at the Cluster barycenter, versus the value of B X. It is seen that the absolute value of J gradually increases toward the neutral plane and has the maximum at the region with B X ~ 3 nt, i.e., just outside the neutral plane. The observed increase of the negative value of J corresponds to the increase of the duskward current. [22] From 03:44:30 UT until the end of subinterval I, the two pairs of the Cluster quartet (Cluster-1, Cluster-4 and Cluster-2, Cluster-3) were located at opposite sides of the CPS (see Figure 4) and Cluster barycenter was inside the current layer. For this period, the CS thickness could be estimated as L = B 0 ΔZ 1-3 /ΔB X(1-3) ~ 0.4 R E, where B 0 is the magnetic field in the outer PS outside the current layer (B 0 ~ 15 nt); ΔZ 1-3 is the difference in the Z locations of Cluster-1 and Cluster-3 and ΔB X(1-3) is the difference in the B X component of the magnetic field measured by both satellites. It is also worth noting that after 03:45 UT, Cluster-1 and Cluster-4 started to observe similar values of the magnetic field. This indicates the increase of the CS tilt. For the period between 03:44:30 and 03:46:40 UT, when the Cluster 3268

5 Figure 4. Details of ion velocity, magnetic field, and current density observed by Cluster s/c between 03:42 and 03:55 UT on 22 September (first and second panels) Three GSM components of the ion velocity observed by Cluster-3 and Cluster-1, (third to sixth panels) absolute value of the magnetic field and three components of the magnetic field measured by the four Cluster spacecraft, (seventh panel) three components of the electric current density, (eighth panel) time profiles of the parallel (J ) and perpendicular (J ) current density, (ninth panel) rb/rb, and (tenth panel) three GSM components of the magnetic field curvature vector measured at the Cluster barycenter. The periods of quadrupole B Y variations are shaded in yellow. The three subintervals of characteristic B Y spatial distributions are shown by vertical dashed lines and marked as I, II, and III. tetrahedron was located inside the current layer ( B X 5 nt), the local coordinate system l, m, n was defined according to the method discussed by Runov et al. [2006]: the l axis is directed along the maximum variance eigenvector (from Maximum Variance Analysis (MVA), applied to the magnetic field at the barycenter); the m axis is aligned along the component of J perpendicular to l, averaged over this period; and the n axis is directed perpendicular to l and m. As a result we obtain l =[ 0.7, 0.6, 0.4], m = [0.6, 0.75, 0.3] and n =[ 0.3, 0.3, 0.9]. Thus, the tilt angles in the (XZ) and(yz) planes:c =atan( n X / n Z ) 18 and =atan ( n Y / n Z ) 18. [23] During subinterval II (between 03:46:40 and 03:49:30 UT), the spatial distribution of the B Y component changed drastically. In the beginning of this subinterval within 03:46:40 03:47 UT, Cluster-1 and Cluster-4 registered still large and negative values of the B Y field at the neutral plane (see Figure 4). But, in comparison with the previous subinterval, the B Y field decreased more sharply outside the neutral plane (see Cluster-2 and Cluster-3 observations in Figure 4). [24] By the middle of this subinterval, the guide field completely disappeared and during three short periods, 03:47:12 03:47:31 UT, 03:47:44 03:48:13 UT, and 03:48:30 03:48:50 UT, quadrupole B Y variations were detected (these periods are shaded by yellow color in Figure 4). During these periods, Cluster spacecraft simultaneously registered a positive B Y variation in the southern CPS and a negative B Y variation in the northern CPS. This spatial distribution of the B Y field can be associated with the tailward Hall current outflowing from a magnetic X line located earthward of the Cluster spacecraft [e.g., Nagai et al., 2001; Runov et al., 2003b]. [25] Around the second period of the quadrupole B Y variation, the absolute value of the magnetic field B decreased down to ~2 nt at the neutral plane (see Cluster-1 and Cluster-4 observations in Figure 4) and it decreased further down to ~0.5 nt by 03:48:30 UT (see Cluster-2 and Cluster-3 observations in Figure 4). This means that the Cluster spacecraft approached closer to the reconnection region. The B Z component of the magnetic field was negative at the neutral plane and experienced strong temporal variations (see Figure 5, right column). [26] By 03:48 UT, the CS thickness decreased down to ~800 km [Artemyev et al., 2008] and the CS tilt in the (YZ) plane increases up to ~40. Since during this time the Cluster tetrahedron was inside the current layer, the tilt angle in the (YZ) plane could be estimated as J Z /J Y. At this time, the Cluster barycenter approached the neutral plane while a strong negative variation of the X component of curv(b) was observed along with the bipolar variations of Y and Z components of curv(b). As opposed to the event of X line tailward retreat, which was analyzed by Runov et al. [2003b], in this case, the amplitude of the variation of the X component of curv(b) was comparable to the amplitudes of the variations of curv(b) Y and curv(b) Z. It is also worth noting that during the periods, when the Cluster barycenter was located in the southern CPS, positive variations of the Z component of curv(b) were detected, while in the northern CPS, negative variations of curv(b) Z were observed. This means that the magnetic loop(s) was closed in the tailward direction. [27] In the beginning of subinterval II, when a guide B Y field was still rather strong near the neutral plane, the parallel component of the electric current density exceeded the value of the perpendicular component and had a maximum in the region with B X ~ 4 nt. During the periods of the quadrupole B Y variations, J ~J and the location of the maximum value of J approached closer to the neutral plane (see Figure 6b). By the end of subinterval II when the quadrupole structure of B Y started to decay and the guide B Y field appeared again, the value of J again became larger than J and the CS tilt angle decreased. Indeed the components of the local orthogonal coordinate system l, m, n, calculated 3269

