Ground-Space Observations of Pc5 Poleward Moving Auroral Arc Pulsations and Field-Line Resonances in the Post-Midnight Sector: THEMIS Observations

We investigate the Pc5 poleward moving auroral arc (PMAA) pulsations (~ 4–5 min period) using the ground-based all-sky imager network and the Time History of Events and Macroscale Interactions during Substorms (THEMIS) A, D, and E satellites, whose footprints were located near the PMAA in the post-midnight sector. The Pc5 PMAA pulsations considered herein occurred in conjunction with the enhancement of the magnetic and electric eld oscillations observed near the equatorial plane of the magnetosphere. The magnetospheric oscillation signal displayed three-cycle oscillations, which correspond primarily to the PMAA pulsations. The value of coherence between the magnetospheric oscillations and the luminosity pulsations was higher than 0.9. Based on these observations, it is suggested that the PMAA pulsations and the magnetospheric eld oscillations are initiated by the same physical mechanism and thus oscillate concurrently by the magnetosphere-ionosphere (M-I) coupling. The satellite data indicated a longer period of magnetospheric oscillations at the higher latitude site. On the other hand, the measured period of the PMAA pulsation was almost constant in the lower latitude region (~ 68.5°-70.0° MLAT), whereas in the higher latitude region (~ 70.0°-70.5° MLAT) it increased with increasing latitude. This signature demonstrates that the oscillations on the lower latitudinal side of the PMAA conformed with the monochromatic frequency eld-line resonance (FLR) where the oscillation period was constant and independent of latitude, whereas the higher latitude side of the PMAA presented a multi-frequency FLR region where the period lengthened with increasing latitude. The Pc5 magnetic pulsations observed on the ground neither exhibited a clear coincidence with the PMAA pulsations nor with the magnetospheric magnetic oscillations. On the other hand, the H component of magnetic pulsations demonstrated a rather similar behavior to that of the ion pressure variation within the magnetosphere. The solar wind speed was signicantly high, approximately 650 km/s, during this event. The magnetospheric magnetic and electric eld oscillations could be triggered simultaneously in a wide region by an impulse such as rapid convection changes caused by the sudden variations of the interplanetary magnetic eld (IMF) Bz, which was observed by the SuperDARN radar and the Geotail satellite.


Introduction
There are two types of auroral luminosity pulsations. The rst type is a short-period pulsation with the main period of a few to a few tens of seconds, which is called pulsating aurora (see the review by Lessard, 2012), and is mainly observed during the recovery phase of a substorm. The second type is a longer period pulsation in the Pc5 (150-600 s) period range, which is called long-period auroral pulsation.
The long-period auroral pulsations have two categories: one is arc aurora luminosity pulsation and the other is a patch aurora luminosity modulation. Previous studies reported that long-period auroral arc pulsations have a good correlation with Pc5 magnetic pulsations and their latitudinal amplitude and phase relation is consistent with the FLR models (Chen and Hasegawa, 1974;Southwood, 1974;Xu, et al., 1993;Samson et al., 1996Samson et al., , 2003Milan et al., 2001;Baddeley et al., 2017). Such evidence suggests that the auroral arc may be related to the eld-aligned current associated with the FLRs (Milan et al., 2001; Samson et al., 2003;Tanaka et al., 2012;Gillies et al., 2018). These physical processes are not well understood yet. On the other hand, the long period modulation of a diffuse patch aurora, including pulsating aurora is thought to be caused by the electron pitch-angle scattering with compressional mode Pc5 pulsations (Oguti et al., 1987;Yamamoto, 1988;Saka et al., 2014).
In this study, we examine the characteristics of arc type long-period auroral pulsations. Long-period auroral pulsations have been observed not only from the ground (Samson et al., 1996(Samson et al., , 2003 The space-ground coordinated observations using high spatial/temporal resolution data can provide important information concerning the long-period Pc5 auroral arc pulsations associated with FLR oscillations. However, to the best of our knowledge, there are no reports of such space-ground coordinated observations. Our motivation to study the long-period Pc5 auroral arc pulsation in the postmidnight sector stemmed from our former work. During our previous observations of pulsating auroras and omega bands (Sato et al., 2015 and 2017) using the dataset compiled by the THEMIS ground-based all-sky imager network, we found clear evidence of long-period Pc5 auroral arc pulsation events in the post-midnight sector at the higher latitude region of the pulsating aurora. Therefore, we focused our current study on the Pc5 auroral arc pulsations observed in the post-midnight sector.
The data were observed by the THEMIS ground-based all-sky imager network in this study. Using the data, we could pick up the fortunate events whereby the footprints of the THEMIS-A (TH-A), THEMIS-D (TH-D), and THEMIS-E (TH-E) spacecraft (using TS05 model [Tsyganenko & Sitnov, 2005]), whose orbits were located near the equatorial plane of the magnetosphere, traversed the eld of view of the all-sky imager and the Pc5 auroral arc pulsations were observed during the interval of 1210-1240 UT on 2nd March 2011.
