Location and Timing of Magnetic Reconnections in Earth's Magnetotail: Accomplishments of the 29‐Year Geotail Near‐Earth Magnetotail Survey

The spacecraft Geotail surveyed the near‐Earth plasma sheet from XGSM = −10 to −31 RE and YGSM = −20 to +20 RE during the period from 1994 to 2022. It observed 243 magnetic reconnection events and 785 tailward flow events under various solar wind conditions during plasma sheet residence time of over 23,000 hr. Magnetic reconnections associated with the onset of magnetospheric substorms occur mostly in the range XGSM = −23 to −31 RE. When the solar wind is intense and high substorm activities continue, magnetic reconnection can occur closer to the Earth. The YGSM locations of magnetic reconnections depend on the solar wind conditions and on previous substorm activity. Under normal solar wind conditions, magnetic reconnection occurs preferentially in the pre‐midnight plasma sheet. Under conditions with intense (weak) solar wind energy input, however, magnetic reconnection can occur in the post‐midnight (duskside) plasma sheet. Continuous substorm activity tends to shift the magnetic reconnection site duskward. The plasma sheet thinning proceeds faster under intense solar wind conditions, and the loading process that provides the preconditions for magnetic reconnection becomes shorter. When magnetic flux piles up during a prolonged period with a strongly northward‐oriented interplanetary magnetic field (IMF) Bz, the time necessary to provide the preconditions for magnetic reconnection becomes longer. Although the solar wind conditions are the primary factors that control the location and timing of magnetic reconnections, the plasma sheet conditions created by preceding substorm activity or the strongly northward IMF Bz can modify the solar wind control.


Introduction
The most dynamic phenomena in Earth's magnetotail are produced by magnetic reconnection in the near-Earth magnetotail.The magnetic field configuration changes prior to and after the onset of magnetic reconnection in the magnetotail, and highly dynamic and high-speed plasma flows are generated by magnetic reconnection in the plasma sheet.The magnetosphere and the ionosphere are strongly coupled with field-aligned currents (electric currents along the magnetic field lines).Intense and dynamic auroral activity starts on the ground, and intense electric currents flow in the ionosphere and create disturbances in the magnetic field on the ground.This entire phenomenon is called a magnetospheric substorm (or simply a "substorm").The onset of a substorm is traditionally defined by a breakup of intense auroral activity on the ground, and the active period after the onset called the "expansion phase" lasts for 30-40 min.Hence, conditions for magnetic reconnection in the magnetotail are formed prior to the onset.When the interplanetary magnetic fields (IMFs) become oriented southward (antiparallel to the magnetic field in the dayside magnetosphere), magnetic reconnection occurs on the dayside.The reconnected field lines are transported tailward with the solar wind flow, and accumulate in the tail.This process is called the "loading process," and it corresponds to the growth phase of a substorm, as determined from ground-level magnetic variations.The loading process typically lasts for 40-min, although its duration is highly variable.Magnetic reconnection apparently can start at any time during the loading process.There is no evident threshold for the onset of magnetic reconnection, and the loading process can continue even after the onset of magnetic reconnection.Magnetic reconnection in the magnetotail proceeds explosively, with an X-line geometry of the magnetic field.Magnetic reconnection is the "unloading process," in which the accumulated magnetic field energy is converted efficiently into kinetic and thermal energies of the plasma.This unloading process creates fast tailward plasma outflows with the southward magnetic field (Bz < 0) tailward of the X-line and fast earthward plasma flows with the northward magnetic field (Bz > 0) earthward of the X-line in the plasma sheet; the typical flow speed is the Alfvén velocity (∼500 km s −1 ).The unloading process provides the energy for various dynamic The Geotail spacecraft was launched on 24 July 1992 (Nishida, 1994).After probing the distant tail during the first two-and-a-half years, Geotail started to survey the near-Earth magnetotail at radial distances of 10-31 R E in October 1994.The main objective of the Geotail mission was to explore dynamical processes in the magnetosphere, especially magnetic reconnection in the magnetotail, at the magnetohydrodynamics (MHD) level.Geotail made the first in situ observations of magnetic reconnection in a fully three-dimensional (3D) mode on 27 January 1996 (Nagai et al., 1998b).These comprehensive plasma measurements including 3D distribution functions (Mukai et al., 1994) and fine-resolution magnetic field measurements (Kokubun et al., 1994) have enabled us to study the ion-electron decoupling process in magnetic reconnection below the MHD level, including Hall Physics and inflowing Hall electrons (Nagai et al., 1998b(Nagai et al., , 2001)), and electron dynamics manifested by electron outflow jets and electron current layers (Nagai et al., 2011(Nagai et al., , 2013b).These findings have been confirmed by recent, more-sophisticated measurements from MMS (e.g., Torbert et al., 2018).In this paper, we investigate the location and timing of magnetic reconnections in the magnetotail from the macroscopic point of view.We do not discuss the details of each magnetic reconnection event, since they have been presented previously in the review by Nagai (2021).Because of its orbital motion, Geotail did not stay the plasma sheet continuously.Furthermore, one of the data recorders on Geotail failed on 25 December 2012, so that we lack approximately 2 hr of data each day after this date.The other data recorder failed on 28 June 2022, so that only the limited data transmitted directly to the Japanese ground station have been available after this date.Although the instruments on Geotail continued to work properly, operation of the spacecraft was terminated on 28 November 2022.Other data gaps occurred due to various instrument problems.Nevertheless, the almost-continuous long-term Geotail observations have enabled us to derive robust results from various studies.
Hints about the answer to the question "where and when does magnetic reconnection occur in the tail?" have been provided by some recent studies.On the basis of in situ observations of magnetic reconnection in the magnetotail with simultaneous magnetic field measurements at geosynchronous altitude and on ground level, Nagai and Shinohara (2021) have shown that the dawn-dusk extent of the magnetic reconnection site, which is equivalent to the length of the X-line, is limited to be approximately 5 R E .The magnetic reconnection site corresponds to the upward (from the ionosphere to the magnetosphere) field-aligned current part of the substorm current wedge.Hence, the magnetic reconnection site occupies only a part of the dawn-dusk extent of the magnetotail (the full width is 40 R E in the mid-tail).Hence, it is natural to ask "where does magnetic reconnection occur in the tail?."The pioneer work by Hones (1979) using early spacecraft  showed that tailward flows with negative Bz were observable over almost the full dawn-dusk extent of the mid-tail at radial distances >30 R E .Plasmoids and traveling compression regions were observed everywhere in the distant tail beyond 100 R E (Hones et al., 1984;Ieda et al., 1998;Moldwin & Hughes, 1992;Slavin et al., 1984).These studies may give the impression that the X-line extends from the dusk magnetopause all the way to the dawn magnetopause.However, recent spacecraft observations do not support such a great length of the X-line for magnetic reconnection in the near-Earth plasma sheet, although they only provide indirect evidence.A duskside preference is found in various reconnection-related-phenomena (e.g., Walsh et al., 2014).Multi-satellite studies have shown that fast earthward flows in the near-Earth magnetotail appear to have a finite dawn-dusk width (e.g., R. Nakamura et al., 2004), although the fast earthward flows observed beyond a radial distance of 10 R E may not be related to the onset of substorm activity (e.g., Ohtani et al., 2006) but instead may occur in the recovery phase of a substorm.One event study has suggested that the flux rope does not extend azimuthally across the magnetotail at the lunar distance (Kiehas et al., 2013).

10.1029/2023JA032023
3 of 31 Nagai and Shinohara (2022) have found that solar wind conditions have evident effects on the onset meridian of substorms.They studied the onset magnetic local time (MLT) meridians on the basis of ground magnetic field data and proton injection data at geosynchronous altitude.Since their study examined only 41 magnetic reconnection events having well-defined footpoints on the ground, they did not fully analyze the locations of magnetic reconnection events in the magnetotail.Their analyses of well-isolated substorms (Figure 8 of Nagai & Shinohara, 2022) showed that intense solar wind conditions result in shorter durations of the growth phase.Long-term observations from Geotail have provided numerous magnetic reconnection events and tailward flow events under various solar wind conditions.Recently, especially since 2015, various spacecraft observations have provided unprecedently precise information on the structure of the solar wind and the onsets of substorms.In this paper, we use these data to pursue the answer to the question "where and when does magnetic reconnection occur in the tail?." The remainder of this paper is organized as follows: Section 2 describes the data sets used in this study.Section 3 describes statistical results for the locations of magnetic reconnection events in the geocentric solar-magnetospheric (GSM) x-y plane obtained from the Geotail survey in 1994-2022.We also present the average variations of the magnetic field and plasma in the central plasma sheet for magnetic reconnection events.Section 4 describes the effects of solar wind conditions on magnetic reconnection events.Section 5 explores the relationships between the solar wind conditions and the durations necessary to create the preconditions for magnetic reconnection.Section 6 presents global pictures of the plasma sheet under different solar wind conditions.Section 7 discusses the significance of the present results and makes some remarks, and Section 8 presents our conclusions.

