Geomagnetic response to solar wind dynamic pressure impulse events at high-latitude conjugate points

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH: SPACE PHYSICS, VOL. 118, , doi: /jgra.50555, 2013 Geomagnetic response to solar wind dynamic pressure impulse events at high-latitude conjugate points H. Kim, 1 X. Cai, 1 C. R. Clauer, 1 B. S. R. Kunduri, 1 J. Matzka, 2 C. Stolle, 2,3 and D. R. Weimer 1 Received 29 April 2013; revised 3 September 2013; accepted 6 September 2013; published 1 October [1] It is commonly assumed that geomagnetic activity is symmetrical between interhemispheric conjugate locations. However, in many cases, such an assumption proved to be wrong. Especially in high-latitude regions where the magnetosphere and the ionosphere are coupled in a more complex and dynamic fashion, asymmetrical features in geomagnetic phenomena are often observed. This paper presents investigations of geomagnetic responses to sudden change in solar wind pressure to examine interhemispheric conjugate behavior of magnetic field variations, which have rarely been made mainly due to the difficulty of facilitating conjugate-point measurements. In this study, using magnetometer data from three conjugate stations in Greenland and Antarctica, solar wind pressure impulse events (>5 npa in <16 min) and their geomagnetic responses, typically seen as magnetic impulse events, have been examined. Our results suggest that asymmetry in ground response patterns between the conjugate locations often shows little correlation with interplanetary magnetic field orientation, season, and ionospheric conductivity, indicating that much more complex mechanism might be involved in creating interhemispheric conjugate behavior. Citation: Kim, H., X. Cai, C. R. Clauer, B. S. R. Kunduri, J. Matzka, C. Stolle, and D. R. Weimer (2013), Geomagnetic response to solar wind dynamic pressure impulse events at high-latitude conjugate points, J. Geophys. Res. Space Physics, 118, , doi: /jgra Introduction [2] The response of the Earth s magnetosphere to a sudden increase in solar wind dynamic pressure (or called solar wind dynamic pressure impulse) and/or interplanetary magnetic field (IMF) provides a unique opportunity to investigate systematically how the solar wind couples to the dayside magnetosphere and ionosphere. Such increases in solar wind dynamic pressure produce sudden impulse (SI) events observed as a rapid increase of the geomagnetic fields at low and middle latitudes and magnetic impulse events (MIEs) at high latitudes [e.g., Araki, 1977; Russell et al., 1994; Lanzerotti et al., 1991; Sibeck, 1993; Sitar et al., 1996] and can cause geomagnetic storm sudden commencement (SC) if associated with a negative IMF Bz [e.g., Araki et al., 2006]. Additional supporting information may be found in the online version of this article. 1 Center for Space Science and Engineering Research, Virginia Tech, Blacksburg, Virginia, USA. 2 National Space Institute, Technical University of Denmark, Copenhagen, Denmark. 3 Now at Helmholtz Centre Potsdam, GFZ, German Research Centre for Geosciences, Potsdam, Germany. Corresponding author: H. Kim, Center for Space Science and Engineering Research, Virginia Tech, Ste. 1000, 1901 Innovation Dr., Blacksburg, VA 24060, USA. (hmkim@vt.edu) American Geophysical Union. All Rights Reserved /13/ /jgra Solar wind dynamic pressure impulses compress the dayside magnetosphere and thus enhance the electric fields in the magnetopause, creating complex current system in both the magnetosphere and the ionosphere including SC as discussed in Araki [1994]andAraki et al. [2009]. Observations at geosynchronous orbit also showed that magnetospheric compression on the dayside was found in response to solar wind dynamic pressure impulses [e.g., Lee and Lyons, 2004; Villante and Piersanti, 2011]. Lee and Lyons [2004] found that IMF orientation plays an important role in geosynchronous magnetic field response to such variations in the solar wind. However, some studies [e.g., Lanzerotti et al., 1991] argued that solar wind dynamic pressure change may not be a major factor of SI events on the ground. [3] At high latitudes, solar wind pressure changes may produce traveling