Cluster observation of plasma flow reversal in the magnetotail during a substorm

. We investigate in detail a reversal of plasma ﬂow from tailward to earthward detected by Cluster at the downstream distance of ∼ 19 R E in the midnight sector of the magnetotail on 22 August 2001. This ﬂow reversal was accompanied by a sign reversal of the B z component and occurred during the late substorm expansion phase as revealed by simultaneous global view of auroral activity from IMAGE. We examine the associated Hall current system signature, current density, electric ﬁeld, Lorentz force, and current dissi-pation/dynamo term, the last two parameters being new features that have not been studied previously for plasma ﬂow reversals. It is found that (1) there was no clear quadrupole Hall current system signature organized by the ﬂow reversal time, (2) the x -component of the Lorentz force did not change sign while the other two did, (3) the timing sequence of ﬂow reversal from the Cluster conﬁguration did not match tailward motion of a single plasma ﬂow source, (4) the electric ﬁeld was occasionally dawnward, producing a dynamo effect, and (5) the electric ﬁeld was occasionally larger at the high-latitude plasma sheet than near the neutral sheet. These observations are consistent with the current disruption model for substorms in which these disturbances are due to shift-ing dominance of multiple current disruption sites and turbu-lence at the observing location.


Introduction
Plasma environment in the Earth's magnetotail has been studied since the early days of the Vela satellites in the late Correspondence to: A. T. Y. Lui (tony.lui@jhuapl.edu) 1960's (see, e.g., Hones, 1973).Plasma flows were inferred from these early measurements even though the plasma detectors and the satellite spin axis orientation were not ideally designed to obtain accurate determination of plasma flow direction.It has been recognized in early substorm research that high-speed plasma flows occur during substorms.The lack of multi-point observations in the magnetotail and the ambiguity in differentiating temporal from spatial variations have posed major obstacles in advancing our understanding of substorm phenomena in the magnetotail.The ISEE-1/2 mission demonstrated the power of multi-point measurements from satellites flying in close formation.Subsequently, the International Solar Terrestrial Physics (ISTP) program has brought the prowess of multi-point magnetospheric observations to the forefront of space research.The Cluster-II mission elevates the capability of ISEE-1/2 and ISTP missions even further by the ability to determine gradients in three-dimensions as well as differentiate temporal from spatial variations.This capability provides valuable constraints in data interpretation on magnetotail dynamics.
It is evident that the new capability of Cluster can help to advance our understanding of substorm phenomena.Several investigations have utilized Cluster to examine substorm phenomena in the tail (e.g., Nakamura et al., 2002;Runov et al., 2003).In this paper, we examine a plasma flow reversal in the magnetotail observed by Cluster on 22 August 2001.Plasma parameters associated with the plasma flow reversal, including the Lorentz force and the current dissipation/dynamo term that have not been investigated before, are studied.These observations are used to evaluate two prominent substorm models that will be tested by the upcoming NASA mission THEMIS.The near-Earth neutral line (NENL) model accounts for this type of plasma flow reversal by the tailward movement of an X-line (Hones, 1979;Baker et al., 1996) for this type of plasma flow reversal by multiple current disruption sites developing at progressively further downstream distances (Lui, 1991(Lui, , 1996)).It is found that the observed features associated with this plasma flow reversal are consistent with the CD substorm model.