Data
We used the data obtained by the Geotail spacecraft during the period from October 1994 to November 2022.The apogee of Geotail was around 50 R E in the near-tail seasons in late 1994 and early 1995, and it then moved down to 30 R E .The perigee of Geotail was approximately 10 R E during the entire period.Magnetic field data were obtained with the magnetic field experiment MGF (Kokubun et al., 1994), and ion and electron data were obtained with the low-energy plasma experiment LEP (Mukai et al., 1994).Full energy-time spectrograms for the ions and electrons from LEP, which are the most fundamental data used to identify magnetic reconnections, can be found on the website of Institute of Space and Astronautical Science (ISAS).The Geotail data are presented in the GSM coordinate system.We used the solar wind data obtained by the ACE spacecraft (after 1998) in this study.In addition, we used the OMNI 1-min data (obtained mainly by the Wind spacecraft) to examine any changes in the dynamic pressure of the solar wind and to exclude any possible ambiguity in the solar wind travel time from the L1 point to the Earth.To examine the solar wind conditions in the near-Earth environment, we also examined data from the Geotail, MMS, and Time History of Events and Macroscale Interactions During Substorms (THEMIS) missions.
To identify the onset of dipolarization in the magnetic field at geosynchronous altitude, we used the magnetic field data from the GOES satellites-mostly GOES-13, GOES-14, and GOES-15 during 2015-2019, and GOES-16 and GOES-17 during 2020-2022.Magnetic field data with a time-resolution of 0.512 s or 0.1 s are given in the VDH coordinate system, where H (pointing northward) is antiparallel to the Earth's dipole axis, D (azimuthal east) is orthogonal to H and to the radius vector to the satellite, and V (nearly radial outward) completes the Cartesian coordinate system.When necessary, we also used energetic electron and proton data from GOES-13, GOES-14, and GOES-15.More detailed information about these electron and proton measurements is available from Nagai et al. (2019).In addition, we used the energetic electron (>200 keV) fluxes observed by the geosynchronous meteorological spacecraft  to monitor electron injections and particle trapping boundary motions (Walker et al., 1976) to identify substorm onsets.We also used the GOES-8, GOES-9, GOES-10, GOES-11, and GOES-12 1-min magnetic field data before 2015.We employed the geomagnetic indices AU and AL (from Kyoto University before 2014 and from SuperMAG after 2015).We also used individual magnetic field data from the ground stations provided by Kyoto University.

Geotail Near-Earth Plasma Sheet Survey
Figure 1 shows the hours of data sampling by Geotail in the plasma sheet during the period 1994-2022.Geotail surveyed the plasma sheet almost equally at radial distances between 10 and 30 R E inside its apogee.The plasma sheet is defined as the region where the plasma β (the ratio of plasma pressure to magnetic pressure) is >0.1 and the ion temperature is >0.1 kV.We exclude the magnetosheath plasmas by inspecting the ion and electron energy-time spectrograms.The region beyond a radial distance of 31 R E was seldom surveyed, since the apogee of Geotail was approximately 31 R E after 1995, and it decreased during this observing period.Inside X GSM = −20 R E , the magnetopause was located near Y GSM = +20 R E on the duskside and near Y GSM = −18 R E on the dawnside, respectively.The Geotail orbits were almost symmetric relative to the Y GSM = 0 axis over the long term, so that there was no significant dawn-dusk asymmetry in the plasma sheet sampling.The instruments on Geotail occasionally stopped near the midnight meridian due to eclipses when the spacecraft entered the Earth's shadow.Plasma data could not be used when solar proton fluxes were high.In addition, there were periods during which data were not available because of instrumental problems or troubles with data transmission.However, these effects did not bias the observations toward any particular site in the near-Earth plasma sheet.The amount of plasma sheet data available totaled 23,896.3hr at a time resolution of 12 s.
It is important to note that the tail observations were obtained during different seasons over the observing span 1994-2022.The magnetotail season of Geotail occurred during winter in 1996, during spring in 2003, during summer in 2010, and during fall in 2021.We ignored any dipole title effects, if they exist.

Magnetic Reconnection Events
In situ observations of magnetic reconnections in the magnetotail from Geotail have been presented in previous studies (e.g., Nagai, 2021;Nagai et al., 1998bNagai et al., , 2001Nagai et al., , 2011Nagai et al., , 2013aNagai et al., , 2013bNagai et al., , 2015a)).The most convincing signature of a magnetic reconnection is electron heating during a flow reversal accompanied by a simultaneous Bz reversal (e.g., a tailward flow with negative Bz becomes an earthward flow with positive Bz).Electron heating can be detected from the ratio of the 5 keV electron flux to the 1 keV electron flux, and we adopt the criterion that this ratio be ≥0.1, which is the same as the criterion used by Nagai et al. (2015b).We obtained 93 such events during 1994-2022, and a list of the events during 1994-2014 is presented by Nagai et al. (2015b).In most events, we observe ion-electron decoupling (e.g., the electron flow speed is faster than the ion flow speed) and/or an electron current sheet (high-speed dawnward flowing electrons) near the neutral sheet.Since this number of events (93) is too small for any statistical study, we have also selected possible magnetic reconnection events.Nagai et al. (2013aNagai et al. ( , 2013b) ) performed superposed epoch analyses of 30 magnetic reconnection events.They have found that the flow reversal from −500 to +500 km s −1 occurs within a span of 2 min and that the plasma characteristics (density and temperature) of subsequent earthward flows are not different from those of the preceding tailward flows.For the present analyses, we first selected fast tailward flows (Vx < −300 km s −1 ) accompanied by negative Bz in the central plasma sheet, using the procedure adopted by Nagai et al. (2015b).We examined the data in the plasma sheet beyond Y GSM = −20 R E and beyond Y GSM = +20 R E .This yielded more than 1,000 tailward flow events, including the magnetic reconnection events.Second, for each event we examined a plot of magnetic field and plasma moment data to determine whether or not tailward flows with negative Bz were followed within 5 min by earthward flows with positive Bz.For some events, Geotail exited from the plasma sheet and entered briefly into the tail lobe, and we then detected earthward flows inside the plasma sheet.Finally, we examined the plasma characteristics using the energy-time spectrograms for both ions and electrons.We also found weak electron heating events inside the plasma sheet.We identified as a possible magnetic reconnection event one in which a flow reversal occurred within 5 min.In this way, we obtained 150 possible magnetic reconnection events.We consider the other 785 events to be tailward flow events.
Figure 2 shows the locations of 93 events identified as magnetic reconnection events with in situ observations (dots), and those of 150 possible magnetic reconnection events (open circles).The number of events per 1 R E y bin and that per 1 R E x bin are presented in the histograms to right and bottom of the main figure, respectively.Most of the in situ observations of magnetic reconnections were obtained inside the plasma sheet from X GSM = −20 to −31 R E (88%, 82 out of 93 events) and Y GSM = −8 to +14 R E (94%, 87 out of 93 events).The distribution of the possible magnetic reconnection events in the GSM x-y plane agrees reasonably well with that of the in situ observation events.Hereafter, we consider all 243 (93 + 150) of these events as magnetic reconnection events.The distribution of all these magnetic reconnection events from Geotail is quite similar to that obtained from MMS observations using different criteria (e.g., Hubbert et al., 2022;Rogers et al., 2023).
Figure 3 shows the locations of the 785 tailward flow events (open circles) and those of the 243 magnetic reconnection events (dots).There are two groups of tailward flow events.In one group of 618 events, the plasma sheet returned to be quasi-stationary after the tailward flow period, and the characteristics of the ions and electrons returned to their pre-event conditions (e.g., the 27 January 1996, event discussed by Nagai et al., 1998b).In other group of 170 events, Geotail entered into the tail lobe just after the tailward flow period.We did not discriminate between these two groups, since we did not find any significant differences in their distributions.The distribution of tailward flow events overlaps well with that of the magnetic reconnection events in the range Y GSM = −15 to +15 R E , where magnetic reconnections are frequently observed.
Figure 3 also shows the average tailward flow velocity vector for each Y GSM = 5 R E bin.We calculated the average velocity vectors for Vx < −300 km s −1 and negative Bz within the 30 min period.In the range Y GSM = −10 to +10 R E , where magnetic reconnections occur frequently, the magnitude of Vy is small.Hence, tailward flow events can be used as signatures of magnetic reconnection events at almost the same Y GSM position.Magnetic reconnections occur rarely in the far dawnside plasma sheet with Y GSM < −8 R E .Tailward flows have a significant negative (dawnward) Vy component in this region.Magnetic reconnections occur less frequently in the far duskside plasma sheet with Y GSM > +14 R E .Tailward flows have a significant positive (duskward) Vy component in this region.Hall MHD simulations (e.g., T. K. M. Nakamura et al., 2012) show that the outflow has a slight positive (duskward) component.It is possible that the X-line in the far duskside and far dawnside plasma sheet may be inclined relative to the y-axis.Several tailward flow events in which the magnitude of Vy was larger than that of Vx occurred near the duskside magnetopause (Y GSM > +20 R E ), although the magnetic field structure was not deformed.It is more likely that tailward flows in the far-duskside plasma sheet originate from magnetic reconnections that occur in the smaller Y GSM meridian of the plasma sheet.
Figure 4 shows the occurrence of magnetic reconnections during the years 1994-2022.This period corresponds to solar cycles 23 and 24 and the beginning of solar cycle 25.The IMF magnitude Bt, the solar wind speed, and geomagnetic activities measured by Kp are also plotted.Since the orbital plane of Geotail was inclined relative to the GSM x-y plane, its residence time in the plasma sheet changed year-by-year.In the beginning of Geotail operations in 1995-1999, the apogee of Geotail was controlled to stay within the plasma sheet during the tail seasons, so that the residence time in the plasma sheet was large and nearly constant.The Geotail residence time had dawnside preference in 2001-2005 and duskside preference in 2011-2015.Furthermore, solar protons arrive in near-Earth space during periods of high solar activity, and the plasma instrument could not operate then, for examples, during big magnetic storms.The paucity of events in 2002 may be caused by these effects.However, we judge that the instrumental turn-off did not produce any significant bias in the statistics used for the entire period of 1994-2022.
In the period from 1995 to 2006, Geotail stayed in the plasma sheet for 7,753.95hr, and it observed 189 magnetic reconnection events (0.024/hr) and 706 (magnetic reconnection + tailward flow) events (0.091/hr).The average Kp was 2.13.In the period from 2011 to 2017, Geotail remained within the plasma sheet for 3,434.13hr, and it The histograms of the number of events per GSM y-bin (the thick line represents the number of magnetic reconnection events).(c) The histograms of the number of events per GSM x-bin (where again the thick line represents the number of magnetic reconnection events).observed 38 magnetic reconnection events (0.011/hr) and 236 events (0.069/ hr).The average Kp was 1.74.The difference in event occurrence during two solar cycles is probably caused by the level of substorm activities.Furthermore, it is found that 942 events (out of 1,028 events) were taken in these two periods.Although several factors may influence the results, the magnetic reconnection events used in the present study were obtained during periods of high solar and geomagnetic activity.
Figure 5 shows the distributions of magnetic reconnection events and tailward flow events in X GSM (top) and Y GSM (bottom).Here, the distributions are restricted to selected areas in order to avoid any effects from the no-event areas.Histograms of the events collected from the whole area are presented in Figures 2 and 3. To obtain the X GSM dependence, we used the 211 (787) events in the range Y GSM = −8 to +14 R E .The plasma sheet residence times have local maxima near both the perigee (∼10 R E ) and the apogee (∼30 R E ), as shown in Figure 1.This same tendency can be seen for any given Y GSM .Similarly, to obtain the Y GSM dependence, we used the 186 (830) events in the range X GSM = −23 to −31 R E .The plasma sheet residence times have broad local maxima around both Y GSM = +13 R E and Y GSM = −10 R E (see Figure 1).This same tendency can be found throughout the region X GSM < −20 R E .The plasma sheet residence time has a broad maximum centered on the Y GSM = 0 meridian inside X GSM > −15 R E .However, these variations are rather smooth.Hence, the characteristics seen in "number of events" are very similar to those seen in "occurrence of events."Magnetic reconnections occur throughout the region X GSM < −15 R E , and the occurrence of magnetic reconnections becomes high around X GSM = −23 R E and appears to saturate near X GSM = −30 R E .Indeed, tailward flows were almost always found inside 30 R E in association with the onset of substorms when Geotail stayed inside the plasma sheet (not near the dawn flank).These characteristics do not change significantly for any selection of the Y GSM range (see also Figures 2 and 3).Magnetic reconnections rarely occur in the near-Earth magnetotail with X GSM > −15 R E .The occurrence of tailward flow events increases almost monotonically as X GSM decreases.This X GSM dependence of the tailward flow events is caused by the cumulative effect of magnetic reconnections.These results from the tailward flow events support the results from magnetic reconnection events.
During a big magnetic storm (Dst was −316 nT and AL exceeded −2,000 nT), Geotail detected tailward convection flows (calculated from E × B, where E is the measured electric field and B is the measured magnetic field) with negative Bz at (−8.60, −0.74, −0.88 R E ) at 21:04 UT on 30 October 2003 (Nagai, 2006).Unfortunately, the LEP ion data were contaminated by intense solar protons and correct moment data (flow velocities) were not available.The electron energy-time spectrogram, the data of which were also contaminated, shows electron heating signatures.Among all the Geotail observations, this event at X GSM = −8.60R E was probably the magnetic reconnection event closest to the Earth (this event is not included in the present study).The Polar spacecraft (with an apogee of 9 R E ) did not find any magnetic reconnections in the near-Earth magnetotail during 2001-2003 (Ge & Russell, 2006).Hence, magnetic reconnection seldom occurs inside X GSM = −10 R E , however, it is possible that events occurring close to the Earth may be hidden.
The occurrence of magnetic reconnections peaks in the range Y GSM = +4 to +8 R E , and broad wings extend from Y GSM = +14 to Y GSM = −8 R E with a duskside preference.This tendency can be seen in the x-y distribution of magnetic reconnection events in Figure 2, where the number of events is presented in each 1 R E bin.The occurrence of tailward flow events also has a broad peak.The occurrence of tailward flows is quite similar to that of magnetic reconnections in the range Y GSM = +14 to −8 R E , indicating that tailward flow events can be used as a proxy for the Y GSM dependence of magnetic reconnection events.The occurrence of tailward flows is approximately three times higher than that of magnetic reconnections.On the far dawnside with Y GSM < −10 R E The histograms of the number of events per GSM y-bin (the thick line corresponds to the number of magnetic reconnection events).(d) The histograms of the number of events per GSM x-bin (where the thick line again corresponds to the number of magnetic reconnection events).and on the far duskside with Y GSM > +14 R E , the occurrence of magnetic reconnections is reduced to 0.1 of that of tailward flows.As indicated in Figure 3, since the tailward flow vector has a duskward component on the far duskside and a dawnward component on the far dawnside, it is likely that tailward flows originate from magnetic reconnections that occur near the central meridian of the plasma sheet.