In this study, we refer to the long-period auroral arc pulsations as "Pc5 poleward moving auroral arc (PMAA) pulsations," which were previously used by Kozlovsky and Kangas (2002) and Tanaka et al., (2012). They referred to it as PMAA to distinguish these phenomena from the poleward moving auroral forms (PMAFs) (Sandholt et al., 1990) that are observed in the dayside cusp region and are caused by the solar wind-magnetosphere reconnection. Here, we report on the observational evidence of the Pc5 PMAA pulsation signals and the magnetospheric magnetic and electric eld oscillations through the groundspace coordinated observations.

Instrumentation
In this study, we use optical all-sky imagers (ASIs) and magnetometers of the THEMIS ground-based observatories (GBO) [Angelopoulos, 2008;Mende et al., 2008]. We also use particles and eld data obtained onboard TH-A, TH-D, and TH-E spacecraft from the electric eld instrument , electrostatic analyzer [McFadden et al., 2008], and uxgate magnetometer .
The coordinated THEMIS/ASIs observations provided auroral images covering broad latitude and longitude (~ 1000 km) ranges with high spatial (~ 1 km near zenith) and temporal (3 s) resolutions. The white light imagers cover a wide wavelength band of ~ 400-700 nm, and the images were projected onto an ionospheric altitude of 110 km. We also use ground magnetometer data with a 0.5 s time resolution.
To examine the morphological and dynamic signals of the shapes and motions of Pc5 PMAA pulsations, we produced movie les using the original 3-s-resolution ASIs data.

− 1. Selection of Pc5 PMAA pulsation event
To identify "Pc5 PMAA pulsation" events in this study, we rst examined a summary plot of the THEMIS data (for example, that of the March 2, 2011, <http://themis.igpp.ucla.edu/summary.php? year=2011&month=03&day=02&hour=0024&sumType=asi&type=keograms>). Our selection method is based on qualitative visual inspection of the images. Pc5 PMAA pulsation-like events can be identi ed in the keogram because most of them display a "repeated poleward moving" structure. Subsequently, we analyzed the auroral shapes using a sequence of ASI images to verify whether the shapes of the auroras resembled an auroral arc. Finally, we selected 119 events during a ve-year period from January 2007 that exhibited the speci c characteristics of Pc5 PMAA pulsations.   Figure 2") also shows the Pc5 PMMA pulsations with three-cycles oscillations, but their luminosity decreased with time. The main occurrence region drifted equatorward with time during the three cycle oscillations. As can be seen from the combined all-sky snapshot images at both observatories (right panel), which show the spatial structure of the auroral forms giving rise to the keogram behavior, the auroral arc elongated approximately along the east-west direction from horizon to horizon, for more than 2000 km, though inclined slightly towards north-south. The north-south width of the auroral arc was narrow; it could be less than 20 km at the zenith of FSIM. FSIM is located at almost the same geomagnetic latitude as FSMI. When the event occurred at 1220 UT, the magnetic local time at FSMI and FSIM were 0325 MLT and 0235 MLT, respectively.
Therefore, the auroral arc observed at FSIM appeared at a lower latitude than that at FSMI, which may indicate that the arc appeared approximately along the auroral oval, as the averaged auroral oval is located at the lowest latitude in the midnight region. YKNF is located approximately 2° higher in MLAT and 5° westward in MLON from FSMI. The Pc5 PMAA pulsations were observed over the zenith of YKNF using keogram and ASI. The magnetogram located at YKNF, to which a bandpass lter of 60-600 s was applied, demonstrated that an increase in the H-component started at ~ 1214 UT and reached its maximum amplitude at ~ 1218 UT. Subsequently, it decreased and reached its negative peak at ~ 1222 UT, which was followed by weak two-cycle oscillations with positive peaks at ~ 1225 UT and ~ 1229 UT. The oscillation amplitude of the H component was much larger than that of the D and Z components. Speci cally, the maximum peak-to-peak (pp) amplitude of the H component was ~ 80 nT. The rst negative peak at ~ 1222 UT indicated by the dotted line coincides with the occurrence of the rst PMAA observed at FSMI. However, it is di cult to nd clear coincidence with the H component of magnetic oscillations at the two subsequent PMAAs, especially at the third PMAA.
The magnetogram at FSIM shown in the panel (d) of " Figure 2", which is located at a lower latitude than YKNF, reveals almost the same signature of the magnetic pulsation as that seen at YKNF, although the amplitude of the magnetic pulsation is smaller than that observed at YKNF. For the relationship to the PMAA pulsations observed at FSIM, the magnetogram also revealed a similar signature as that seen at YKNF and FSMI.