6 Figure 5. (left column) Spatial distributions of the B Y field and (right column) spatial distributions of the B Z component observed by the four Cluster spacecraft during the three subintervals marked in Figure 4 as I, II, and III. The magnitude of B X was used as a proxy for the distance from the neutral plane. for the period between 03:48:15 and 03:49:30 UT were l = [0.9, 0.3, 0.3], m = [0.2, 0.95, 0.2], and n =[ 0.3, 0.15, 0.95]. So that the tilt angle in the (YZ) plane decreased to ~9. [28] During the last subinterval III (between 03:49:30 and 03:55 UT), the bell-like spatial distribution of the guide field was observed again with B Y ~ 7 nt at the neutral plane (see Figure 5, left column). The CS thickness L estimated for the period of the neutral plane crossing between 03:49:30 and 03:50 UT increased to ~0.2 R E. The components of the l, m, n coordinate system calculated for the period between 03:49:30 and 03:49:55 UT were l = [0.9, 0.35, 0.3], m =[ 0.4, 0.9, 0.15], and n =[ 0.3, 0.15, 0.95], and the tilt angle in the (YZ) plane was ~9. [29] In the beginning of subinterval III between 03:49:30 and 03:49:55 UT, a significant negative value of the fieldaligned current directed mostly duskward was observed (see Figure 4). The spatial distribution of J had a maximum outside the neutral plane in the region with B X ~ 3 nt, and thus, it was similar to the distribution of J (B X ) observed during subinterval I (see Figure 6b). After 03:50 UT, the absolute value of the parallel current decreased to zero value. [30] Summarizing the observations of the magnetic and current structures during the interval of interest, we may assume that both spatial and temporal changes in the magnetic configuration were peculiar for this event. The field-aligned electric current directed duskward dominated during the periods of a guide B Y field. In the spatial distribution of the field-aligned current density, a south-north asymmetry was clearly observed. We may also suggest that magnetic reconnection started earthward of Cluster spacecraft after 03:42 UT and evolved on the closed CPS field lines. This suggestion is also confirmed by the features of the particle dynamics, which we will discuss in the next section. Figure 6. Spatial distributions of the (a) Y component of the electric current density and the (b) field-aligned current density versus the value of B X at the Cluster barycenter. The spatial profiles of the current density observed during the three subintervals (I, II, and III) are indicated by different colors shown in the right part of the figure. 3270

7 3.2. Particle Dynamics [31] In this section we analyze the ion velocity distribution functions observed by Cluster-1 (Cluster-3 provides similar observations) and the electron pitch angle distributions measured by the four Cluster spacecraft between 03:42 and 03:55 UT. As previously discussed, the topology of the magnetic field changed significantly during this interval. Since these changes may affect the particle dynamics, we consider ion and electron behavior during the three subintervals characterized by the different spatial B Y profiles discussed in the previous section (see Figure 5). [32] Before analyzing in detail the particle dynamics observed during the interval of interest, we stress the key points of these observations, which are important for the further discussion of the mechanism possibly responsible for the strong enhancement of the B Y field near the neutral plane. As shown below, between 03:44 and 03:51 UT, the Cluster spacecraft was located inside the ion diffusion region and the following features of the particle dynamics were detected during this period: [33] 1. The complicated structure of ion velocity distribution functions, which were composed of (i) a hot and almost isotropic background plasma; (ii) a strongly accelerated ion flow with energies up to HIA energy threshold, moving in the tailward direction, and (iii) quasi-adiabatic and nonadiabatic populations of kev ions moving mostly duskward; [34] 2. The nonadiabatic ion population mainly contributed to the parallel (duskward) electric current, which was the dominating current component during the subintervals of a strong B Y field observations near the neutral plane (subintervals I and III); [35] 3. No electron beams were detected during the periods of a strong guide B Y field. Instead of this, the signatures of electron parallel heating (flat-top type of pitch angle distributions) were detected near the neutral plane and at the CPS magnetic field lines. Only isotropic electron distributions without flat-top or beam features were observed on the magnetic field lines of the outer PS; [36] 4. The anisotropy in the field-aligned direction of the electron pitch-angle distributions was either very small or preferentially at 180. So that these electrons could not produce the observed negative field-aligned current; [37] 5. Multipoint observations of ion and electron distributions show that a magnetic reconnection probably occurred at the closed CPS magnetic field lines earthward of Cluster spacecraft. The reconnected field lines were closed tailward of the Cluster location forming a plasmoid-like magnetic configuration Subinterval I: Bell-like Spatial Structure of BY. Ion and Electron Observations Between 03:42 and 03:46:40 UT [38] In Figure 7 we present 2-D cuts of the ion velocity distribution functions measured by the HIA spectrometer onboard Cluster-1 and plotted in the (V 1,V ), (V X,V Y ), and (V 1,V 2 ) planes. In this figure, we also present the time profiles of B X, B, and the parameter of nonadiabaticity k introduced by Büchner and Zelenyi [1989]. The q parameter ffiffiffi R k was calculated according to the equation:k ¼ r, where i R is the radius of curvature of the magnetic field lines and r i is the ion Larmor radius. The radius of curvature R of the magnetic field lines was calculated from the magnetic field curvature vector curv(b) =(b r)b. The time profiles of the k parameter were calculated for 1 kev ions (displayed in Figure 7 by the dashed lines) and for 5 kev ions (displayed by the solid lines) from Cluster-1 (shown in black) and Cluster-3 (shown in green) data. [39] In the beginning of the subinterval, between 03:42 and 03:44 UT, both Cluster spacecraft were located in the southern PS and the parameter k was 1.0 even for the 5 kev ions. This means that the majority of the ion population was magnetized. The ion velocity distribution functions measured during this period were formed by a hot and almost isotropic plasma (not shown). No earthward or tailward ion flows were detected. [40] From 03:44:20 UT, a tailward flow of accelerated ions started to be observed at the Cluster location. By this time, the value of the k parameter at Cluster-1 location decreased to ~2.0 for 5 kev ions (this time is marked in Figure 7 by the dotted line). From this time, the structure of the ion velocity distribution functions observed by both Cluster spacecraft became more complicated: it was formed by three ion populations (see in Figure 7 the ion distributions measured by Cluster-1 at 03:44:20 UT): [41] 1. a hot and almost isotropic background population; [42] 2. a field-aligned flow formed by ions with energies 5 kev W < 20 kev moving with a negative parallel velocity, i.e., mostly duskward; [43] 3. highly accelerated ions with energies up to HIA energy threshold moving perpendicular to the magnetic field mostly in the tailward direction. [44] No signature of the convection of the entire plasma population was observed. The velocity distribution function for ions with energies 1 kev was almost isotropic while the more energetic ions experienced a complicated dynamics including a chaotization in the region of k approximately a few units. In the cut of the ion velocity distribution function in (V 1,V 2 ), a phase bunching is observed for ions with W > 10 kev. This ion population is formed by the quasi-adiabatic (k ~1.0) and the nonadiabatic ions with energies >20 kev (k < 1.0) [Alexeev and Kropotkin, 1970]. [45] As the parameter k decreased further, the perpendicular and the field-aligned ion components observed in the velocity distributions became stronger and expanded in the velocity space. These populations were detected in the ion velocity distributions until the end of this subinterval (see in Figure 7 ion velocity distribution function measured at 03:46:24 UT). [46] In Figures 8a 8c, the three GSM components of the ion perpendicular velocity (HIA data) are presented together with the corresponding components of the electron perpendicular velocity and the drift velocity obtained from the 4 s averaged values of the magnetic field and the electric field data from the Electric Field and Wave (EFW) instrument [Gustafsson et al., 2001] onboard Cluster-1. It is seen from the figure that the electron component seems to fulfill the frozen-in condition as long as the ion population demonstrates a significant increase of the duskward velocity component that reached ~400 km/s by the end of this subinterval and well exceeded the drift velocity calculated as (E B)/ B 2. This means that the Cluster spacecraft penetrated into the ion diffusion region. 3271