Average Magnetic Field and Plasma Variations for Magnetic Reconnections in the Central Plasma Sheet
As described below, we constructed the average magnetic field and plasma variations for magnetic reconnections.We used the 582 magnetic reconnection and tailward flow events observed in the central plasma sheet from Middle: (e) Numbers of tailward flow events (with magnetic reconnection events represented by thick lines) in the full dawn-dusk tail, (f) those in the duskside tail, and (g) those in the dawnside tail for each year.Bottom: (h) Total hours of Geotail plasma sheet residence time in the full dawn-dusk tail, (i) those in the duskside tail, and (j) those in the dawnside tail for each year.
X GSM = −23 to −31 R E and Y GSM = −10 to +10 R E .The central plasma sheet is defined as the region with −10 nT < Bx < +10 nT.However, this criterion for the magnetic field did not change any of the results, except for the values of Bx and By. Figure 6 shows a superposed epoch analysis over the period from −120 to +120 min.The zero epoch is the start of tailward flows with negative Bz.
Several works have already investigated signatures of the growth phase in the plasma sheet, and the results appear to be well established.However, only a few results have been reported beyond 20 R E .Using Geotail data at the radial distances 20-50 R E , Nagai et al. (1997) found an evident increase in the total pressure (magnetic pressure + plasma pressure) during the growth phase, which is caused mainly by an increase in the density.Shukhtina et al. (2014) also used Geotail data, at radial distances from 15 to 32 R E, and they found an increase in the total pressure and a decrease in Bz during the growth phase.Inside a radial distance of 20 R E , the pressure increase is easily detected, along with a decrease in Bz indicating a change in the magnetic field configuration (e.g., Yushkov et al., 2021).It is well-known that the development of a more taillike configuration of the magnetic field, which corresponds to a decrease in Bz, is a representative growth phase signature at the geosynchronous altitude of 6.6 R E (e.g., Nagai, 1982).Hence, we presented the results in the central plasma sheet relevant to the investigations in this study.
Figure 6 confirms that an increase in the total pressure and a decrease in Bz prior to the zero epoch are major characteristics of the growth phase in the central plasma sheet.Although we have carried out various analyses of plasma flows, we have not found any evident characteristics in the plasma flow data before the zero epoch.Since there is dawn-dusk asymmetry in Vy, there are two data plots for Vy in Figure 6, one for Vy on the dawnside (Y GSM = −10 to 0 R E ) and one for Vy on the duskside (Y GSM = 0 to +10 R E ).The velocity Vy is always positive on the duskside, while Vy is almost zero and can become negative on the dawnside.These Vy patterns can be seen in any conditions (see Section 6).Although clear-cut magnetic reconnection events have been presented in past studies (e.g., Nagai, 2021;Nagai et al., 2013b;Nagai & Shinohara, 2021), no evident patterns emerged from the plasma flow data.While the flow direction is highly variable, the x-direction flow becomes earthward in the long-term average (the y-direction flow depends on Y GSM ; see Section 6).It is possible that no significant changes occur in the flow pattern during the growth phase.Irregular variations in Bx and By near epoch zero are caused by our selection criterion of −10 nT < Bx < +10 nT, since these variations are not seen without imposing any criteria for Bx.Note that the magnetic field in the southern hemisphere is reversed in the averaging.
Figure 7 shows the behavior of Bz and of the total pressure at nine X GSM locations.The events were obtained from the range Y GSM = −10 to +10 R E for X GSM > −32 R E and from the range Y GSM = −20 to +20 R E for X GSM < −32 R E .Note that the vertical scale for the range X GSM = −10 to −20 R E is different from those for the others.The events at X GSM < −32 R E were newly selected from the interval from November 1994 to February 1995, when the apogee of Geotail was 50 R E .We adopted the events in the range Y GSM = −20 to +20 R E , using the criteria discussed in Section 3.1.Geotail stayed in the plasma sheet for 417.3 hr during this period and observed 45 tailward flow events (without any in situ reconnection signatures).A decrease in Bz prior to epoch zero is less evident for X GSM < −30 R E and an increase in the total pressure is less evident for X GSM < −32 R E .Furthermore, a spike in the total pressure at the zero epoch becomes evident for X GSM < −30 R E .This spike is produced by an increase in the density (at almost constant temperature) at the heads of the tailward flows, as shown in Figure 6.The density increase is caused by compression of pre-existing plasma sheet plasmas by fast flows.However, a pile-up of Bz, which is one evident signature of a plasmoid (e.g., Hones et al., 1984), is not created, probably because the observation point is close to the source region.The initial distant tail survey (Nagai et al., 1994) has shown that only a southward dip of Bz is common inside 60 R E and a bipolar magnetic structure becomes typical beyond 60 R E .These features of Bz and the total pressure provide additional evident that the occurrence of magnetic reconnection saturates near X GSM = −30 R E , as shown in Section 3.2.We found no magnetic reconnection events in the range X GSM = −31 to −50 R E according to our criteria (electron heating), although the survey was limited to 1994 and early 1995.