Herein We de ne a half-width of luminosity as W(half) = W((I p +I g )/2), where I p is the peak luminosity and I g is the background luminosity. W(half) corresponds to the latitudinal width where the luminosity reduces to half of the peak luminosity. In the case of (b), I p is ~ 6000 and I g is ~ 4200; so, (I p +I g )/2 = ~ 5100. W(half) becomes ~ 0.6°. It is found that the W(half) was approximately the same at (a), (b), and (c). That is, the half-width of luminosity did not depend on the growing, maximum, and declining phases.
The upper panel of " Figure 4" shows the latitudinal distribution of the maximum auroral luminosity at each sampling time on the keogram for PMAA-1, PMAA-2, and PMAA-3. That is, each gure shows the maximum luminosity of aurora considering its latitude during the one-cycle of PMAA. It was found that the luminosity at PMAA-1 during the growing and maximum phases was lower than that at PMAA-2 and 3. Moreover, the luminosity at PMAA-2 was higher during the maximum and declining phases than that at PMAA-1 and 3 Furthermore, it was found that the luminosity at PMAA-3 was somewhat equal to that at PMAA-2 during the growing phase until ~ 69.1°, after which it formed a broad peak at approximately 69.1°-69.4°. After the peak, the luminosity dropped sharply during the declining phase from ~ 69.4°. This demonstrates that the main region of PMAA-3 drifted further to the lower latitude side than that of PMAA-1 and PMAA-2. The latitude of the luminosity maximum during one cycle may correspond to the central region on the FLRs. The signatures shown above demonstrate the dynamic overall activity of PMAA during this event.
The lower panel of " Figure 4" shows the relative phase on the luminosity oscillation of PMAA versus magnetic latitude. In this gure, phase "zero" is the latitude where the luminosity was maximum in PMAA- To examine the PMAA feature in relation to the FLR model, it may be important to investigate a latitudinal dependency of the repetition period of PMAA pulsations between PMAA-1, PMAA-2, and PMAA-3. " Moreover, during the interval between PMAA-2 and PMAA-3, the period on the lower latitude side between the line-190 and line-218 was constant, ~ 300 sec, whereas on the higher latitude side, it is di cult to nd a periodicity because no luminosity peak exists during the PMAA-3. Such a characteristic of low latitude portion as the constant period without latitudinal dependence, e. g., monochromatic oscillation, ts to the FLR model proposed by Chen and Hasegawa (1974) and Southwood (1974), though the period between PMAA-2 and 3 is longer than that between PMAA-1 and 2. On the other hand, the period at higher latitude side of PMAA, between PMAA-1 and 2, lengthens with increasing latitude. This characteristic suggests that the period is related to the local led line length.
3-2-2. Field-line resonance signals observed in space " Figure 6" shows auroral snapshot images observed at FSMI during the time period from the growing phase of the auroral arc pulsation at (a) 1226:00 UT to the declining phase at (c) 1228:09 UT thorough the maximum phase at (b) 1226:57 UT plotted with the footprints of TH-A, TH-D, and TH-E spacecraft at a 110 km altitude based on the TS05 model. The footprints were calculated by the TS05 model using following parameters: Dst, -28.0 nT; solar wind dynamic pressure, 2.0 nPa; interplanetary magnetic eld (IMF) By, 0.5 nT; and Bz, as -1.5 nT. In each snapshot image, the footprints were shown with the triangle symbol under the half-hour trajectory during 1210-1240 UT (see the spatially expanded plot in " Figure  6"). Although TH-D was located in the most westward side and TH-A was located at the highest latitude side, the footprints of the three satellites were located very close to each other, as shown in the plot gures of " Figure 6". The model-calculated footprint was well within the elds of view of the ASIs where the auroral arc was detected, and the footprints were located on the poleward side from the PMAA during most of time in this study. The snapshot image at 1226:00 UT ((a) in " Figure 6") demonstrates that the auroral arc with higher luminosity at the eastern part was located a few tens of kilometers poleward from the zenith. At 1226:57 UT ((b) in " Figure 6"), the luminosity enhancement region elongated westward and moved in the poleward direction. Hence, the footprints of the spacecraft were located more equatorward relative to (a). At 1228:09 UT ((c) in " Figure 6"), the portion of the auroral arc moved more poleward with decreasing luminosity, so the footprints of the spacecraft were located more equatorward than in (a) and (b), and were located at just poleward boundary of the PMAA. A faint PMAA-3 appeared at the lower latitude side of the PMAA-2.