8 Figure 7. (top) Two-dimensional cuts of the ion velocity distribution functions observed by Cluster-1 (HIA data) and plotted in (first row) (V 1, V ), (second row) (V X, V Y ), and (third row) (V 1, V 2 ) planes and measured with 12 s resolution at the moments indicated above each plot and (bottom) time profiles of the (first panel) absolute value of the magnetic field, the (second panel) X and (third panel) Y components of the magnetic field measured by the four Cluster spacecraft, and the (fourth panel) parameter of nonadiabaticity (k) calculated for 1 kev (dotted lines) and 5 kev (solid lines) ions from HIA data onboard Cluster-1 (shown by black color) and Cluster-3 (shown by green color). The red arrows indicate the moments of ion velocity distribution function measurements. 3272

9 Figure 8. (a c) Three components of the ion and the electron perpendicular velocity and the corresponding components of the drift velocity calculated as (E B)/B 2 from the electric and magnetic field data onboard Cluster-1and time profiles of the parallel and perpendicular components of the rb calculated for the local vector of the magnetic field at (d, e) Cluster-1 and (f, g) Cluster-2 locations (shown by black solid lines) and plotted together with the corresponding components of the ion current density (shown by black dotted lines in Figures 8d and 8e) and the electron current density (shown by grey dotted lines in Figures 8f and 8g). [47] The parameter k for electrons was much larger than 1.0 during the entire subinterval (not shown), so that electrons were magnetized. Nevertheless, the electron dynamics was more time and spatial dependent than the ion dynamics and reflected the temporal and/or spatial changes in the source. As a result, several types of electron pitch angle distributions were registered at different times and at different magnetic latitudes (B X ) by the four Cluster spacecraft. [48] Figure 9 shows the electron pitch angle distributions plotted for electrons with 0,90, and 180 pitch angles. Electron distributions were accumulated by the Plasma Electron and Current Experiment (PEACE) [Johnstone et al., 1997] over one spin period (4 s). The start of the corresponding spin period is indicated above each distribution. Similarly to the analysis performed by Nakamura et al. [2008], we used different symbols to represent the different types of electron distributions: square parallel heating with a shoulder structure (flat-top) at energies larger than 0.5 kev; empty oval parallel anisotropy but not as clear as a flat-top case; filled oval not a flat-top case without the anisotropy; and triangle a beam. The color of the symbols represents the dominant direction: red 0, blue 180, and black no preference. In Figure 9 we present the examples of each type of electron pitch angle distributions. Also, on the B X time profile measured by the corresponding Cluster spacecraft, we mark with these symbols the intervals when the different types of distributions were observed. This plot therefore provides spatial (inferred from the B X value) and temporal changes of the electron distributions in the vicinity of the neutral plane. [49] At the beginning of the subinterval (03:42:00 03:43:30 UT), when the Cluster spacecraft were in the southern PS, no signature of parallel heating (flat-top feature) or beam-like structure was detected in the electron pitch angle distributions. After 03:43:30 UT, Cluster-1, Cluster-4, and Cluster-2 consecutively reached the southern CPS (the region with B X >

10 Figure 9. Electron pitch-angle distributions observed during subinterval I. The parallel slices of electron phase density are shown by solid black lines, the antiparallel slices by solid grey lines, and the perpendicular ones by grey dotted lines. The pitch angle distributions were measured with a 4 s resolution and the start of each 4 s bin is indicated above each distribution. The different types of the distributions are displayed by the symbols on the B X time profile observed by the corresponding Cluster satellite and plotted in the bottom of the figure. We used the following symbols: square parallel heating with a shoulder structure (flat-top) for energies > 0.5 kev; empty oval parallel anisotropy without a flat-top feature; filled oval not a flat-top case without the anisotropy; and triangle a beam. The color of the symbols represents the dominant direction: red 0, blue 180, and black no preference. nt) and started to observe flat-top type distributions similar to the one measured by Cluster-1 at 03:43:40 UT (see Figure 9). Simultaneously at higher magnetic latitudes, Cluster-3 did not register a flat-top feature in the electron distribution. This indicates that the parallel heating of the electrons occurred at a source located on the CPS field lines and the magnetic field lines belonging to the PS were not involved in this process. [50] Near the neutral plane, the temporal changes in the electron pitch angle distributions became more pronounced. Namely, the flat-top feature in the electron distributions measured near the neutral plane emerged between 03:44:28 and 03:44:40 UT (Cluster-1 and Cluster-4 observations) and disappeared during the next neutral plane crossings (at 03:44:45 03:45:33 UT). However, during this time in the southern CPS, flat-top electron distributions were observed. This indicates the presence of a pronounced south-north asymmetry in the electron parallel heating. [51] The parallel heating started to be observed again near and at the neutral plane after 03:45:38 UT. By the end of the subinterval, the parallel heating of the electrons became stronger: the shoulder-like decrease started from ~1 kev. [52] Summarizing the electron observations made by the Cluster spacecraft in the CPS and at the neutral plane, we may assume that the strongest temporal changes occurred around the neutral plane. In the southern part of the CPS in the region with B X between 6 and 3 nt, all Cluster spacecraft detected a flat-top feature in the electron distributions at different times. Most likely, the parallel heating of the electrons did not cease on these field lines. It is interesting to note that no signatures of electron beams were detected during the entire subinterval. [53] During the main part of this subinterval, the anisotropy in the field-aligned direction of the electron pitch angle distributions was either very small or preferentially in 180. These electrons could produce the positive parallel current and so they are hardly responsible for the negative parallel current calculated as (rb) at the Cluster barycenter. In Figures 8d 8g we present a comparison of the field-aligned and perpendicular components of J calculated as rb at Cluster-1 and Cluster-2 locations and the corresponding components of the ion and the electron current density calculated as j =env (where V is the particle bulk velocity) from HIA and PEACE data obtained by Cluster-1 and Cluster-2, respectively. For a better comparison with the ion and electron currents, the parallel and perpendicular components of rb were calculated for the local vector of the magnetic 3274