Examinations of Solar Wind Conditions
Nagai and Shinohara (2022) have shown that the solar wind energy input, VBs (expressed as −Vx × Bs, where Vx is the x component of the solar wind velocity, and Bs is the southward component of the IMF Bz), and the IMF Bz behavior are the major factors that control the onset locations of magnetic reconnections in the magnetotail.Since the Geotail events in the magnetotail were not fully studied in their work, we examined here the solar wind conditions for the magnetic reconnection and tailward flow events selected in Section 3. We used the Geotail events for the period 1998-2022, for which ACE data were available.We calculated the propagation time as the ACE geocentric solar ecliptic (GSE) x value divided by the daily averaged solar wind GSE Vx value (Nagai et al., 2005;Nagai & Shinohara, 2022).We also examined the OMNI data (mostly provided from Wind observations).However, we could not perform any detailed quality checks for the ACE data, since in most cases solar wind data were not available in the near-Earth environment.Hence, there are ambiguities in the solar wind data themselves, and we therefore show only the average variations of the solar wind conditions.Since we examined the solar wind conditions for different event groups (magnetic reconnection events, tailward flow events, isolated events) separately, the consistency can be tested, and we discuss the major findings from our analyses.

Factors That Control the Magnetic Reconnection Site in Y GSM
As in the study by Nagai and Shinohara (2022), we here examine the solar wind conditions for three groups of magnetic reconnection events in the Y GSM range where the occurrence of magnetic reconnections is high (Section 3): the duskside plasma sheet from Y GSM = +8 to +14 R E , the premidnight plasma sheet from Y GSM = 0 to +8 R E , and the dawnside plasma sheet from Y GSM = −8 to 0 R E .We selected events in the range X GSM = −23 to −31 R E to avoid any X GSM effects.These groupings are based on the results shown in Figure 5, and slight changes in the groupings make no significant differences.Nagai and Shinohara (2022) used the following four groups selected in terms of MLT: 19-22 MLT, 22-23 MLT, 23-24 MLT, and 00-04(03) MLT.The footpoint of Geotail around Y GSM = +8 R E was located near the 22 MLT meridian and that around Y GSM = −8 R E was located near the 02 MLT meridian (Nagai & Shinohara, 2022).In the present study, we combined the central two MLT groups in order to obtain sufficient number of events.Figure 8 shows the average variations of the IMF Bz, the solar wind energy input VBs, and the auroral electrojet index AL for the magnetic reconnection events (157 events in 1998-2022) for the period from −240 to +240 min around epoch zero.Here, the zero epoch is the time when Geotail detected tailward flows leading to electron heating (the signature of magnetic reconnection) or a flow reversal.Figure 9 shows the results for tailward flow events (464 events, not including the magnetic reconnection events).As expected, the AL index clearly shows the onset of substorms at the zero epoch in both figures.We can recognize next three characteristics, which are found in Nagai and Shinohara (2022), in Figures 8 and 9: 1.Strong solar wind energy input occurs before the onset in the Y GSM = −8 to 0 R E (dawnside) group.2. The solar wind energy input before the onset becomes weaker in the positive Y GSM range (pre-midnight).3. The IMF Bz is continuously southward in the larger Y GSM range (duskside), resulting in continuous substorm activity.
In addition, we examined isolated events in which the average values of AL for 1 hr and for 2 hr prior to epoch zero did not exceed −100 nT.We presume that no onset occurred during the 2 hr prior to the selected event.Using these criteria, we obtained 126 magnetic reconnection and tailward flow events in the region from X GSM = −23 to −31 R E and Y GSM = −8 to +14 R E .The results of the three Y GSM groups for the period from −360 to +120 min around epoch zero are presented in Figure 10.According to these event selection criteria, the AL activity declined during the 2-hr period, and VBs became almost zero for 1 hr.The VBs value in the range Y GSM = −8 to 0 R E is only slightly larger than that in the range Y GSM = 0 to 8 R E .However, we note that the IMF Bz became significantly negative just prior to epoch zero for the events in the range Y GSM = −8 to 0 R E .For the events in the range Y GSM = +14 to +8 R E , the VBs value becomes small and the average IMF Bz is positive even prior to epoch zero.Furthermore, substorm activities are high prior to the quiet period in the range Y GSM = +8 to +14 R E .These results support the three findings described previously.

Factors That Control the Magnetic Reconnection Site in X GSM
We examine 124 magnetic reconnection events in the plasma sheet from Y GSM = 0 to +12 R E to minimize the Y GSM dependence.Since the events close to the Earth (inside X GSM = −17 R E ) are investigated in Section 7, we examine the solar wind conditions for three groups in the range X GSM = −17 R E to −31 R E .The first group includes the events in the range X GSM = −17 R E to −23 R E , where magnetic reconnection occurs relatively close to the Earth.Since the occurrence of magnetic reconnections beyond X GSM = −23 R E is almost constant, we divided the events in this region into two groups, simply to get equal event numbers.The results are presented in Figure 11.Magnetic reconnections occur closer to the Earth as the solar wind energy input increases.It is likely that the strong southward IMF Bz prior to the onset mainly contributes an increase in the solar wind energy input.This result is consistent with the previous study on the basis of the limited period 1995-2003 (the solar minimum and the solar maximum of cycle 23) by Nagai et al. (2005).

Methodology
Fortunately, there were several cases during 2020-2022 in which a northward-oriented IMF Bz continued and then executed a simple southward turning (see Figure 12).We searched for step-like southward turnings of Bz in the solar wind with the ACE data for 2015-2022 and for which Geotail observed the magnetic field structure detected by ACE in the solar wind.First, we selected the events from the ACE data for which the average VBs was 0.1 mV m −1 for 1 hr and then a sudden southward turning of the IMF Bz occurred.The southward-oriented IMF Bz continued for more than 40 min.Second, we required that Geotail be in either the solar wind or the magnetosheath, and we examined whether or not the magnetic field structures at the two points (the locations of ACE and Geotail) were reasonably matched in Bx, By, Bz, and Bt.We examined the data from MMS (the apogee was less than 30 R E ) for some events when Geotail data were not available.We also checked some events for which data from all three spacecraft were available.When the ACE and Geotail data matched, the MMS data usually matched perfectly as well.We determined the onset of magnetic reconnection from onset of dipolarization in the magnetic field at geosynchronous altitude (Nagai & Shinohara, 2021).This onset signature at geosynchronous altitude is the most reliable tool, since any pseudo-onsets as well as the major onsets of substorms can be detected easily (e.g., Nagai, 1982;R. Nakamura et al., 1994).We obtained 61 cases in 2015-2022 (38 cases in 2020-2022).

The 10 April 2022, Event
Figure 12 shows an event observed on 10 April 2022.We time-shifted the solar wind data in this figure to represent the data at X GSM = 0 R E .In this event, Geotail was located upstream at X GSM = +27 R E , and we shifted its data by 6.2-min using the solar wind velocity from ACE.For this event, the 54.9-min time shift of the ACE data using only its X GSE location and the daily average solar wind velocity did not require any correction for data overlap.The time shift of the ACE data to match the Geotail data is usually less than 5 min in other cases.The magnetic structures at both locations (ACE and Geotail) matched perfectly until 04:40 UT.We time-shifted all the data to represent the data at X GSM = 0 R E for simplicity, since we do not know adequately the time delays for the effect of the IMF on the evolution of the magnetotail.The OMNI IMF Bz data (from Wind, not presented here) are a few minutes advanced.Since the OMNI data represent the solar wind conditions around X GSM = +10 R E , we expect them to be advanced by approximately 2-min.The MMS data were also available, and they matched the ACE and Geotail data perfectly for this event.
The first dipolarization in the magnetic field occurred at 04:00 UT at GOES-17.Although the data are plotted as 1-min average values, data with 0.1-s values clearly showed a sudden onset in the H and V components, with rapid fluctuations, as typical signatures for the onset (e.g., Nagai et al., 2019).Furthermore, GOES-16 detected a deflection in the D component (not shown here), indicating the development of field-aligned currents.We examined the 0.512-s magnetic field data for all onsets.We also examined the ground-level magnetic field data and the AU and AL indices.In this event, another large dipolarization occurred around 04:30 UT, with a mid-latitude positive bay at US magnetic stations on the ground.In this study, we took the time of onset to be that associated with the first dipolarization on the nightside at geosynchronous altitude.Because of the localization of onset signatures at geosynchronous altitude (e.g., Nagai, 1982), we have approximately a 5-min ambiguity in the onset time.Although it is possible to have a maximum 10-min ambiguity, this does not affect any of the conclusions in this study.