" Figure 7" shows the spacecraft orbit con guration projected onto three different planes (X-Y, X-Z, and Y-Z) in the geocentric solar magnetospheric (GSM) coordinate system during 1210-1240 UT on 2nd March 2011. The details of the orbit of TH-A, TH-D, and TH-E at 1210 UT on the X, Y, and Z planes with GSM coordinates are the following: X is ~ -6.2 Re, ~ -6.0 Re, and ~ -6.0 Re; Y is ~ -9.8 Re, ~ -9.5 Re, and ~ -9.8 Re; and Z is ~ 1.4 Re, ~ 1.0 Re, and ~ 1.0 Re. These orbits demonstrate that, at ~ 0410 MLT, the spacecraft were located near the northern hemisphere' s magnetic equatorial region in the post-midnight sector. Regarding the relative orbital location among the three satellites, it is worth noting that TH-D was located at almost the same position as TH-E in the X and Z coordinates, but they were separated by ~ 0.2 Re in the Y direction, and the position of TH-A was located at almost the same position as TH-E in the Y plane, but was separated by 0.4 Re northward in the Z plane from TH-D and TH-E.
In the following analysis, general mean-eld-aligned (MFA) coordinates were used. In this system, the Z component is parallel to the average direction of the ambient magnetic eld, the Y component is azimuthally perpendicular to the magnetic meridian (westward), and the X component is roughly in the radial direction. Since the magnetic eld near the equatorial plane was stretched, deviating from a dipolelike topology. We found that, when the data were plotted on GSM coordinates, the X and Y components of the magnetic eld intensity were comparable to the total magnetic eld intensity (not shown here the gure). This signal suggests that the eld-line topology was extremely stretched to tail-ward. Therefore, the MFA coordinate system can display more physically understandable wave signals than the GSM coordinate system. The average magnetic eld was calculated as sliding averages of 600 points (30 min It was found that the oscillation of the Ex started ~ 3 minutes earlier than that of the luminosity enhancement of the Pc5 PMAA. Subsequently, the signal displayed three-cycle oscillations and a weak increase at the fourth cycle oscillation, which could correspond to the fourth cycle oscillation with weak luminosity, as shown in the panel (a) of " Figure 8". Furthermore, it was observed that the oscillation amplitude attained a maximum at the second cycle, which may correspond to the occurrence of PMAA-2, whose luminosity was highest among three cycles of PMAA, as demonstrated in " Figure 4". It was noticed that after the start of the oscillation at ~ 1218 UT, the oscillation at TH-A exhibited a phase lag with respect to that of TH-D and TH-E. Moreover, this relative phase lag between TH-A and TH-D lengthened during the following oscillations. This implies that the period of the Ex oscillation in TH-A was longer than that in TH-D and TH-E. Such signatures are more clearly found in the Y component of magnetic eld (By) variations as seen in the panel (d) of " Figure 8" This signature is the same as the periodicity of PMAA observed at the higher latitude side (~ 70.0°-70.5° MLAT), as depicted in " Figure 5" where the period between PMAA-1 and PMAA-2 lengthened with increasing latitude. This signature suggests that the footprint of TH-A could be located in the region of the higher latitude side of the PMAA. Accordingly, this characteristic suggests that the period relates to the local led line length. The oscillation phase lag at the eld line of the poleward location suggests that these waves were generated by FLR. However, the FLR model (Chen and Hasegawa, 1974;Southwood, 1974) postulates that waves are monochromatic, that is, the wave period is the same everywhere, as demonstrated in " Figure 5." Therefore, the features observed at TH-A did not agree with the monochromatic frequency FLR model, instead they indicated that each eld line oscillates at a different eigen-frequency. This is discussed further in Session 4. Furthermore, before the oscillations started, the base-line intensities at TH-A, TH-D, and TH-E were ~ − 6 mV/m, − 9 mV/m, and − 3 mV/m, respectively. The peak-to-peak amplitudes of TH-D and TH-E were almost the same during the entire oscillation period, whereas the amplitude at TH-A was ~ 20-30% larger than that at TH-D and TH-E during the rst cycle, and became comparable to that at TH-D and TH-E during the second and third cycles. It is worth noting that the relative intensity difference between TH-D and TH-E was rather high in comparison to the By as seen in the panel (d) of " Figure 8". It was found that the oscillation amplitude at TH-A was larger than that at TH-D and TH-E, whereas the amplitudes of TH-D and TH-E were approximately equal. Moreover, the oscillation amplitude attained a maximum at the second cycle, which may correspond to the PMAA-2, in the same manner as demonstrated in the Ex behavior. The fundamental-mode Alfven waves have the property that the magnetic node is located at the equatorial region; therefore, it is reasonable that the amplitude at TH-A was larger than at TH-D and TH-E, because TH-A was located at ~ + 1.4 Re in the Z coordinate and TH-D and TH-E were located more equatorward at ~ + 1.0 Re in the Z coordinate, as shown in " Figure 7." It was also observed that the oscillation of TH-A exhibited a phase lag with respect to that of TH-D and TH-E. The TH-A lag time from TH-E at the positive rst, second, third, and fourth peaks were ~ 20 s, ~ 30 s, 8 0 s, and ~ 90 s, respectively. Moreover, the relative phase lag between TH-D and TH-E was less than ~ 20 s in all four peaks.