11 Figure 10. Pitch angle distributions of the electron phase density observed during subinterval II. The format of the figure is the same as that of Figure 9. One more type of electron distribution is observed during this interval. This type is characterized by the domination of the 0 electrons in the energy range below 1 kev and of the 180 electrons in the energy range larger than 1 kev (a value of the corresponding anisotropy should exceed 40%, at least, at several energy channels). On the B X time profile, such distributions are displayed by blue diamonds. The time profile of B was added in the bottom of the figure. field at Cluster-1 (Figures 8d and 8e) and Cluster-2 (Figures 8f and 8g) locations. Figure 8d shows that the time profiles of the ion parallel current and (rb) are in a rough agreement. The correspondence between the electron parallel and perpendicular current components and the corresponding components of rb is poor. Thus, we may conclude that the observed negative parallel current was produced mainly by ions Subinterval II: Hall-like Spatial Structure of BY. Ion and Electron Observations Between 03:46:40 and 03:49:30 UT [54] During this subinterval the spatial structure of B Y was completely different from the one observed during the previous subinterval. Now in the vicinity of the neutral plane B Y is close to zero and it increases outside the neutral plane. Such a spatial profile of B Y could be formed by an electric current flowing at the neutral plane in the radial direction. [55] By the middle of this subinterval, the absolute value of the magnetic field observed at the neutral plane decreased down to ~1 2 nt, and as a result, the parameter of nonadiabaticity k became less than 1.0 even for 1 kev ions. During this subinterval, the ion velocity distribution functions measured by Cluster-1 and Cluster-3 at different magnetic latitudes were quite similar and formed by three main components (see in Figure 7 the ion velocity distribution functions measured at 03:47:51 UT by Cluster-1): [56] 1. a hot background ion population; [57] 2. a perpendicular component of ions with energies 1 10 kev moving mostly duskward; [58] 3. a field-aligned flow of highly accelerated ions (up to HIA energy threshold) moving mostly tailward and partially duskward. [59] It is worth noting that the field-aligned flow of highly accelerated ions was observed even near the neutral plane. This may indicate that the magnetic separatrix was located close to the neutral plane. [60] From Figures 8a 8c, it is seen that during this subinterval, the ion population demonstrates a significant increase of the duskward and tailward components of the perpendicular velocity up to ~400 km/s that well exceeds the drift velocity calculated as (E B)/B 2. These observations indicate that the Cluster spacecraft were in the ion diffusion region. [61] Figure 10 presents the electron pitch angle distributions and their types indicated by symbols and shown on 3275

12 the B X time profile in the same format as in Figure 9. One more type of electron distribution was observed during this subinterval. This type is characterized by the domination of the 0 (or 180 ) electrons in the energy range below 1 kev and of the 180 (or 0 ) electrons in the energy range larger than 1 kev (a value of the corresponding anisotropy should exceed 40%, at least, at several energy channels). On the B X time profile, such distributions are displayed by blue (or red) diamonds. [62] Between 03:46:40 and 03:46:50 UT, Cluster-1 and Cluster-4 were located near the neutral plane in the northern CPS and registered similar electron pitch angle distributions with a flat-top and a shoulder-like decrease starting from about 0.5 kev for the 0 electrons and with the 180 portion slightly exceeding the 0 portion in the energy range larger than 1 kev (see in Figure 10 the pitch angle distribution measured by Cluster-1 at 03:46:43 UT). At this time Cluster-2 and Cluster-3 were located at higher magnetic latitudes in the southern PS and detected a 0 beam of electrons with energies ~0.9 kev. This beam was directed tailward indicating on the source location earthward of the Cluster spacecraft. During this period, the spatial distribution of B Y started to change: at the location of Cluster-2 and Cluster-3, the value of B Y decreased to zero and changed its sign, so that the opposite signs of B Y were registered near the neutral plane and in the southern PS region indicating on an enhancement of the J X current within the Cluster tetrahedron. [63] The tailward moving electron beam was observed only during few spin periods and then it disappeared. Then, until 03:47:20 UT, only the flat-top distributions were observed in the southern CPS. The next observation of a 0 electron beam with energy ~0.6 kev was made by Cluster-1 and Cluster-4 in the southern CPS around 03:47:20 UT (see in Figure 10 the pitch angle distribution measured by Cluster-1 at 03:47:20 UT). The observation of this beam coincides with the observation of the first quadrupole B Y variation (in Figure 10 the intervals of the quadrupole B Y are shaded in yellow color). Again, this beam was observed during only a few spins and then it disappeared. Simultaneously, at the outer PS magnetic latitudes (at B X ~ 15 nt), electron distributions without beam or flattop features were detected (see in Figure 10 the pitch angle distribution measured by Cluster-3 at 03:47:19 UT). This indicates that the process of the electron beam generation or parallel heating occurred mainly on the CPS field lines. [64] During the second observation of the quadrupole B Y structure, the electron beam appeared again. In the southern CPS (at B X ~ 2 nt), it was directed tailward and had energy ~0.8 kev. Almost simultaneously (at 03:47:46 UT) in the southern PS (at B X ~ 10 nt), Cluster-2 observed a cold electron beam (~0.2 kev) directed earthward. Such spatial distribution of the electron beams is in agreement with the one expected in the region tailward of a magnetic X line [e.g., Nagai et al., 2001; Fujimoto et al., 2001]. However, in this event, the energy of the tailward moving electrons, which could be accelerated near the X line, was significantly smaller than the one reported in earlier studies. [65] It is worth noting that Cluster-1, which between 03:47:40 and 03:47:53 UT was located in the southern CPS at a little bit higher magnetic latitudes than Cluster-4, did not observe the electron beam. Instead it observed a flat-top electron distribution. This indicates the strong spatial localization of the tailward beam. Unfortunately between 03:47:53 and 03:48:05 UT during the neutral plane crossing by Cluster-1, the analysis of the electron pitch angle distributions was impossible due to the poor angle coverage. [66] During the last period of the quadrupole B Y (03:48:35 03:48:50 UT) near the neutral plane, Cluster-1 and Cluster-4 observed flat-top distributions with a stronger parallel heating for the 180 electrons. During this time, Cluster-2 observed in the southern CPS a tailward streaming beam with energy ~0.5 kev. After 03:48:51 UT, this beam disappeared and until the end of this subinterval, Cluster-2 registered only flat-top distributions without the beam feature. Simultaneously, in the southern PS, no parallel heating of the electrons was detected (Cluster-3 observations). [67] Around 03:48:59 UT, both Cluster-1 and Cluster-4 observed an earthward streaming electron beam in the northern part of the CPS. This beam had almost the same energy (~0.5 kev) as the tailward streaming beam observed before in the southern CPS by Cluster-2. Thus, it is most likely that Cluster-1 and Cluster-4 detected the electron beam reflected at the magnetic closure tailward of the Cluster spacecraft. This beam was observed during only one spin period and then disappeared, although Cluster-1 and Cluster-4 stayed at similar magnetic latitudes until the end of the subinterval. We may assume that by the end of subinterval II, the Cluster spacecraft exited the region where the Hall effect was significant. [68] Summarizing the electron observations provided by the four Cluster spacecraft between 03:46:40 and 03:49:30 UT, we may assume a transient enhancement/reduction of the Hall effect in the ion diffusion region. The enhancements of the Hall effect manifested themselves in the transient observations of the quadrupole B Y variations and electron beam generation. The source of the beam generation and electron parallel heating (possibly magnetic reconnection) was located on the CPS magnetic field lines (within the region with B X < 10 nt) earthward of the Cluster spacecraft. [69] Finally, it must be stressed that during this subinterval, a more or less good conformity was observed between the parallel component of the ion current and the corresponding component of rb (see Figures 8d and 8e). The agreement between the electron parallel and perpendicular current components and the corresponding components of rb is much worse. This discrepancy could be caused by the presence of fine electron structures with spatial scales smaller than the characteristic scale of the Cluster tetrahedron Subinterval III: Bell-like Spatial Structure of BY. Ion and Electron Observations Between 03:49:30 and 03:52:00 UT [70] During the last subinterval the negative value of the guide B Y increased again and reached its maximum value approximately 6 nt at the neutral plane, forming the belllike shape of the B Y spatial profile (see Figure 5). [71] Around 03:50 UT, Cluster-1 and Cluster-4 crossed the neutral plane moving toward the southern CPS. Both in the northern and the southern parts of the CPS as well as in the close vicinity of the neutral plane, Cluster-1 observed ion velocity distribution functions resembling the ones observed previously during subinterval I, when a similar bell-like structure of a guide B Y was detected (compare in Figure 7 the ion velocity distribution functions measured at 03:49:55 UT and at 03:44:20 UT). 3276