Statistical Studies
Figure 13 shows the IMF Bz, the solar wind energy input VBs, and the AU and AL indices for the five groups of the selected 61 events.Each group is defined by its average VBs value for the initial 40 min after the southward turning of the IMF Bz, (approximately, 1, 2, 3, 4, and >5 mV m −1 ; see (c) Ave VBs of Figure 13).In some events, the VBs profiles exhibit positive spikes, even when the 1-hr average VBs was less than 0.1 mV m −1 .In these events, the VBs values reached their average values within the initial 10 min, and in most cases they remained around the average values and did not return to zero.We examined many similar events in the ACE data.Further stringent limitations appear not to be needed for statistical studies, if we do not include events in which there were fluctuations around the zero level after the IMF Bz became southward.We use these 61 events for the first statistical study.The final results show that the average VBs value in a certain given time period cannot alone determine the growth phase interval (from the southward turning of the IMF Bz to the onset of the substorm).
Figure 14 shows the relationship between the growth phase interval and VBs.In this plot, we used the 40-min average VBs values.We found no significant differences when we used other time spans (e.g., 20, 30, or 60-min).It is known that northward turning of the IMF Bz sometime triggers the onset of substorm activity (e.g., Rostoker, 1983).It is also possible that some solar wind changes can trigger an onset (e.g., McPherron, 2023).For 44 events (the dots in Figure 14), the IMF Bz continued to be oriented southward, 12 events occurred with Bz turning northward (the circles in Figure 14), and five events had some other changes around the onset (also shown by circles in Figure 14).In this study, we did not find any differences among these groups.We found a linear trend for the 56 data points with VBs in the range 0.0-4.0 mV m −1 , with a minimum time delay of approximately 30 min.The correlation coefficient is 0.60 for this limited data set.This is the trend expected on the basis of the study by Nagai and Shinohara (2022).Although there are only five data points with VBs > 4.0 mV m −1 , they give the somewhat surprising result that the data points do not have short time delays.
In order to check this unexpected result for the large VBs results, we selected 38 events during 1998-2014 with 40-min average values of VBs > 3.0 mV m −1 .For this study, we had to use 1-min GOES data since only these low time-resolution data were available.We also used a clearly isolated positive bay events observed by the Japanese magnetic stations Kakioka, Memambetsu, and Kanoya.Figure 15 shows the relationship between the growth phase interval and VBs for the combined 38 + 5 data points.Although it is possible that the error for each individual point may be larger, the time delay clearly can become large even for the same VBs value.
We also examined the solar wind density and the solar wind dynamic pressure for the events in Figure 15.The results for the solar wind density are presented by the three groups (<7.0, 7.0-15.0,and >15.0/cc) in Figure 15.We used the 40-min average solar wind data values.The results for the dynamic pressure are very similar, since the dynamic pressure is almost proportional to the solar wind density in most cases.The solar wind density and dynamic pressure are generally large when the magnitude of IMF (VBs) is large.There are no evident relationships between the solar wind density (dynamic pressure) and the growth phase interval, as clearly seen in the events in the range VBs 3.0-4.0mV m −1 .
We examined the solar wind conditions for seven shorter events and four longer events.The results are presented in Figure 16.In general, a large VBs event (a large southward-oriented Bz event) was preceded by a period with a large northward-oriented Bz, since the IMF magnitude did not change significantly in almost all cases.From these data, we found that a large time-delay event was preceded by a prolonged period with a large northward-oriented, relatively constant, IMF Bz.For a smaller time-delay event, the IMF Bz gradually turned southward before the zero crossing, and the IMF Bz was variable during the period of the northward-oriented IMF Bz.
We also examined the temporal behavior of the IMF Bz for the events with smaller VBs. Figure 17 shows the results from the events in Figure 14 with VBs values in the range 0.5-2.0 mV m −1 .There are 13 data sets for events with time delays of 40-70 min and 10 data sets for those with time delays of 80-110 min.These two groups have prolonged periods with a northward-oriented IMF Bz, and the average IMF Bz is 2.0 nT for the group with the shorter time delay and 3.4 nT for the group with the longer time delay.Hence, the existence of a northward-oriented IMF Bz prior to the IMF Bz turning southward is another factor in determining the growth phase interval.For the event selection, we require that the events should have a period with small VBs (less than 0.1 mV m −1 ) prior to a clear southward turning of the IMF Bz.Hence, we did not examine any events for which the IMF Bz was highly variable, since it was difficult to select reasonable variables for deducing causality.

Average Magnetic Field and Plasma Characteristics
It is not simple to derive the basic state of the central plasma sheet prior to the start of plasma sheet thinning; in other words, prior to the growth phase of a substorm.We have therefore constructed the mean structure of the magnetic field and the plasma in the central plasma sheet under the following conditions: the IMF Bz is almost 10.1029/2023JA032023 20 of 31 zero, VBs is less than 0.5 mV m −1 , the magnitude of Vx in the plasma sheet is small, and the AE values are small.The results are essentially the same, except for the magnitude of Vx.When there is no restriction on Vx, the Vx values become large (positive).Hence, we show the results under the restrictive conditions that −150 km s −1 < Vx < +150 km s −1 and −10 nT < Bx < +10 nT.Note that in this study the criterion for a fast tailward flows (in Section 3) is Vx < −300 km s −1 .When the IMF Bz remains oriented northward for a prolonged period, the plasma sheet conditions change significantly; we therefore discuss this case separately below.
Figure 18 shows the magnetic field vectors (Bx, By) projected onto the GSM x-y equatorial plane.Since we imposed the condition −10 nT < Bx < +10 nT, the magnitude of the magnetic field can be neglected.In addition, we have inverted the signs of Bx and By in the southern hemisphere.The magnetic structure is almost symmetric relative to the Y GSM = 0 axis, which is the reason why we adopted the GSM coordinate system, not including aberration, in this study.Furthermore, the footpoints of magnetic reconnection events near the meridian Y GSM = 0 used in this study mapped near the midnight meridian (Nagai & Shinohara, 2021).The density, ion temperature, and total pressure have the symmetrical distributions relative to the Y GSM = 0 axis (not shown here, since these parameters seem to be unimportant in this study).Figure 19 shows the plasma (ion) flow vectors (Vx, Vy) projected onto the GSM x-y equatorial plane.Since these data were obtained under the condition −10 nT < Bx < +10 nT, the flow can be considered to be a convection flow (the flow is perpendicular to the local magnetic field).The Geotail plasma data had a time resolution of 12-s and the magnetic field direction near the neutral sheet can change within one plasma moment sampling.The calculated values of Vperp (the flow velocity perpendicular to 12-s averaged magnetic field) therefore may not be physically correct, and accordingly they were not used, even though they were practically the same as the raw values of Vx and Vy.The average value of Vx is positive (earthward) in the full magnetotail, and it has a small local minimum around Y GSM = 0 R E .
The value of Vy is positive (duskward) on the duskside and mostly negative (dawnward) on the dawnside (Y GSM < −5 R E ).We find this Vy pattern to persist even when fast flows were included.In the region X GSM < −20 R E , the value of Vy increased rapidly in the range Y GSM = 0 to +5 R E , and it had a broad local maximum at Y GSM = +5 to +10 R E .
Figure 20 shows the structure of Bz.The value of Bz on the duskside is generally smaller than that on the dawnside.This Bz pattern indicates that the magnetic flux crossing the equatorial plane is small and almost constant in the duskside region X GSM < −24 R E .If we assume that convection is enhanced by a southward IMF Bz, the magnetic flux would be transported duskward from the premidnight region.Thinning the plasma sheet is most likely to proceed in the range Y GSM = 0 to +5 R E , since there is a steep gradient in the Vy pattern there.