The compressional mode wave at the Z component of magnetic led (Bz) oscillations is plotted in the panel (e) of " Figure 8." This eld-aligned component of the magnetic eld demonstrated that both the baseline and oscillation amplitude were larger at TH-A than at TH-D and TH-E. The relative amplitudes at TH-D and TH-E were approximately equal. Such characteristics are reasonably explained by the relative location of the spacecraft. That is, TH-A was located at the higher latitude side (in the northern hemisphere) from the magnetic equator, than TH-D and TH-E. Furthermore, TH-A data demonstrated that a negative oscillation started from ~ 1218 UT and reached a negative peak at ~ 1220 UT, after which it formed a positive peak at ~ 1222 UT, whereas TH-D and TH-E presented weaker negative and positive peaks almost simultaneously with TH-A. Subsequently, the waveforms at TH-A displayed a phase lag from TH-D and TH-E signals in the same manner as documented above.
The panel (f) of " Figure 8" presents the variation in ion pressure based on Electrostatic Analyzer observation made by three satellites. It was found that the pressure was gradually increased with time until ~ 1218 UT at the location of all three satellites, after which the pressure decreased with time and started the long-period oscillations. The pressure at TH-E was higher than that at TH-D and TH-A during the entire event. This suggests that TH-E would be located closer to the central plasma sheet than the TH-D and TH-A because the ion pressure escalates at that location. It is interesting to note that the decrease in ion pressure started at ~ 1218 UT, which is when the Ex and By oscillations started. The luminosity peaks, which are marked with vertical long and short dash lines in the panel (b) of " Figure 8 phase difference (lower panel) between the two data. It can be seen that the coherence had a high value of more than 0.9 at the period of ~ 200-300 s, and the phase difference was ~ 150°. The power spectrum did not show a clear peak power at the speci ed period in this cross-correlation analysis.
" Table 1" shows the summary of the coherence around the period of ~ 200-300 s between the luminosity pulsations at lines 223, 220, 210, and 200 observed on the ground, and the X, Y, and Z components of the magnetic and the X and Y components of the electric eld oscillations observed onboard the TH-A, TH-D, and TH-E satellites. Higher coherence columns, > 0.8 are shown in different colors. From this, the magnetic eld (Bx, By, Bz) and electric eld (Ex, Ey) variations and the luminosity pulsation at line 220 showed the highest correlation when compared with the other luminosity lines for the TH-D and E satellites, and that at line 223 for the TH-A. Meanwhile, the coherence with the luminosity at line 200 had the lowest value for the TH-A, D, and E satellites. From these results, we can draw an important conclusion that the luminosity pulsations around line 220 and 223 correlated well with the magnetic and electric oscillations observed in the magnetosphere. This indicates that the real footprints of the TH-A, TH-D, and TH-E satellites may be located near the eld line of lines 223 and 220. Moreover, the footprints calculated by the TS05 model shown in " Figure 6" were located at the poleward side on the PMAA, which showed good agreement with the results shown in " Table 1 ). Furthermore, the periods between the rst and second peaks at lines 210, 215, and 220 were estimated to be ~ 265 s, ~ 265 s, and ~ 275 s, respectively. The periods between the second and third peaks at lines 210 and 215 were ~ 300 s and ~ 293 s. The lines 210 and 215 are located in the monochromatic frequency FLR region where the period is constant and independent of latitude. Moreover, line 220 is located in the multi-frequency FLR region where the period dilates with increasing latitude, as plotted in " Figure 5." The panel (c) of " Figure 10" shows the Y component of the magnetic eld observed at TH-A, TH-D, and TH-E. The periods at TH-D and TH-E are approximately equal, whereas the period at TH-A is longer than that at TH-D and TH-E. This signature suggests that the TH-D and TH-E were located in the monochromatic frequency FLR region, but the TH-A was located at a multi-frequency FLR region. It is worth noting that the repetition period of the magnetic oscillation was ~ 10-25% longer than that of the Pc5 PMAA pulsations.

3-3. Relation between FLR and Solar wind parameters
The source mechanism of the FLR that has received the most attention is surface instabilities at the magnetopause (such as Kelvin-Helmholtz instability (KHI) and pressure impulse in the solar wind plasma) [Rostoker and Sullivan, 1987;Engebretson et al., 1998;Kivelson and Southwood,1985]. If the ground-based Pc5 PMAA pulsations are the result of this instability, it is likely that their characteristics depend on solar wind conditions. Therefore, we now examine this relationship. The panels (a) and (b) of " Figure 13" show the Y component of the magnetic eld and the Y component of ow velocity, respectively, observed onboard the TH-A. The panels (c) and (d) of " Figure 11" show the Y and Z components, respectively, of interplanetary magnetic eld (IMF), the panels (e) and (f) of " Figure 11" show the solar wind ow speed and pressure, and the panel (g) of " Figure 11" shows the AE index. The solar wind and IMF parameters were taken from the time-shifted OMNI-1 min data. The highlighted region indicates the time interval we examined. It is found from (a) and (b) that few FLR cycle signals were sporadically observed before and after our observations. Corresponding to these FLR oscillations, the Pc5 PMAA pulsations were also observed on the ground as shown in " Figure 1". By looking at IMF Bz, it is found that the FLR phenomena in this study were observed during the period of positive Bz after a recovery of about 1.5 hours of negative Bz, which started at ~ 1045 UT. It is worth noting that a positive Bz (~ 2 nT) was observed at ~ 1220 UT when the FLR phenomena were observed. The most distinct characteristic is that the solar wind speed was very high, approximately 650 km/s, during the observations. We will take a look on solar wind conditions in more detail for this event using Geotail data.