Overview of activity
The ground-station magnetic activity on 22 August 2001 around the plasma flow reversal interval is shown by the AU/AL indices given in Fig. 1.Within 08:00-14:00 UT, the AL index showed the strengthening of the auroral electrojet signifying substorm development.The plasma flow reversal event studied here occurred during late substorm expansion at the time marked by the vertical dashed line.
Cluster satellites were at a downstream location of about (-18.9, -3.4,0.9) R E in GSM coordinates at 09:50 UT, i.e., in the midnight sector of the magnetotail.The dimension of the tetrahedron formed by the four Cluster satellites was ∼2000 km, with C3 (SAMBA) at its apex south of the tetrahedron base.Substorm onset seen by the IMAGE/FUV imager was at ∼09:20 UT, with a sizeable auroral bulge developed at ∼09:26 UT.This and the subsequent auroral activity around the flow reversal are shown in Fig. 2. Cluster crossed the neutral sheet from north to south during this interval, as shown in Fig. 3 with measurements from C3. Shown in the panels are the number density, the ion temperature (red trace), the x-and y-components of plasma flow, the y-components of electric field and -V ×B (cross-product of plasma flow and magnetic field -red trace), the x-and y-components of the magnetic field, and the z-component of the magnetic field.The plasma, electric field, and magnetic field measurements on Cluster were taken by CIS (Rème et al., 2001), EFW (Gustafsson et al., 1997), andFGM (Balogh et al., 2001), respectively.The plasma data used here are from CIS/HIA.The measurements show several dipolarizations of the magnetic field and plasma flow reversals.It also can be noted that the y-components of electric field and -V ×B matched well in general but significant deviations can also be seen intermittently.

2.2
Magnetic signatures and timing sequence of the plasma flow reversal event There were some large-scale MHD oscillations of the plasma sheet previously studied by Volwerk et al. (2003), Louarn et al. (2004), andFruit et al. (2004).There was a notable plasma flow reversal from tailward to earthward near the end of the substorm expansion phase at ∼09:54 UT that was not discussed by the previous studies.Figure 4 shows in expanded time scale the perpendicular plasma flow V x and V y components from CIS/HIA on C1 (RUMBA) and C3, from CIS/CODIF on C4 (TANGO) (no data from CIS on C2), the magnetic field B x and B z components from all four spacecraft.Different standardized colors are used to denote data from different spacecraft, as indicated by the colored labels at the top left corner.In panel (d), the traces are offset for clarity with the corresponding zero level shown.The plasma flow reversal was generally accompanied by a reversal in the B z component.The signatures of the V x and B z in this event are similar to the flow reversal and X-line encounter interval of 09:47-09:51 UT on 1 October 2001 reported by Runov et al. (2003).In particular, there was a period of very low speed flow within the flow reversal interval while the B z component changed sign from negative to positive.
A quadrupole B y signature is expected around an X-line (Sonnerup, 1979), i.e., B x B y <0 tailward of the X-line (before flow reversal) and >0 earthward of it, valid for both north and south of the current sheet.In order to see this, one may note that tailward of an X-line, B y <0 north of the neutral sheet (B x >0) and B y >0 south of it (B x <0).Therefore, the product B x B y <0 in both north and south of the neutral sheet.Similarly, earthward of an X-line, B y >0 north of the neutral sheet and B y <0 south of it.The product B x B y >0 in both north and south of the neutral sheet.Thus, the magnetic perturbation associated with the Hall current system can be revealed better by the sign of the productB x B y , eliminating the confusion introduced by plasma sheet flapping.This product, normalized by the B x magnitude, is shown in Fig. 5 with the flow reversal time for the corresponding satellite indicated by the arrow below the trace.There are significant deviations from the expected trend.For example, for C1, the parameter was only intermittently negative prior to flow reversal and was definitely positive ∼1 min before the flow reversal.For C3, the trend was more consistent with expectation, but there were many significant deviations as well.The trend for C4 was similar to C1, with the parameter turning positive nearly 2 min before flow reversal.Also, the expected sequence of flow reversal for a tailward moving plasma flow source based on the X GSM location of the Cluster configuration is C1-C3-C4.The observed time sequence is C3-C1-C4 instead.Although the flow reversal timing is based on CIS/HIA data, we have also examined the CIS/CODIF data and found them to give almost identical timing.