13 Figure 11. Pitch angle distributions of the electron phase density observed during subinterval III. The format of the figure is the same as that of Figure 9. Until 03:50:10 UT, the k parameter was ~1.0 even for 1 kev ions, so that the phase bouncing observed for ions with energies higher than 5 kev was most likely produced by the nonadiabatic ion population. [72] After 03:50:10 UT, the k parameter increased and reached ~4.0 for 5 kev ions and ~6.0 for 1 kev ions at Cluster-1 location. However, a chaotic ion population with energies 5 kev < W < 20 kev moving mostly duskward was still clearly observed in the ion velocity distributions until 03:51:10 UT along with a more energetic (up to HIA energy threshold) perpendicular moving ion component. After 03:51:10 UT and until the end of the subinterval, both Cluster-1 and Cluster-3 registered only hot and isotropic ion velocity distributions. [73] From Figure 8 it is seen that between 03:49:30 and 03:51 UT, the ion perpendicular velocity exceeded the drift velocity calculated as (E B)/B 2, so that the Cluster spacecraft were still in the ion diffusion region. After 03:51 UT, the Cluster spacecraft went out of the diffusion region and the observed ion perpendicular velocity became of the order of the drift velocity calculated from the magnetic and electric field data. [74] Figure 11 shows the electron pitch angle distributions observed during this subinterval by the four Cluster spacecraft. The format of this figure is the same as that of Figure 9. In the electron pitch angle distributions measured in the beginning of this subinterval (between 03:49:40 and 03:49:44 UT) by Cluster-1 and Cluster-4, a 0 beam with energy ~0.6 kev was detected. This beam was observed only during one spin period when both Cluster spacecraft were located in the northern part of the CPS very close to the neutral plane (in the region with B X ~2 1 nt). We can hardly infer from these observations whether this beam is produced due to the Hall effect or if it represents a beam generated earlier, having experienced a bouncing motion along the closed magnetic field lines. [75] After 03:49:44 UT, Cluster-1 and Cluster-4 crossed the neutral plane, moving toward the southern CPS, and observed flat-top type distributions in which the 180 portion exceeded the 0 portion at energies higher than 1 kev. Similar pitch angle distributions were observed by Cluster- 2 at higher magnetic latitudes in the southern CPS and in the PS (in the region with B X between 6 and 12 nt). During this time (between 03:49:35 and 03:50 UT), Cluster-3 was located in the southern part of the PS at higher magnetic latitudes than Cluster-2 and observed electron pitch angle distributions without beam or flat-top features. This again confirms that the electron parallel heating occurred mostly on the CPS field lines. [76] Between 03:49:30 and 03:50:10 UT when the Cluster spacecraft were located inside the parallel current layer, the 180 anisotropy in the electron distributions prevailed over the 0 one, except for the short interval of the 0 beam observation. Since the portion of 180 electrons cannot contribute to the negative parallel current observed during this period, we may assume again that the main contribution to this current was produced by ions. This situation is similar to the one which took place during subinterval I, when a similar bell-like structure of B Y was observed. [77] The observed correspondence between the time profiles of the perpendicular current calculated as (rb) at Cluster-1 and Cluster-2 locations and the corresponding profiles of the ion and electron perpendicular currents is worse. Probably both ions and electrons contribute to the perpendicular current component. 4. TC-1 Observations Earthward of the Reconnection Region [78] Figure 12 shows TC-1 observation between 03:42 UT and 03:55 UT: 2-D cuts of the ion velocity distribution functions in (V 1,V ), (V X,V Y ), and (V 1,V 2 ) planes, the three GSM components of the ion velocity and the magnetic field observed. During the entire interval, TC-1 was located in the southern PS and did not approach the neutral plane. [79] Before 03:44:42 UT, TC-1 observed mostly a hot and isotropic ion population (not shown). During this time, the X component of the ion velocity was approximately zero. Plasma moved northward with V Z ~ 50 km/s and duskward 3277