convection vortices (TCVs) or transient stationary responses [Friis-Christensen et al., 1988]. While on the dayside at low and middle latitudes, ground responses tend to occur in a rather simple fashion (i.e., step functionlike waveform in the H component), more complex geomagnetic signatures are observed at high latitudes (both auroral and cusp regions) owing to field-aligned currents (FACs) connecting to the ionosphere [McHenry and Clauer, 1987; Friis-Christensen et al., 1988; Glassmeier et al., 1989; Glassmeier and Heppner, 1992; Glassmeier, 1992; Le et al., 1993; Araki, 1994; Lanzerotti et al., 1991; Sibeck et al., 2003]. Clauer and Petrov [2002] suggested the Kelvin- Helmholtz instability as a source of TCVs based on their

2 statistical observations of the relationship between solar wind speed and vortex activity. [4] While the effect of solar wind dynamic pressure impulses on SI events or MIEs is still not fully understood, it appears that the geomagnetic field responses to solar wind pressure impulse events are associated with the field-aligned currents (FACs) and their resulting ionospheric currents (i.e., Pedersen and Hall currents). Observations by Kikuchi et al. [2001], Araki [1977], Lanzerotti et al. [1991], Araki et al.[2006], and Araki et al. [2009] reported an adjacent two-pulse structure with opposite sense in MIEs at high latitudes suggesting it is provided by a pair of FACs and FAC-produced ionospheric currents. Ground magnetic field signatures due to FAC-produced Hall currents have been investigated by modeling work in McHenry and Clauer [1987] and Lanzerotti et al. [1986]. Kivelson and Southwood [1991] presented a modeling work demonstrating traveling vortical motions associated with FACs driven by pressure changes at the magnetopause. A model simulation by Kataoka et al. [2004] also presented generation mechanism of pairs of FACs due to the deformation of the magnetopause caused by a localized density pulse. Belenkaya et al. [2004] suggested a model showing the distribution of field-aligned currents in the transition current system produced from the high-latitude magnetosphere in association with a sudden pressure increase and simultaneous northward IMF turning. [5] Observations of interhemispheric conjugate behavior of MIEs (and TCVs) at high latitudes suggest their association with FACs directed in the same direction in both hemispheres [e.g., Lanzerotti et al., 1990, 1991]. A modeling scheme shown in their study demonstrates how FACs directed in the same direction in both hemispheres produce the vertical component perturbations in the conjugate hemispheres. Multi-instrument observations at conjugate points reported by Kataoka et al. [2001] showed that an MIE accompanied by TCVs was associated with upward FACs with soft electron precipitation that was located near the trailing edge of the Hall current loop. They found no solar wind dynamic pressure impulse during the event, suggesting an abrupt IMF cone angle change and its interaction with the bow shock as an indirect trigger of the MIE. Murr et al. [2002] reported for the first time that the spatial structure of auroral precipitation and a TCV event was directly related and found that the overall nature of the TCV event at the conjugate points is similar. They concluded that the TCV is a conjugate phenomenon which occurs on closed field lines. Fillingim et al. [2011] reported that the Time History of Events and Macroscale Interactions during Substorms (THEMIS) and Polar satellites, and their ground support at conjugate points in both hemispheres observed ionospheric responses to a hot flow anomaly (HFA), suggesting that the origin of the magnetic perturbation and resulting auroral emission is the deformation of the magnetopause due to HFA-magnetosphere interaction. [6] Simultaneous observations in both hemispheres, however, are rare and challenging mainly due to the lack of a dense array and limited high-quality data in the Southern Hemisphere. Moreover, some of the studies mentioned above used stations that are not exactly conjugate. While it has often been assumed that our understanding of phenomena in one hemisphere can be simply mapped to the other hemisphere, several