Response of the Magnetotail to the Southward Turning of the IMF Bz
To investigate the effects of the intensity of solar wind energy input, it is desirable to identify the growth phase interval.The period during which the magnetic field becomes more taillike and the particle fluxes drop at geosynchronous altitude may be a good candidate for this interval.Unfortunately, we could find only a few events when Geotail stayed within the plasma sheet.Reliable data from geosynchronous altitude were obtained only after 2015, but Geotail did not remain in the central plasma sheet for reasonably long periods after 2015.It is almost impossible to use ground-level magnetic field or AU/AL indices to determine the growth phase interval, as they often miss the onsets of substorms.In this study, we therefore selected the time intervals just after the southward turning of the IMF Bz using the ACE data.We used the selection criteria that the 1-hr average VBs prior to a clear southward turning of Bz was less than 0.1 mV m −1 and that the 1-hr average VBs then became larger than 0.4 mV m −1 .We also required that Geotail be located in the magnetotail; that is, from X GSM = −20 to −31 R E and Y GSM = −8 to +8 R E .We were able to select 89 intervals during 1998-2021.Unfortunately, Geotail did not stay continuously in the central plasma sheet (−10 nT < Bx < +10 nT); rather, it stayed mostly in the outer plasma sheet and/or in the tail lobe.Hence, we present only the variations of the behavior of Bz and of the total pressure (magnetic pressure + plasma pressure).Other parameters did not appear to provide any useful information.
During the growth phase, the total pressure increases and Bz decreases inside the plasma sheet even at radial distances >20 R E (Section 3.3).
To explore the effects of the VBs intensity, in Figure 21 we show the results for VBs = 0.4-1.2mV m −1 and for VBs = 1.2-2.0mV m −1 .The magnetic field Bz and the total pressure on the duskside (Y GSM = 0 to +8 R E ) and on the dawnside (Y GSM = −8 to 0 R E ) are represented in two ways: raw data sets and as level-adjusted data, with the values adjusted to those at epoch zero.A decrease in Bz indicates a more taillike change of the magnetic field configuration (e.g., Fairfield & Ness, 1970), which indicates the progress of plasma sheet thinning.An increase in the total pressure indicates the accumulation of magnetic flux in the tail (e.g., Shukhtina et al., 2014).It is possible that this process may proceed uniformly in the whole magnetotail, since the total pressure distribution is almost symmetric relative to the Y GSM = 0 axis.For smaller VBs (the top panel), a decrease in Bz and an increase in the total pressure are more evident on the duskside.The response appears to be time delayed.Later, the total pressure decreases first on the duskside in association with the onset of substorms.For the larger VBs (the bottom panel), a decrease in Bz starts immediately on the dawnside.The total pressure increases parallel on the duskside and on the dawnside.Later the total pressure decreases first on the dawnside in association with the onset of substorms.These characteristics are rather robust, and they did not change even when we changed the selection criteria for grouping the data.It is likely that any changes induced by the southward turning of the IMF Bz shift dawnward as the VBs intensity increases.

The State of the Plasma Sheet During the Prolonged Period With a Northward-Oriented IMF Bz
The results in Section 5.3 imply that conditions in the plasma sheet are different for the long time-delay group and the short time-delay group.It is well known that a northward-oriented IMF Bz controls the plasma sheet conditions (e.g., Wang et al., 2006).We therefore examined the data of the magnetic field and the plasma (ion) moments in the central plasma sheet (−10 nT < Bx < +10 nT) during 1998-2022.We selected time intervals during which the IMF Bz continued to be oriented northward for more than 1 hr.The northward-oriented IMF Bz occasionally continued for more than 10 hr.However, we found no time evolution in the plasma conditions in our data sets, since it was not possible to distinguish between spatial variations due to the orbital motion of Geotail and temporal variations.We therefore show only the average magnetic field and plasma parameters in the magnetotail in the range from X GSM = −25 to −30 R E .
Figure 22 shows the spatial variations of the average Bx, By, Bz, ion density, and ion temperature in the central plasma sheet for four groups.Three of these groups have the following average values of the IMF Bz > 1.0, >3.0, and >5.0 nT.For comparison, we also collected the data when the flow speed was small (−100 km s −1 < Vx < 100 km s −1 ), which correspond to plasma sheet conditions during non-substorm ("quiet") intervals.Since we have imposed the criterion (−10 nT < Bx < +10 nT) for the central plasma sheet, the values of Bx and By are very similar in these four groups.The most striking feature appears in Bz; the magnitude of Bz increases over the entire range of Y GSM as the IMF Bz increases.This indicates that magnetic fluxes are accumulating in the mid-tail plasma sheet during the period with the northward-oriented IMF Bz.The plasma density increases for the northward IMF Bz.This characteristic is particularly evident on the dawnside.The increase in plasma density is less evident for the smaller northward-oriented IMF Bz on the duskside.Although the plasma temperature decreases, and this behavior does not exhibit any dependence on the magnitude of the IMF Bz.The singular behaviors of the ion density and temperature at Y GSM = +10 R E for the IMF Bz > 5 nT group were greatly affected during the 27 September 2016, event which had an interval with a northward-oriented IMF Bz (+10 nT) just after a strong southward-oriented IMF Bz and intense substorm activity.