14 Figure 12. TC-1 observations on 22 September 2004 between 03:42 and 03:55 UT. (top) Twodimensional cuts of the ion velocity distribution functions plotted in (first row) (V 1, V ), (second row) (V X, V Y ), and (third row) (V 1, V 2 ) planes and measured at the moments indicated above each plot and (bottom) three components of the ion velocity and the magnetic field. The red arrows indicate the moments of ion velocity distribution function measurements. with V Y ~ 100 km/s suggesting the presence of a dawn-dusk and northward electric field. [80] After 03:44:42 UT, ion velocity distribution functions show an enhancement of the field-aligned component (>5 kev) moving with negative parallel velocity. This ion population was moving duskward and earthward (see Figure 12). Around 03:46 UT, a positive pulse of B Z was observed. This positive B Z variation could be related with the arrival of a dipolarization front propagating earthward [e.g., Runov et al., 2009], which indicates the onset of a reconnection process tailward of the TC-1 location [e.g., Sitnov et al., 2009; Birn et al., 2011]. The dipolarization front was observed by TC-1 during ~1 min and then B Z relaxed to the zero level. If one assumes that this dipolarization front and the tailward moving plasmoid, whose tailward edge (positive B Z pulse) was observed by the Cluster spacecraft around 03:44:50 UT had a common source and propagated with similar velocities, then the distance ΔX DSP between TC-1 and the possible X line could be roughly estimated as ΔX DSP =(ΔX DSP-Cluster + V X Δt)/2 ~ 5 R E. Here, ΔX DSP- Cluster is the radial distance between TC-1 and the Cluster spacecraft, V X is the propagation velocity of the positive B Z pulse, estimated from Cluster observations (see section 3.1), and Δt is the time delay in the observations of this pulse at Cluster and TC-1 locations. Thus, the location of the X line could be roughly estimated at X ~ 15 R E. This is also in an agreement with the estimation made using the time lag (~20 s) between the start of the observations of the tailward and earthward ion flows at Cluster and TC-1 locations, respectively. [81] Simultaneously with the positive B Z pulse, the Y component of the magnetic field experienced a strong negative variation up to 10 nt. This negative B Y variation at TC-1 location roughly coincided with the transformation of the negative guide B Y into the quadrupole-like B Y structure observed by Cluster. Since TC-1 was located in the southern 3278

15 Figure 13. Observations of energetic particles between 03:42 and 03:55 UT by Cluster and TC-1. Time profiles of the fluxes of (first panel) energetic protons (>95 kev) and (second panel) energetic electrons (>50 kev) measured by the RAPID spectrometers onboard the four Cluster satellites, the (third panel) Z component of the magnetic field measured by the four Cluster spacecraft, (fourth panel) electron flux in the energy range kev measured by HEED detector onboard the TC-1 spacecraft, and the (fifth panel) Z component of the magnetic field measured at TC-1 location. PS, the observed negative B Y variation may be caused by the Hall current generated in the earthward side of the reconnection region. [82] The ion velocity distribution functions observed between 03:44:42 and 03:46:20 UT were formed by two components: (i) a hot and isotropic background population and (ii) a field-aligned ion population with energies >5 kev propagating with negative parallel velocity mostly earthward (see in Figure 12 the ion velocity distribution measured at 03:45:55 UT). Between 03:46:20 and 03:46:36 UT when the maximum of the positive B Z pulse was registered, a strong pitch angle scattering is observed in the earthward moving ion component (see in Figure 12 the distribution measured at 03:46:20 UT). [83] Around 03:47 UT, TC-1 exited to the outer part of the southern PS probably due to a sudden CS thinning (B X decreased down to 24 nt). Now three components could be distinguished in the ion velocity distribution functions (see in Figure 12 ion distribution measured at 03:47:17 UT): [84] 1. an isotropic background population which was colder than the one observed deeper in the PS; [85] 2. a strong field-aligned flow of highly accelerated ions moving earthward; [86] 3. a perpendicular component of ions with lower energies (<5 kev) moving dawnward. [87] These features in the ion distributions were observed until 03:47:50 UT. After 03:47:50 UT, the dawnward moving ion component disappeared, and until 03:49:03 UT, only two ion populations the cold and isotropic background population and the field-aligned flow of highly accelerated ions were observed. [88] It should be noted that although the background ion population was almost isotropic indicating that TC-1 was still located on closed magnetic field lines, a tailward moving (reflected near the Earth) flow of highly accelerated ions was not observed. This indicates that TC-1 crossed the magnetic separatrix between the field lines connected with the acceleration source (probably near-earth magnetic reconnection) located tailward of TC-1 and earthward of Cluster spacecraft and the field lines, which were not mapped to the acceleration source, closing farther in the tail. In other words, the magnetic separatrix was embedded into the PS field lines (see Figure 14a in section 6). [89] After the passage of the first dipolarization front, the B Z component again started to increase and reached its maximum value ~13 nt around 03:50 UT. After this, B Z decreased down to ~6 nt and was stable until the end of the interval of interest. This dipolarization event was completely different from the previous one. It was longer: the duration of the B Z increase lasted ~4 min and fluctuated strongly. After reaching the maximum value, B Z did not relaxed to zero level but remained positive until the end of the interval of interest. This positive B Z enhancement could be related with the negative B Z enhancement observed by Cluster. Both perturbations were prolonged in time and highly fluctuating. During this time, the negative B Y increased further and around 03:48 UT, it reached its maximum negative value approximately 16 nt. [90] After 03:48 UT, the negative value of B Y started to decrease, and by the time when B Z reached its maximum, B Y became almost zero (around 03:50 UT). At 03:49 UT, TC-1 immersed deeper into the PS, and by 03:50 UT, it approached the CPS (B X ~ 5 nt). In the ion velocity distribution functions measured between 03:49:03 and 03:49:50 UT, the flow of highly accelerated ions was neither purely field-aligned nor perpendicular and moved earthward and partially duskward (in Figure 12 see the ion distribution measured at 03:49:44 UT). After 03:50 UT, the ion velocity distribution function became mostly hot and isotropic (not shown). [91] Summarizing the TC-1 observations, we may conclude that the acceleration process, which generated the highly accelerated ion flow in the southern PS, occurred on closed field lines tailward of TC-1 location. From the analysis of TC-1 and Cluster observations, we may infer that the ion acceleration process started around 03:44 UT. 5. Energetic Particle Observations [92] Figure 13 presents the time profiles of the fluxes of energetic protons (W > 95 kev) and electrons (W > 50 kev) measured by the Research with Adaptive Particle Imaging Detector (RAPID) [Wilken et al. 2001] onboard the four Cluster spacecraft, the time profiles of the B Z component measured by the four Cluster spacecraft, the time profile of the electron flux in the energy range kev measured by the High Energy Electron Detector (HEED) [Cao et al., 2005; Liu et al., 2005] onboard the TC-1 spacecraft, and 3279