Current associated parameters during the plasma flow reversal
The parameters associated with current density during the flow reversal interval are shown in Fig. 6.The curlometer technique (Dunlop et al., 1988) is used to determine these parameters.The j y component stayed positive while the j x component reversed sign from positive to negative at plasma flow reversal.Interestingly, there was a sign reversal of the y-and z-components of j ×B at flow reversal while the xcomponent did not show a clear reversal.This indicates that the magnetic stress in the y-and z-directions changed Ann.Geophys  at flow reversal but not in the x-direction.The current dissipation/dynamo term j y E y was quite variable, indicating the occasional presence of the dynamo effect (negative value) in addition to dissipation (positive value).The electric field component used in this term is the average over the four spacecraft.Furthermore, this quantity is computed in the GSE coordinate system since only two components of the electric field were measured on the GSE xy-plane.In this way, we avoid making any assumption on the unmeasured component to infer the three components of the electric field.
The computed value of div B during this period typically lies within ±5 nT/10 3 km and exceeds ±10 nT/10 3 km only for a few data points (Runov et al., 2003).The above characteristics concerning current-associated parameters would be unchanged even if points with div B exceeding ±5 nT/10 3 km were removed.These characteristics are thus not affected by potential inaccuracies in the current density computation.

Electric field signatures during the plasma flow reversal
Another feature is the observed electric field, which is expected to have a duskward component near an X-line and the highest value closest to the neutral sheet.However, Fig. 7 shows the E y component (the most reliable component from EFW) was intermittently dawnward (negative), especially at C3, which was generally the closest satellite to the neutral sheet among the four as judged by the magnitude of the B x component.For instance, during two intervals (09:54:25-09:54:40 and 09:55:10-09:55:20 UT), the electric field was large at C1 where the B x component was large (i.e., far from neutral sheet) while it was small at C3 where B x component was small (i.e., close to the neutral sheet).

Summary
This study presents detailed examination of a plasma flow reversal event seen by Cluster in the magnetotail during substorm expansion when simultaneous global viewing of auroral activity is available from IMAGE.The analysis includes examination of the Lorentz force j ×B and the current dissipative/dynamo term j y E y associated with the plasma flow reversal.These two parameters have not been examined by previous reports of plasma flow reversals.This plasma flow reversal, from tailward to earthward, was accompanied by southward magnetic field component reversing to northward and occurred during the late phase of substorm expansion.Close examination shows several interesting features: (1) there was no clear sign of the quadrupole magnetic perturbations from a Hall current system organized by the flow reversal time; (2) the x-component of the Lorentz force did not change sign at plasma flow reversal although the other two components did; (3) the timing analysis of plasma flow reversal among the Cluster satellites was inconsistent with tailward movement of a plasma flow source; (4) the electric field occasionally was dawnward, opposite to the current density direction and indicative of a dynamo action occasionally; and (5) the electric field occasionally had a larger value far from the neutral sheet than close to it.

Comparison with NENL model predictions
There are two prominent substorm models, namely, the NENL and CD models, that depict magnetotail disturbances at the late stage of substorm expansion.In the NENL model, plasma flow reversal signifies the tailward retreat of an Xline.More specifically, this model predicts the following signatures: (1) detection of the quadrupole magnetic perturvations from a Hall current system organized by the plasma

Comparison with CD model predictions
In the CD model, plasma flow reversal is produced by development of multiple current disruption sites at progressively further downstream distance in the magnetotail.Current disruption is crudely speaking equivalent to generation of a dawnward current.This can produce southward magnetic field tailward of the current disruption site when the ambient magnetic field prior to activity is weak (Lui, 2001)  organized by the flow reversal time from this scenario.The x-component of the Lorentz force is highly dependent on the turbulent distributions of the current density and the B z component resulting from the nearby current disruption sites and is thus not expected to have an orderly pattern.The change in the y-and z-components of the Lorentz force suggests that the dominant activity site shifted from duskward to dawnward of Cluster and from below to above Cluster.This shift is entirely consistent with the expected dawnward expansion of substorm activity.Indeed, the auroral activity seen by IMAGE/FUV indicates that a new activity site developed at ∼09:53 UT to the east of the existing activity site.The new activity site intensified at ∼09:55 UT (Fig. 2g) and the old activity site faded away at ∼09:57 UT (Fig. 2h).In addition, from the ion velocity distribution measured by CIS/HIA on C3, there was another ion population flowing duskward in the interval (09:54-09:55 UT) when the plasma flow reversed (note the positive V y during this time in Fig. 4b).This feature is also consistent with another plasma flow source developed to the dawnside of Cluster.Furthermore, the centroid of the Cluster tetrahedron moved from south of the neutral sheet to north of it, consistent with the expected change in the z-component of Lorentz force that points toward the neutral sheet.The changing direction and the variable strength of the electric and magnetic fields are expected from turbulence arising from current disruption (Lui et al., 1988), which can give rise to dawnward electric field and dynamo effect, distinct from magnetic reconnection where only duskward electric field and dissipation are expected.Current disruption in the two-dimensional particle simulation of a thin current sheet exhibits current filamentation and fluctuating electric fields as well as current sheet bifurcation (Figs. 31 and 32 in Lui, 2004).The instability is tentatively identified as the cross-field current instability (Lui et al., 1991;Yoon et al., 2002).The simulation result is reproduced here in Fig. 8.Further tests of the CD substorm model and the NENL model will be made by the upcoming NASA mission THEMIS.