Geotail satellite was located at the upstream region near the dayside magnetopause. The orbit at 1200 UT on the X, Y, and Z planes with GSM coordinates was 19.6 Re, ~ -22.0 Re, and ~ -6.1 Re. The panel (a) of " Figure 12" shows the X, Y, and Z components of the magnetic eld observed at TH-A. The Pc5 oscillations were observed from ~ 1218 UT at the vertical line in this gure, as shown in " Figure 8." The IMF magnetic eld and the solar wind speed data obtained by Geotail was signi cantly different from the data obtained from OMNI, as depicted in " Figure 11". Sudden changes were observed at ~ 1219 UT on IMF By (from ~ -3 nT to ~ 2 nT) and Bz (from ~ 2 nT to ~ -2 nT). Similar sudden changes were also found in the solar wind velocity at Vy (from ~ 20 km/s to ~ 50 km/s) and Vz (from ~ 30 km/s to ~ -10 km/s).
Even if we assumed that these drastic changes in the IMF and solar wind speed could trigger the FLR oscillations, their timing would be inconsistent. Speci cally, because the Vx is ~ 660 km/s, it takes approximately 4-5 minutes when the rapid change of the solar wind observed at Geotail reaches TH-A. Therefore, under the general/standard solar wind condition, we have to exclude the possibility that the magnetospheric magnetic and electric eld oscillations observed at ~ 1218 UT onboard TH-A were affected by the solar wind discontinuity observed at 1219 UT onboard Geotail. However, when we checked the polar cap Super Dual Auroral Radar Network (SuperDARN) ( Figure 13," we found that the polar cap convection enhanced to more than ~ 800 m/s from ~ 1218 UT. These signatures obtained from the SuperDARN data indicate a typical convection pattern under the sudden negative changes in IMF Bz (Ruohoniemi and Greenwald, 1998;Shepherd et al., 1999). To explain why the discontinuity arrived at Earth's magnetopause earlier than Geotail, we can speculate that the surface discontinuity of the solar wind had large tilt in the Vx direction. Under this condition, the sudden negative changes in the IMF Bz may cause the sudden enhancement of magnetospheric convection, which could have affected the generation of the FLR oscillations, and caused the thinning of plasma sheet, as shown in " Figure 8." The negative Bz also caused the expansion of the auroral oval, after which, the primary occurrence region of PMAA drifted equatorward with time, as shown in " Figures 2, 4, and 8."

Discussion
In this study, we reported on the observational evidence of the Pc5 PMAA, magnetic pulsations, and the magnetospheric magnetic and electric eld oscillations through the ground-space coordinated observations. We examined Pc5 magnetic pulsations from the ground at YKNF (69.3° MLAT, 302.7° MLON) and FSIM (67.2° MLAT, 294.4° MLON), as depicted in " Figure 2." The oscillation signatures observed at the two observatories were somewhat similar, although the amplitude of the magnetic pulsation at YKNF, where the PMAA pulsations were observed over the zenith of the observatory, was larger than that at FSIM. Moreover, the H component of magnetic variations observed at the two observatories did not exhibit a clear coincidence with the PMAA pulsations. Previous studies reported that the Pc5 magnetic pulsations and auroral luminosity pulsations showed good correlation, and that both pulsations have a latitudinal phase shift variation (e.g., Xu, et al., 1993;Milan, et al., 2001;Samson et al., 1996Samson et al., , 2003. On the other hand, this event showed that the magnetic variation on the ground neither exhibited a clear coincidence with the PMAA pulsations, nor with magnetospheric magnetic By oscillations, as demonstrated in " Figure   8." These signatures suggest that the magnetic pulsations observed on the ground could not correspond directly to the PMAA pulsations. Furthermore, it was found that the H component of magnetic pulsations observed on the ground demonstrated a rather similar behavior to that of the ion pressure variation in the magnetosphere. As examined above, the magnetic pulsations observed on the ground exhibited complex signatures that were not well correlated with the PMAA, and so far we do not have a clear explanation for the apparent correlation between the ground magnetic eld and the magnetospheric pressure. Clarifying the physical relationship between these variations requires further work. As demonstrated in " Figure 5," the periodicity of the Pc5 PMAA pulsations presented an intriguing signature. The recurrence periods on the lower latitude region (~ 68.5°-70.0° MLAT) between PMAA-1 and 2, and PMAA-2 and 3 were almost constant ~ 265 s and ~ 300 s. These signatures demonstrate that the lower latitudinal region of the PMAA pulsations conformed to the monochromatic frequency FLR region, where the period is constant and without latitudinal dependence. Such a monochromatic FLR model was proposed by Chen and Hasegawa (1974) and Southwood (1974). The phase-latitude pro les of the PMAA pulsations at the lower latitude region (~ 68.5°-70.0° MLAT), as shown in " Figure 4," also demonstrated that these signatures tted to the monochromatic frequency FLR model, as reported by Milan et al, (2011) and Samson et al., (2003). However, the period increased from ~ 265 s (between PMAA-1 and 2) to ~ 300 s (between PMAA-2 and 3). It is suggested that the FLR condition, such as eldline length and/or plasma density, could change during the time interval. On the other hand, in the higher latitude region (~ 70.0°-70.5° MLAT), the period lengthened with increasing latitude, which indicates that the higher latitude side of the PMAA exhibits a multi-frequency FLR region and that the period was proportional to the local eld line length. It means that the two different types of FLR oscillations were excited simultaneously. To the best of our knowledge, there are no reports of such FLRs observations.