16 the time profile of B Z observed by TC-1 between 03:42 and 03:55 UT. [93] Between 03:43 and 03:44 UT, all Cluster spacecraft registered a flash of energetic electrons. The flash was observed during the positive B Z variation registered by Cluster between 03:42 and 03:45 UT. During this time, TC-1 did not observe high-energy electrons. At TC-1 location, the increase of the energetic electron flux started to be observed from 03:46 UT, at the time of the first dipolarization front arrival (Figure 13). However, the maximum of the energetic electron flux was reached around 03:48:30 UT and it was associated with the second positive B Z enhancement. These electrons could be accelerated by inductive electric fields during the dipolarization process, which is able to energize electrons to a few hundreds of kev [e.g., Williams et al. 1990; Birn et al., 2004]. The possible contributions to the electron energization from a betatron and Fermi acceleration were also considered by, e.g., Khotyaintcev et al. [2011] and Fu et al. [2011]. [94] The increase of the energetic electron flux at TC-1 location approximately coincided in time with the increase of the energetic ions observed by Cluster between 03:48 and 03:50 UT. At the Cluster location, the energetic ions were observed ~4 min later than the first flash of energetic electrons, so that these two flashes of energetic particles could be related to different onsets of acceleration. During this time, the B Z component at the Cluster location already became negative (Figure 13). [95] The second flash of energetic electrons was observed at the Cluster location more or less clearly only by Cluster- 1 and Cluster-4 satellites around 03:51:15 UT. At this time both spacecraft experienced a short excursion into the deep PS (at B X ~ 9 nt) while Cluster-2 and Cluster-3 stayed in the outer PS (at B X ~ 12 nt). The amplitude of this enhancement was smaller than the amplitude of the first electron flash. The B Z component was negative and strongly fluctuating. Probably these electrons were energized by the island surfing mechanism [Oka et al., 2010] in the course of their trapping inside the secondary small-scale magnetic islands generated in the reconnection region [e.g., Drake et al., 2006; Eastwood et al., 2007]. No signature of energetic ion acceleration was observed during this period. 6. Discussion and Test Particle Simulations [96] Regarding the observations made by the Cluster and the TC-1 spacecraft in the active PS and CS between 03:42 and 03:55 UT on 22 September 2004, several features should be pointed out. It was a geomagnetically quiet period ( AL < 50 nt), in spite of the signatures of a significant energy dissipation in the near-earth magnetotail observed by all spacecraft. The tailward flow of accelerated ions began to be observed at the Cluster location around 03:44:20 UT and the earthward flow commenced at TC-1 location around 03:44:42 UT. Between 03:42 and 03:45 UT, a positive B Z variation and then its southward turn were observed by the Cluster spacecraft. The timing analysis applied to Cluster observations of the positive B Z pulse around 03:45 UT shows its tailward propagation with V X ~ 190 km/s. This may indicate the propagation of a slowly moving plasmoid [e.g., Nishida et al., 1986]. TC-1, in turn, detected two dipolarization fronts with some delay relatively to Cluster observations. Taking for the first dipolarization front a similar value of the propagation velocity, we may roughly estimate the source (probably an X line) location between Cluster and TC-1 at X ~ 15 R E. [97] One of the unique features of this event is the strong negative variation in B Y observed first by the Cluster spacecraft at X ~ 18 R E and then, with a few minutes delay, by the TC-1 spacecraft at X ~ 10 R E. The negative variation of the B Y field was almost two times larger than the IMF negative B Y (see Figure 2) and obviously had temporal character. Thus, we may assume an internal CS source of this guide field. [98] In this paper we do not intend to provide a comprehensive explanation of the origin of the strong B Y observed in the CS, but we would like to point out some peculiarities of the magnetic field and particle dynamics and make some assumptions about the possible source of this B Y enhancement. First of all, it is worth noting that during the period of interest, the negative B Y field was not observed simultaneously by Cluster and TC-1. At the Cluster location, the maximum negative value of B Y was observed near the neutral plane, so that the spatial profile of B Y had a bell-like shape, at least, during subintervals I (03:42 03:46:40 UT) and III (03:49:30 03:55 UT). However, TC-1 observed almost zero B Y until ~03:46 UT possibly because it was located in the outer part of the PS (at B X ~ 15 nt) where B Y was close to zero at the Cluster location as well. [99] During subinterval II (03:46:40-03:49:30 UT), the guide B Y field disappeared at the Cluster location. This change had obviously a temporal character and it was related with a change in the current structure due to the appearance (or increase) of the Hall current in the tailward side of a near-earth magnetic X line. Indeed, the quadrupole structure of B Y usually associated with the tailward Hall current outflow was observed by the Cluster spacecraft during three short periods: between 03:47:12 and 03:47:31 UT, between 03:47:44 and 03:48:13 UT, and between 03:48:30 and 03:48:50 UT (see Figures 4 and 5). Around 03:48 UT, the TC-1 spacecraft also observed an increase of negative B Y (see Figure 12). Since around this time TC-1 was located in the southern PS (in the region with B X between -10 nt and -23 nt) but earthward of the possible reconnection region, the observed negative B Y variation could also be caused by the Hall current outflowing from the X line and directed earthward [e.g., Nagai et al., 2001; 2003; Runov et al., 2003b]. [100] However, in contrast with previous observations of the quadrupole B Y associated with the Hall current system near a magnetic X line [see, e.g., Runov et al., 2003b; Borg et al., 2005], in our case the appearance of the Hall current had a bursty character. It must also be stressed that the tailward ion flow at the Cluster location started at 03:44:20 UT and the earthward flow was observed at TC-1 location starting at 03:44:42 UT. This indicates that the reconnection process generating these flows started between TC-1 and Cluster locations at around 03:44 UT, i.e., significantly earlier than the quadrupole B Y variations. Moreover after 03:44 UT, the ion population observed at the Cluster location demonstrated violation of the frozen-in conditions (see Figures 8a 8c). This indicates that by this time the Cluster spacecraft had already penetrated into the ion diffusion region. However, no signatures of quadrupole B Y variations indicating the presence of the Hall current were observed by Cluster until 03:47:15 UT. Also in the electron pitch angle 3280