3.4
Comparison with an X-line encounter reported previously For comparison purposes, we have also examined the observations of flow reversal event documented by Runov et al. (2003).They have demonstrated well the existence of the quadrupole magnetic perturbations for their event.We have found that the z-component of the Lorentz force in their event was the largest component and was directed toward the neutral sheet, similar to the event reported here.However, the x-component of the Lorentz force was directed earthward during tailward flow and reversed to tailward directed during part of the earthward flow interval.This behavior of the xcomponent of the Lorentz force is opposite to the expected directions for the reconnection configuration.In addition, the timing sequence of plasma flow reversal from tailward retreat of an X-line based on the X GSM location of satellites  Fig. 8. Evolution of current density and the current dissipation/dynamo term in 2-1/2D particle simulation of current disruption (Lui, 2004).
Each panel shows a yz-plane cut of the current sheet.As time progresses, the initial relatively uniform current density breaks up, exhibiting both dissipation (j y E y >0) and dynamo effect (j y E y <0).
should be C3-C1-C4 while the observed sequence is C3-C4-C1.Furthermore, the electric field was directed dawnward intermittently during the flow reversal interval accompanied by negative j y E y term, i.  Lui et al. (1978), to relate the E y component with the electric field normal to separatrix surface in the reconnection plane.In their analysis, the E y component plays a pivotal role in particle acceleration and thus should be associated with dissipation.The occurrence of negative j y E y is therefore unexpected.Therefore, not all the observed features in the 1 October 2001 event are consistent with tailward retreat of an X-line.

Concluding remarks
It is unclear how general the above findings are for plasma flow reversals.More events will be examined in the future.However, at the very least, this report demonstrates that for this event, plasma flow reversal from tailward to earthward accompanied by magnetic field reversing from southward to northward, is consistent with multiple current disruption sites dominating a given observation point as envisioned by the current disruption substorm model.

Fig. 1 .
Fig. 1.The AU (upper) and AL (lower) indices on 22 August 2001 and the relative positions of four Cluster satellites in the XY-and YZ-planes in GSM coordinates.

Fig. 5 .
Fig. 5.The paarameter B x B y /|B x | to check for the signature of the quadrupole magnetic perturbation from the Hall current system associated with magnetic reconnection.The arrow below each trace indicates the plasma flow reversal time for that satellite.

Fig. 6 .
Fig. 6.Plasma parameters associated with current density during the plasma flow reversal.The unit for j ×B components is nT-nA/m 2 .

Fig. 7 .
Fig. 7. Comparison between the simultaneous measurements of E y and B x from all four satellites during the plasma flow reversal.
e. dynamo effect.The electric field occasionally became larger at the high latitude of the plasma sheet than near the neutral sheet, as reported by Cattell et al. (2005) and Wygant et al. (2005) for this event.Cattell et al. (2005) used the burst mode data to propose the detection of electron holes near the plasma sheet boundary as a feature consistent with magnetic reconnection.Wygant et al. (2005) considered the tilting of the current sheet in the yz-plane, a wavy profile of the current sheet first discovered by