From the luminosity distribution analysis as graphed in " Figure 3b," the latitudinal half-width of the arc luminosity was ~ 0.6°. It was also found that its size was somewhat constant during the enhancement of PMAA from the growing phase to the declining phase. It is widely known that the auroral arc luminosity is proportional to the intensity of eld-aligned current (FAC) (e.g., Borovsky, 1993). If a half-width of luminosity corresponds to the latitudinal width of the FAC, the features mentioned above suggest that the width of FAC was invariable from growth to the declining phase of PMAA. The latitudinal width of the FAC was ~ 0.6 degree at that moment. The FAC region moved poleward in association with the movement of PMAA within the latitude from ~ 68.5° to ~ 70.5° MLAT. The generation of PMAA may closely relate to FLR; therefore, the location of the luminosity maximum during one cycle of PMAA may correspond to the central resonance region of the FLR, where the eld of the line matches the resonance frequency.
In the following, we discuss the in uence of the magnetospheric magnetic and electric eld oscillations observed at the THEMIS spacecraft on the Pc5 PMAA pulsations in the ionosphere and the magnetosphere-ionosphere (M-I) coupling processes to connect both.
Using the data of THEMIS ground-space coordinated observations, we found that the Pc5 PMAA pulsations observed on the ground occurred in conjunction with the enhancement of the magnetic and electric eld oscillations observed near the equatorial plane of the magnetosphere. The magnetic and electric eld signal displayed three-cycle oscillations and a weak increase at the fourth cycle oscillation, similar to the main PMAA pulsations and the fourth cycle oscillation with weak luminosity. It is found the oscillation amplitude showed a maximum at the second cycle that corresponded to the occurrence of the PMAA-2 whose luminosity was highest among the three-cycle main pulsations. It was also found that the coherence between the magnetic and electric eld oscillations and the luminosity pulsations had a high value of more than 0.9. From these observations, it is suggested that the PMAA pulsations and the magnetospheric eld oscillations were initiated by the same physical mechanism (presumably originating in the magnetospheric side) and thus they oscillated concurrently through the M-I coupling.
The PMAA signal showed that the arc elongates a few thousand kilometers in the east-west direction, but only a few tens of kilometers in the north-south direction. Such features are similar to those of a stable discrete auroral arc that is commonly observed in the evening sector (Gillies et al., 2014). These signals suggest that both the discrete stable arc and the PMAA were generated by a similar mechanism. The generation mechanism of a stable discrete auroral arc is thought to be associated with the upward eldaligned electric eld, which accelerates auroral electrons and enhances the auroral luminosity.
Many experimental and theoretical studies have focused on investigating the mechanism that produces such a eld-aligned electric eld. Examples include static magnetosphere-ionosphere couplings, mirroring of electrons, electrostatic double layers, and dispersion in kinetic or electron inertia Alfven waves in the auroral acceleration region at ~ 1-2 RE (see the review by Borovsky, 1993). However, such physical processes are still under discussion. If we apply a eld-aligned electric eld model to the generation mechanism in this study, the Pc5 PMAA pulsations could be associated with the eld-aligned electric eld that could be caused by the magnetospheric magnetic and electric eld oscillation.