17 Figure 14. (a) An illustration of the possible scenario of the enhancement of negative guide B Y field near the neutral plane tailward of the reconnection region due to the presence of a tailward (earthward) electric current (J) in the northern (southern) part of the PS. The red lines represent the magnetic separatrix between field lines reconnected between Cluster and TC-1 spacecraft and PS field lines, which are closed tailward of Cluster location and are not involved into the reconnection process. The region of ion nonadiabatic dynamics is shaded by blue color. (b) An illustration of the peculiarities of the nonadiabatic ion dynamics in the CS under the presence of a negative guide B Y with a bell-like spatial profile. The solid grey line shows a magnetic field line. An ion coming from the southern PS (its trajectory is shown by the red dotted line) experienced a quasi-larmor rotation in the CS around the finite negative B Z so that the field line lies inside the ion quasi- Larmor circle (shown by the black dotted line) and then, at the point of the second separatrix crossing (2), is ejected to the northern PS and moves tailward. An ion coming to the CS from the Northern Hemisphere (its trajectory is shown by the green dotted line) starts to rotate in the CS plane so that its original field line does not lie inside the quasi-larmor circle. The partial magnetization of the ion by a guide field impedes its successive motion duskward. Such ions pass smaller distances in the duskward direction and depending on their pitch angles are ejected either back to the Northern Hemisphere or to the southern PS. distributions measured by Cluster before ~03:46:50 UT, no beams, which could be responsible for the Hall current, were observed. Instead, during this period, only the signatures of electron parallel heating on the CPS field lines were detected. These observations may be explained by the possible thermalization of the electron beam and reduction of the Hall current predicted in simulations of magnetic reconnection in the presence of a guide field by Pritchett and Coroniti [2004]. [101] The bursty observations of the quadrupole B Y at the Cluster location were possibly related to the expansion of the reconnection region and its approaching to the Cluster spacecraft. Indeed, the absolute value of the magnetic field B observed at the neutral plane between 03:47:40UT and 03:48:30UT significantly decreased and became 1 ntin comparison with the B magnitude observed at the neutral plane during its previous and next crossings by the Cluster satellites. From these observations, we may deduce that in the presence of a guide B Y, the Hall effect becomes important and the Hall current effectively affects the guide B Y only very close to the electron diffusion region. [102] The negative guide B Y field with the bell-like shape of the spatial profile appeared again at the Cluster location after 03:49:30 UT (see subinterval III in Figures 4 and 5). These changes in the B Y field obviously had a temporal character and were related to the disappearance or the significant reduction of the Hall current at Cluster and probably at TC-1 location. Since the tailward flow of accelerated ions was observed by the Cluster spacecraft until 03:52 UT, we may assume that the reconnection region shrank in space but did not disappear yet. It is worth noting that during this event, no signature of the X line tailward retreat was observed. [103] According to Ampere s law, the bell-like B Y profile observed during subintervals I and III implies a negative (tailward) current J X in the northern part of the PS and a positive (earthward) J X in the southern part of the PS (see the cartoon in Figure 14a). Indeed, in the beginning of the interval of interest when the Cluster quartet was still in the southern PS (in the region with B X 10 nt), a positive value of the J X current density ~1 3 na/m 2 was observed (see Figure 4). Then the Cluster barycenter approached the neutral plane and we cannot infer if this current continued in the southern PS. [104] Now let us consider the peculiarities of the ion dynamics observed by the Cluster spacecraft in the CPS and near the neutral plane between 03:42 and 03:55 UT, which may contribute to the generation of the J X current in the PS. The CS started to thin from the beginning of this interval and by the middle of subinterval II, it became of the order of several hundreds of kilometers [Artemyev et al., 2008]. In such CS, the kev ions become unmagnetized. Indeed by 03:44:30 UT, the parameter of nonadiabaticity k became ~1.0 for 5 kev ions (see Figure 7) even in the presence of the guide B Y.Thisresults in the partial chaotization [Büchner and Zelenyi, 1989] of the preexisting population of kev ions in the CPS. As a result in the ion velocity distribution functions measured in the CPS during subintervals I and III, the three ion populations, described in section 3.2, were observed (see Figure 7). As the parameter k decreased further, the nonadiabatic ion population moving duskward became more pronounced and expanded in the velocity space. [105] During this event, a significant negative parallel current was observed. The time profile of its current density calculated as (rb) at the Cluster barycenter was in a good agreement with the time profile of the ion field-aligned current density J ion (see Figure 8d). The contribution of the electrons to (rb) is not so evident. Indeed, during the main parts of subintervals I and III, the parallel anisotropy in the electron pitch angle distributions either was absent or was observed in the 180 electron component. These electrons could not carry negative parallel current. The absence of an evident electron contribution to the observed electric current could be related to the thermalization of electron beams by electromagnetic turbulence in the ion diffusion region leading to the reduction of the Hall current, as it was shown in simulations of magnetic reconnection under the presence of a guide field [e.g., Pritchett and Coroniti, 2004]. [106] In the ion velocity distribution functions measured during this event by both Cluster-1 and Cluster-3, no signatures of the convection of the entire PS population across 3281

18 Figure 15. (left column) The distributions of the parameter of nonadiabaticity k, calculated for 5 kev ions at Cluster-1 (shown by black circles) and at Cluster-3 (shown by green circles) locations versus B X observed by the corresponding satellite during the three subintervals. (right column) The distributions of the ion density, which was calculated from HIA data for energy range 5 32 kev (data from Cluster-1 and Cluster-3 are displayed by black and green crosses, respectively) versus B X. the magnetic field were observed. The significant increase of the ion positive V Y and negative V X was due to the appearance of energetic ion populations consisting of locally unmagnetized ions moving mostly duskward and of an energetic ion population accelerated in the tailward direction at a remote source (in the near-earth X line). Observations of similar kinetic features for bursty bulk flows were reported earlier by Chen et al. [2000]. The duskward moving ion population produced the negative parallel current observed during this event Test Particle Simulations [107] In this event, a slight south-north asymmetry was observed in the spatial distribution of the parallel current density (Figure 6), in the spatial distributions of the k parameter and the density of the high-energy ions (calculated from HIA data for energy 5 32 kev range; see Figure 15) during the periods of a guide B Y (subintervals I and III). In order to explain these peculiarities, we analyze the kinetic features of test particle trajectories calculated in the presence of a negative guide B Y with the bell-like spatial profile. For a comparison, we present the results of simulations made without a guide B Y (Figures 16a and 16b) and in the presence of a guide B Y (Figures 16c and 16d). [108] To study the kinetic features of the particle dynamics in a thin CS under the presence of a strong negative B Y with a bell-like spatial profile, we analyze test particle trajectories in a prescribed magnetic field, which was uniform along X and Y directions: B X ¼ B X 0 tanhðz=lþ; B Y ¼ B Y0 cosðzp=2lþ; Z L B Y ¼ 0; Z > L B Z ¼ B Z0 ¼ const [109] HereL is the width of the CS and B X0, B Y0,andB Z0 are the components of the magnetic field at the boundaries of the 3282

19 GRIGORENKO ET AL.: CURRENT SHEET WITH A GUIDE BY FIELD Figure 16. Test particle trajectories of nonadiabatic ions interacting with the CS. The trajectories were calculated for six pitch angle values (their values are presented in radians and shown by different colors) in a prescribed magnetic field and plotted in the (XY) and (YZ) planes. No guide BY field and BZ < 0, ions are coming to the CS from the (a) northern PS and from the (b) southern PS. A negative BY field exists in the system and has a bell-like shape of the spatial profile with a maximum negative value at the neutral plane, BZ < 0, ions are coming to the CS from the (c) northern PS and from the (d) southern PS. A trajectory of a magnetized electron (shown by the black line) marks a field line. 3283