Discussion and Remarks
Magnetic reconnection is most likely caused by plasma instabilities in the thinned current sheet embedded in the central plasma sheet, although the ultimate causes of the onset of magnetic reconnection have not yet been identified.It is therefore desirable to determine where and when the thinned current sheet is formed in the plasma sheet.The comprehensive plasma sheet observations from Geotail in the region from X GSM = −10 to −31 R E and Y GSM = −20 to +20 R E have revealed the most preferred locations for magnetic reconnections in the magnetotail under various solar wind conditions.They have also enabled us to identify the factors that determine the time duration required to create the preconditions necessary for magnetic reconnection to occur.The major findings of this study, some of which were discovered in previous studies (e.g., Nagai et al., 1998bNagai et al., , 2005;;Nagai & Shinohara, 2022), are as follows: 1. Magnetic reconnection occurs mostly in the range X GSM = −23 to −30 R E .The occurrence of magnetic reconnections is likely to be reduced at X GSM < −30 R E .2. Magnetic reconnection can occur closer to the Earth than X GSM = −20 R E under conditions of intense solar energy input during continuously high substorm activity.3.Under normal solar wind conditions, magnetic reconnection occurs most frequently in the pre-midnight sector of the plasma sheet from Y GSM = 0 to +8 R E .This region corresponds to the sector with 22-24 MLT. 4. Intense solar wind energy input conditions tend to shift the magnetic reconnection site dawnward, so that magnetic reconnection can occur in the post-midnight sector of the plasma sheet from Y GSM = −8 to 0 R E . 5. Weak solar wind energy input conditions tend to shift the magnetic reconnection site duskward.Continuous substorm activity appears to amplify this effect.6.The time duration required to prepare the preconditions necessary for magnetic reconnection is controlled primarily by solar wind energy input.Intense solar wind energy input can create the preconditions for magnetic reconnection during a shorter time interval.7. Intense solar wind energy input requires a strong southward-oriented IMF Bz, and it is often preceded by a period with a strong northward-oriented IMF Bz, since the magnetic field magnitude (Bt) is usually almost constant in the solar wind.During the periods with a strong northward-oriented IMF Bz, magnetic flux piles up in the near-Earth plasma sheet, enhancing Bz in the plasma sheet.This can totally change the preconditions in the plasma sheet and delay the onset of magnetic reconnection.Since the conditions in the plasma sheet, especially those in the central plasma sheet, cannot be specified for the Geotail observations prior to and just after the southward turning of the IMF Bz, we cannot unambiguously deduce any definite causal relationship.Even with Geotail observations extending over 29 years, it is difficult to collect adequate samplings of the plasma sheet for the period of an "ideal IMF Bz time history."The average pictures of the plasma sheet constructed in Section 6.1 nevertheless can help us to investigate mechanisms for creating the preconditions that result in magnetic reconnection.However, since it is possible that the average pictures of the plasma sheet may be caused by frequent occurrences of magnetic reconnection at any particular site in the plasma sheet, the causality of magnetic reconnection may not be addressed correctly.
The Y GSM dependence of the magnetic reconnection site is likely produced by the basic convection structure in the plasma sheet.The magnetic field structure in Bx and By is basically symmetric relative to the Y GSM = 0 axis.The magnetic flux crossing the equatorial plane, which is represented by Bz, decreases as the radial distance increases.An asymmetry in Bz becomes evident at X GSM < −25 R E , and Bz is reduced in the pre-midnight sector (Figure 20).The velocity of the convection flow is fast in the pre-midnight sector, while the direction of that flow is rather random in the post-midnight sector.A gradient in the velocity field Vy exists in the pre-midnight sector.If the convection deduced from the southward-oriented IMF Bz is enhanced even uniformly in the plasma sheet, duskward transport of the magnetic flux can proceed rapidly from the pre-midnight sector, thinning the plasma sheet there.The intensity of solar wind energy input is likely to control the convection pattern.Weak solar wind energy input enhances convection in the duskside sector of the plasma sheet, while intense solar wind energy input can enhance convection efficiently near the midnight tail.Hence, the intense solar wind conditions lead to an occurrence in the post-midnight sector of the plasma sheet, but they do not exclude any occurrence of magnetic reconnection in the pre-midnight sector of the plasma sheet.This possibility should be pursued further using plasma flow data for the central plasma sheet.Unfortunately, it appears to be difficult to analyze this topic further with the present Geotail data.Intense solar wind energy input can efficiently create the preconditions for magnetic reconnection in the plasma sheet, so that the duration of the growth phase (loading process) can become shorter.
When a strong, northward-oriented IMF Bz continues for a long time, that is, for an hour or more, conditions in the plasma sheet conditions are changed significantly, and magnetic flux is piled up in the region where magnetic reconnection forms preferentially under the normal conditions (Figure 22).However, even when the magnetic flux piles up, the local minimum of Bz in the magnetic field is located in the pre-midnight sector.The results for isolated events indicate that the Y GSM dependence of the magnetic reconnection site on the intensity of solar wind energy input does not change significantly.Hence, the effect of a northward-oriented IMF Bz simply changes the duration of the growth phase and requires a longer period of time for the thinning of the plasma sheet.This may give the impression that the duration of the growth phase is highly variable and irregular.
It is conceivable that the plasma sheet density may affect the preconditions for magnetic reconnection.As shown in Figure 22, the density in the plasma sheet increases, when a strong, northward-oriented IMF Bz continues for a long time.However, we note that the plasma density did not increase significantly in the plasma sheet when the density of the solar wind was not high (roughly <10 cm −3 ) even for a large northward-oriented IMF Bz.The events with the large VBs (>3 mV m −1 ) values in Figure 16, the solar wind density was higher than 10/cc for 3 events and less than 10/cc for 4 events for the shorter time delay events, and it was higher than 10/cc for 2 events and less than 10/cc for 2 events.It is likely that the role of the density in the plasma sheet might be minor.In the 582 magnetic reconnection and tailward flow events from X GSM = −23 to −31 R E and Y GSM = −10 to +10 R E (Figure 6), the average plasma sheet density prior to the epoch zero was less than 0.2/cc for the 346 events and less than 0.4/cc for the 484 events, respectively.Only the five events had the plasma density higher than 1.0/cc.Indeed, there were only 87 days on which the density became higher than 1.0/cc in the same region of the plasma sheet during the period from 1994 to 2022, and 53 days occurred in 1995-2000.Hence, the roles of the density and the temperature in the plasma sheet would be investigated using adequate data sets.
It is not easy to explore mechanisms for the X GSM dependence of the magnetic reconnection site.The earthward convection speed increases only slightly from X GSM = −25 to −31 R E , and because of the paucity of data sampling, reliable convection speed data were not available in the Geotail data beyond X GSM = −31 R E in the Geotail data.The duskward convection speed in the premidnight plasma sheet has a slight local maximum around Y GSM = + 8 R E at X GSM = −21 to −25 R E (Figure 19).It is possible that magnetic flux may be transported most efficiently from the premidnight sector beyond X GSM = −25 R E .It is conceivable that the convection pattern may shift earthward under conditions of intense solar wind energy input.Furthermore, magnetic reconnection close to the Earth tends to occur after continuous substorm activity.Hence, it is also conceivable that continuous substorm activity can change the basic structure of the plasma sheet.
On the basis of Geotail observations during the period from November 1994 to July 1996, Nagai et al. (1998b) reported that tailward flows generated by magnetic reconnection were observed mostly in the pre-midnight sector of the plasma sheet from X GSM = −20 to −30 R E , (see also Nagai & Machida, 1998).These data were obtained under conditions of less intense solar wind energy input.Here, we present results for the Y GSM dependence of the occurrence of magnetic reconnections from two different data sets, one obtained in 1995-1997 and the other in 1998-2000 (Figure 23).The occurrence of magnetic reconnections was reduced significantly in the post-midnight sector (Y GSM < 0 R E ) during 1995-1997, while magnetic reconnection occurred even in the post-midnight sector during 1998-2000.The average VBs value for a 40-min period prior to the onset was 0.71 mV m −1 for the events in 1995-1997 and 1.10 mV m −1 for the events in 1998-2000.The average magnetic field magnitude Bt and solar wind speed V were 4.81 nT and 410.22 km s −1 , respectively, during 1995-1997, and were 7.01 nT and 436.93 km s −1 , respectively, during 1998-2000 (see Figure 4).Although this illustrates the importance of solar wind energy input, we caution that statistics cannot be based on such short, limited period of solar activity.Nagai et al. (2005) reported that the sites of magnetic reconnections are primarily controlled by solar wind energy input.The results from the plasma sheet of X GSM = −17 to −31 R E (Figure 11) support this conclusion.However, the magnetic reconnection events inside X GSM = −17 R E indicate that previous substorm activities may provide conditions for the onset of magnetic reconnection.There are 13 magnetic reconnection events (including only tailward flows with negative Bz) in the range X GSM = −10 to −15 R E and 17 events in the range X GSM = −15 to −17 R E .Figure 24 presents solar wind conditions and substorm activities for these events.Since the events inside X GSM = −15 R E occurred in the range Y GSM = −6 to +12 R E , the same Y GSM range are used for other two groups.The average AL index exceeded −400 nT prior to the onset for the events in the range X GSM = −10 to −15 R E .In the range X GSM = −17 to −20 R E , 29 events (out of 54 events) occurred after relatively smaller substorm activity (e.g., Nagai et al., 1998a).However, it should be noted the substorm activity prior the epoch zero in the range X GSM = −17 to −23 R E is higher than those in the range X GSM < −23 R E (see Figure 11).Furthermore, inside the radial distance of 15 R E , magnetic reconnections (tailward flow events) appear to occur in both the dusk sector and the dawn sector, but not near the midnight meridian (Figures 2 and 3).A map of the average Bz shows a local maximum near the midnight meridian (Figure 20).The occurrence of magnetic reconnections close to the Earth seems to be affected by the structure of Bz in the near-Earth plasma sheet.It is likely that the magnetic reconnection site is not particularly close to the Earth even during an extremely large substorm when it occurs just after a quiet period.Prolonged and/or intense substorm activities appear to make conditions in the plasma sheet favorable to the onset of magnetic reconnection, although their effects are not easily evaluated.
It would be interesting to consider how the magnetic reconnection site is selected under given solar wind conditions on the basis of the previous considerations.Figure 25 shows the average solar wind and AL conditions for six plasma sheet locations.Since the occurrence of magnetic reconnections increases until X GSM = −23 R E , and it remains constant beyond X GSM = −23 R E , we therefore consider the two X GSM groups NEAR and FAR, with the boundary between them located at X GSM = −23 R E .Based on the results in Section 3, we also consider the following three Y GSM groups: DUSK (Y GSM = +14 to +8 R E ), PREmidnight (Y GSM = +8 to 0 R E ), and DAWN (Y GSM = 0 to −8 R E ).Among the three FAR groups and the three NEAR groups, the Y GSM location is selected according to the intensity of the solar wind energy input VBs (or the magnitude of the southward-oriented IMF Bz) just prior to the onset of magnetic reconnection, as discussed in Section 3. When we compare the NEAR groups to the FAR groups, the X GSM location is selected according the intensity of the solar wind energy input VBs (or the magnitude of the southward IMF Bz) just prior to the onset, as discussed in Section 3. Furthermore, the VBs values are larger for the NEAR-PRE group and the NEAR-DAWN group than they are for the FAR-DAWN group.The NEAR-DUSK group is somewhat problematic.When substorm activity continues and solar wind energy input becomes large, the magnetic reconnection site is located in either the PRE or the DAWN group.We note that the value of AL was the lowest before the event in the NEAR-DUSK group.This probably occurred because continuous substorm activity changed the plasma sheet conditions.Hence, with the intensity of the solar wind energy input and the previous substorm activity history, we could predict the most likely site of magnetic reconnection in the magnetotail, however, this prediction cannot be quantitatively justified and it cannot rule out any possibility of occurrence of magnetic reconnection in any different sites.

Conclusions
The Geotail spacecraft surveyed the near-Earth plasma sheet form X GSM = −10 to −31 R E and Y GSM = −20 to +20 R E for the period from 1994 to 2022.It observed 243 magnetic reconnection events and 785 tailward flow events under various solar wind conditions over two solar cycles in its the plasma sheet residence time of 23,896.3hr.Under normal solar wind conditions, magnetic reconnection occurs in association with the onset of substorms in the range X GSM = −23 to −31 R E .Under intense solar wind conditions and continuous substorm activity, magnetic reconnection can occur closer to the Earth than X GSM = −20 R E .The location of magnetic reconnection in Y GSM has a clear dependence on the solar wind and on previous substorm activity.Magnetic reconnection occurs in the pre-midnight sector of the plasma sheet in the usual solar wind energy input conditions.During conditions of intense solar wind energy input, magnetic reconnection can occur in the post-midnight sector of the plasma sheet.As the solar wind energy input weakens, the location of the magnetic reconnection site shifts to the duskside plasma sheet.When the IMF Bz is continuously oriented southward and substorm activity continues, the location of the magnetic reconnection site shifts farther duskward.It is likely that intense (weak) solar wind conditions efficiently thin the plasma sheet in the central (duskside) magnetotail.Plasma sheet thinning proceeds faster under intense solar wind conditions, and the loading process that creates the preconditions for magnetic reconnection becomes shorter.The plasma sheet conditions are greatly affected by the time history of solar wind variations, and magnetic flux accumulates in the mid-tail during the period with the northward-oriented IMF Bz period.Probably even the intense convection induced by the southward IMF Bz cannot cause immediate plasma sheet thinning when the magnetic flux accumulates in the mid-tail plasma.The duration of the preconditioning, as well as the location of the magnetic reconnection site, cannot be determined uniquely with a single parameter from the solar wind.

Figure 1 .
Figure 1.Geotail plasma sheet residence times for each 1 R E × 1 R E box in the geocentric solar-magnetospheric (GSM) x-y plane.The radius of each circle represents the residence time in hours.

Figure 2 .
Figure 2. (a) The locations of magnetic reconnection events in the geocentric solar-magnetospheric (GSM) x-y plane (X GSM = −10 to −31 R E ) observed by Geotail during 1994-2022.A dot indicates a magnetic reconnection event identified by in situ observations, and an open circle indicates a possible magnetic reconnection event (see the text).The outer boundary represented by dashed lines indicates areas of the plasma sheet where Geotail stayed for ≥10 hr in each 1 R E × 1 R E box.(b)The histograms of the number of events per GSM y-bin (the thick line represents the number of magnetic reconnection events).(c) The histograms of the number of events per GSM x-bin (where again the thick line represents the number of magnetic reconnection events).