It was found that the compressional component of the magnetic eld (Bz) oscillations displayed high coherence with the luminosity pulsations as shown in " Figure 8" and " Table 1". It is known that the long period modulation of diffuse patch aurora/pulsating aurora is thought to be caused by the electron pitchangle scattering with compressional mode Pc5 pulsations (Oguti et al., 1987;Yamamoto, 1988;Saka et al., 2014). To the best of our knowledge, reports concerning compressional magnetic pulsations causing a discrete auroral arc are somewhat limited to sudden commencement (SC) related auroral events (e.g., Liu et al., 2011;2013). The contribution of the compressional magnetic pulsation to the Pc5 PMAA pulsation is intriguing; however, it requires further study.
Our study may offer important observational results to investigate the mechanism of producing a eldaligned electric eld from the magnetic and electric eld oscillation, particularly the magnetosphereionosphere (M-I) coupling processes. The relationships between the eld-aligned electric elds in the electron acceleration region at an altitude of ~ 1-2 RE and the magnetic and electric eld oscillations near the equatorial plane in the magnetosphere are essential to consider the generation of this phenomena. To investigate this relationship in future works, coordinated observations are necessary for simultaneous observations of the magnetic and electric eld waves near the equatorial plane, the eldaligned current and the precipitating electron ux in the ionosphere, and the optical PMAA pulsations, as well as a numerical simulation study.
In previous studies, three-generation mechanisms of the FLR oscillations have been postulated. The rst is a KHI caused by the velocity shear between the solar wind and the magnetopause (Southwood, 1974;Chen and Hasegawa, 1974), the second is the pressure impulse/oscillation in the solar wind plasma (Kivelson and Southwood,1985), and the third is the Alfven impulse caused by the sudden magnetospheric convection changes by the sudden IMF Bz changes (Kozlovsky and Kangas, 2002). In " Figures 11 and 12", we showed that the solar wind speed was very high, approximately 650 km/s, during the event on the 2nd March 2011. This suggests that the KHI could be the most likely cause of the FLR.
On the other hand, it was found from the panel (c) and (d) of " Figure 8" that the electric and magnetic eld oscillations on the three satellites started simultaneously with synchronized waveforms, and subsequently, the waveform at the TH-A (located at higher latitudes) displayed a phase lag relative to those at TH-D and TH-E. The oscillation period of the magnetic and electric eld at higher latitudes was longer than that at lower latitudes, which demonstrates the oscillations of individual magnetic shells.
This suggests that the eld line oscillations were triggered simultaneously in a wide region by an impulse, such as rapid convection changes caused by the sudden variations of the IMF Bz observed by SuperDARN radar and Geotail satellite, as shown in " Figures 12 and 13". Further observations are required to understand their generation mechanism.

Summary
This study constitutes the rst coordinated observation of the Pc5 PMAA pulsations (~ 4-5 min in period) and magnetic pulsations on the ground and the magnetic and electric eld oscillations onboard THEMIS spacecraft near the equatorial plane of the magnetosphere, whose footprints were located near the Pc5 PMAA pulsations in the post-midnight sector.
The optical characteristics of the Pc5 PMMA pulsations demonstrated that the east-west aligned arcs, more than 2000 km long, moved poleward with clear three-cycle oscillations within a period of ~ 4.5 min. The north-south width of the arc was narrow, less than 20 km, at the minimum scale. lengthened. This signature demonstrated that the oscillations on the lower latitudinal side of the PMAA conformed with the monochromatic frequency FLR region, where the period is the constant and without latitudinal dependence. Such a monochromatic FLR model was proposed by Chen and Hasegawa (1974) and Southwood (1974). Meanwhile, the higher latitude side of the PMAA presented a multi-frequency FLR region where the period increased with increasing latitude. The phase-latitude pro le of the PMAA pulsations on the lower latitude region also demonstrated that these signatures tted with the monochromatic frequency FLR model.
Pc5 PMAA pulsations occurred in conjunction with the enhancement of the magnetic and electric eld oscillations in the magnetosphere. This magnetospheric oscillation signal displayed three-cycle oscillations and a weak increase at the fourth cycle oscillation, which corresponded to the main PMAA pulsations and the fourth cycle oscillation with weak luminosity, respectively. It was also found from the cross-correlation analysis that the coherence between the magnetospheric magnetic and electric eld oscillations and the luminosity pulsations had a high value of more than 0.9. From these observations, it is suggested that the PMAA pulsations and the magnetospheric eld oscillations were initiated by the same physical mechanism (presumably originating in the magnetospheric side), and thus, they oscillated concurrently through the M-I coupling.
The coordinated satellite data also indicated that the period of the magnetospheric magnetic and electric eld oscillation was longer at the higher latitude site, thereby demonstrating a multi-frequency FLR region.
The solar wind speed was signi cantly high, approximately 650 km/s, during this event. The magnetospheric magnetic and electric eld oscillations could be triggered simultaneously in a wide region by an impulse, such as rapid convection changes caused by the sudden variations of the IMF Bz observed by the SuperDARN radar and Geotail satellite.        Spacecraft orbit con gurations in three different planes (X-Y, X-Z, and Y-Z) in the GSM coordinate system.