Figure 3 .
Figure 3. (a) The locations of tailward flow events (with negative Bz) in the geocentric solar-magnetospheric (GSM) x-y plane (X GSM = −10 to −31 R E ) observed by Geotail during 1994-2022.A dot indicates a magnetic reconnection event, and an open circle indicates a tailward flow event.The outer boundary represented by dashed lines indicates the area of the plasma sheet where Geotail stayed for ≥10 hr in each 1 R E × 1 R E box.(b) The average tailward flow velocity vectors for each 5 R E bin.(c)The histograms of the number of events per GSM y-bin (the thick line corresponds to the number of magnetic reconnection events).(d) The histograms of the number of events per GSM x-bin (where the thick line again corresponds to the number of magnetic reconnection events).

Figure 4 .
Figure 4. Top: (a) Monthly mean sunspot number, (b) interplanetary magnetic field (IMF) magnitude, (c) solar wind speed, and (d) geomagnetic index Kp over the period from 1994 to 2022.Middle: (e) Numbers of tailward flow events (with magnetic reconnection events represented by thick lines) in the full dawn-dusk tail, (f) those in the duskside tail, and (g) those in the dawnside tail for each year.Bottom: (h) Total hours of Geotail plasma sheet residence time in the full dawn-dusk tail, (i) those in the duskside tail, and (j) those in the dawnside tail for each year.

Figure 5 .
Figure 5. Top: (a) Distributions of the occurrence of tailward flow events (as fractions of the plasma sheet residence time in hours), (b) the number of tailward flow events, and (c) Geotail plasma sheet residence time in the range X GSM = −10 to −31 R E .Bottom: (d) Distributions of the occurrence of tailward flow events, (e) number of tailward flow events, and (f) Geotail plasma sheet residence time in the range Y GSM = −20 to +20 R E .The distributions of magnetic reconnections are also shown, using thick lines (scales at right).

Figure 6 .
Figure 6.The average magnetic field variations (a) Bx, (b) By, and (c) Bz, and plasma variations (d) Vx, (e) Vy, (f) density, (g) temperature, and (h) total pressure Pt, constructed from 585 tailward flow events (including magnetic reconnection events) in the central plasma sheet from X GSM = −23 to −31 R E and Y GSM = −10 to +10 R E .The velocity Vy at Y GSM = 0 to +10 is represented by a solid curve, while Vy at Y GSM = −10 to 0 R E is represented by a dashed curve.The growth phase signatures decreasing Bz and increasing total pressure are evident in the period from −30 to 0 min.

Figure 7 .
Figure 7.The magnetic field Bz and the total pressure in the central plasma sheet from X GSM = −10 to −40 R E .The number of events in each location is presented at the right corner of each right box.

Figure 8 .
Figure 8.(a) Variations of the average interplanetary magnetic field (IMF) Bz, (b) solar wind energy input VBs, and (c) auroral electrojet index AL for three geocentric solar-magnetospheric (GSM) y-range groups of magnetic reconnection events for the period from −240 to +240 min for the data set from 1998 to 2022.

Figure 9 .
Figure 9. (a) Variations of the average interplanetary magnetic field (IMF) Bz, (b) solar wind energy input VBs, and (c) auroral electrojet index AL for three geocentric solar-magnetospheric (GSM) y-range groups of tailward flow events for the period from −240 to +240 min for the data set during 1998-2022.

Figure 10 .
Figure 10.(a) Variations of the average interplanetary magnetic field (IMF) Bz, (b) solar wind energy input VBs, and (c) auroral electrojet index AL for three geocentric solar-magnetospheric (GSM) y-range groups of magnetic reconnection and tailward flow events for the period from −360 to +120 min for the data set during 1998-2022.

Figure 11 .
Figure 11.(a) Variations of the average interplanetary magnetic field (IMF) Bz, (b) solar wind energy input VBs, and (c) auroral electrojet index AL for three geocentric solar-magnetospheric (GSM) x-range groups of isolated magnetic reconnection and tailward flow events for the period from −360 to +120 min for the data set during 1998-2022.Note that the scale for (b) VBs is different from those in Figures 8-10.

Figure 12 .
Figure 12. (a)The solar wind energy input VBs from ACE, the interplanetary magnetic field (IMF) data (b) Bx, (c) By, (d) Bz, and (e) Bt from ACE (thick curves), and from Geotail (thin curves), the auroral electrojet indices (f) SMU and SML, and the magnetic field data (g) H and (h) V from GOES-16 (thin curves) and GOES-17 (thick curves) at geosynchronous altitude for the period from 01:00 UT to 06:00 UT on 10 April 2022.The vertical dashed line at 03:02 UT indicates the southward turning of the IMF Bz and the vertical dashed lined at 04:00 UT indicates the onset of dipolarization in the magnetic field at geosynchronous altitude, which corresponds to the onset of the substorm.The ACE data are shifted by 54.9 min and the Geotail data are shifted by 6.2 min (see text).

Figure 13 .
Figure 13.(a) Interplanetary magnetic field (IMF) Bz, (b) solar wind energy input VBs, (c) average VBs, and auroral electrojet indices AU (d) and AL (e) for five step-like VBs groups during the period from −180 to +120 min.Zero epoch corresponds to the southward turning of the IMF Bz.

Figure 14 .
Figure 14.Time delay of the onset of dipolarization at geosynchronous altitude from the southward turning of the interplanetary magnetic field (IMF) Bz as a function of the 40-min VBs value for the 61 events selected in 2015-2022.The correlation coefficient for all 61 events is −0.43, while the correlation coefficient for 56 events in the VBs range from 0.0 to 4.0 mV m −1 is −0.60.The dots indicate the events in which the IMF Bz does not change around the onset (continuously southward), while the open circles indicate the events in which either northward turning of the IMF Bz occurs around the onset or some other irregular change occurs.

Figure 15 .
Figure 15.Time delay of the onset of substorms from the southward turning of the interplanetary magnetic field (IMF) Bz as a function of the 40-min VBs value (>3.0 mV m −1 ) for events observed during 1998-2022.The dots indicate the events in which the solar wind density is larger than 15.0/cc, the double circles indicate the events in which it is 7.0-15.0/cc,and the open circles indicate the events in which it is less than 7.0/cc.

Figure 16 .
Figure 16.(a) Interplanetary magnetic field (IMF) Bz, (b) average IMF Bz, (c) solar wind energy input VBs, and (d, e) auroral electrojet indices AU and AL for seven events with time delays of 0-40 min (left panels) and four events with time delays of 70-120 min (right panels) from Figure 15 for the period from −240 to +240 min.Zero epoch corresponds to the southward turning of the IMF Bz.

Figure 17 .
Figure 17.(a) Interplanetary magnetic field (IMF) Bz, (b) average IMF Bz and Bt (thick curves), (c) solar wind energy input VBs, and (d, e) auroral electrojet indices AU and AL for seven events with time delays of 40-70 min (left panels) and four events with time delay of 80-110 min (right panels) from Figure 14 for the period from −240 to +240 min.Zero epoch corresponds to the southward turning of the IMF Bz.

Figure 18 .Figure 19 .
Figure 18.The average fields Bx and By in the central plasma sheet.

Figure 20 .
Figure 20.The average Bz field.The Bz fields with magnitudes of <2.5 nT are represented by thick circles.

Figure 21 .
Figure 21.Average values of the interplanetary magnetic field Bz (a, g) and of VBs (d, j) from the solar wind data.The variations of tail Bz (b, c, h, and, i) and of the total pressure (magnetic pressure + plasma pressure) in the magnetotail (e, f, k, and l) are shown for two VBs groups: VBs = 0.4-1.2mV m −1 and VBs = 1.2-2.0mV m −1 .Duskside events (27 events and 22 events) are represented by thin lines, while dawnside events (29 events and 11 events) are represented by thick lines (see the text).

Figure 22 .
Figure 22.Spatial variations of the average values of the magnetic field component (a) Bx, (b) By, and (c) Bz, (d) the ion density, and (e) the ion temperature in the central plasma sheet from X GSM = −25 to −30 R E and Y GSM = −17.5 to +17.5 R E .The data correspond to period in which the interplanetary magnetic field (IMF) Bz > +5 nT (indicated by "5"), IMF Bz > +3 nT ("3"), and IMF Bz > +1 nT ("1").Data obtained during quiet periods (see the text) are indicated by "q."

Figure 23 .
Figure 23.(a, d) Distributions of the occurrence of tailward flow events (as fraction of the plasma sheet residence time in hours), (b, e) the number of tailward flow events, and (c, f) the Geotail plasma sheet residence times in hours from Y GSM = −20 to +20 R E .The distributions of magnetic reconnections are represented by thick lines (scales at right).Top: Data from the period 1995 to 1997.Bottom: Data from the period 1998 to 2000.

Figure 24 .
Figure 24.(a) Variations of the average interplanetary magnetic field (IMF) Bz, (b) solar wind energy input VBs, and (c) auroral electrojet index AL for three geocentric solar-magnetospheric (GSM) x-range groups of magnetic reconnection events and tailward flow events for the period from −480 to +120 min.The events occurring at Y GSM = −6 to +12 R E are used.Note that the scales are different from those in Figures 8-11.