findings in the previous studies KIM ET AL.: CONJUGATE MAGNETIC IMPULSE EVENTS [e.g., Murr et al., 2002] indicate that it is possible that the two hemispheres may respond differently to a solar wind discontinuity. In addition, it is not yet clearly known whether ground responses are of totally magnetospheric or the combination of magnetospheric and ionospheric effects. [7] Therefore, conjugate point ground observations will help to answer whether interhemispheric symmetry in ground response is expected if the FACs are one of the controlling factors. The main objective of this study is to find the dynamics between solar wind pressure impulse events and their geomagnetic responses at magnetically conjugate highlatitude ground stations and to examine if there is symmetry or asymmetry in geomagnetic field disturbances between the conjugate locations. Geomagnetic signatures such as SI and MIE are often associated with solar wind pressure impulse events. However, solar wind pressure impulse events do not necessarily produce geomagnetic events [e.g., Sitar et al., 1996]. In our study, solar wind dynamic pressure impulse events, observed in space, are first examined from ACE data and their link to geomagnetic signatures is then studied using the ground magnetometer array at highlatitude conjugate points for the time period of Our study is accented by the ground observations of the responses to the solar wind sudden pressure impulse events using high time resolution data at more strictly chosen three conjugate pairs at high latitudes (interhemispheric conjugate point differences are within < 1 ı magnetic latitude (MLAT) and <5 ı magnetic longitude (MLON)) to examine interhemispheric behavior of MIEs Data Set and Methodology [8] Solar wind pressure sudden impulse events are identified from ACE level 2 solar wind data (temporal resolution of 64 s) obtained from February 1998 to October Chosen are the events in which the solar wind dynamic pressure (= n p v 2 sw ) increases at least 5 npa in less than 16 min and is steady (standard deviation of the pressure change or <0.5) for 1 h before and 20 min after the pressure impulse event. IMF orientations are not considered when choosing sudden impulse events although ACE magnetic field data, provided in Geocentric Solar Magnetospheric coordinates, are also included in the survey. [9] GOES 8, 9, 10, 11, and 12 satellite magnetic field data (the time resolution of 60 s) were also examined to find inner magnetospheric responses to pressure impulse events. The GOES satellite magnetometers conform to the PEN coordinate system, in which H P is a magnetic field vector component pointing northward, perpendicular to the orbit plane (parallel to Earth s spin axis) and H E points earthward, being perpendicular to H P. H N completes the Cartesian coordinate and points eastward. [10] Geomagnetic responses to solar wind pressure impulse events were surveyed using the data set from various ground-based fluxgate magnetometers including International Real-Time Magnetic Observatory Network (INTERMAGNET), Magnetometer Array for Cusp and Cleft Studies (MACCS), Greenland west coast stations, and their geomagnetic conjugate stations in Antarctica (Automatic Geophysical Observatories, AGO; British Antarctic Survey, BAS). Three conjugate pairs, each of which is within MLAT < 1 ı and MLON < 5 ı,are

3 Table 1. Geographic and Geomagnetic Locations of the Magnetometer Stations Used in This Study a Geographic Geomagnetic Station Station Code Group Latitude Longitude Latitude Longitude L MLT MN in UT Conjugate pair 1 Kangerlussuaq STF DTU Space AGO P03 P03 PENGUIn Conjugate pair 2 Iqaluit IQA NRCan BAS LPM M BAS Conjugate pair 3 Pangnirtung PGG MACCS South Pole SPA PENGUIn Low latitude Kourou KOU IPGP a The geomagnetic model parameters are obtained using the 2003 International Geomagnetic Reference Field (IGRF) corrected geomagnetic (CGM) model. DTU Space is National Space Institute at Technical University of Denmark; PENGUIn, Polar Experiment Network for Geospace Upper Atmosphere Investigations; NRCan, Natural Resources Canada; BAS, British Antarctic Survey; MACCS, Magnetometer Array for Cusp and Cleft Studies; and IPGP, The Institute of Earth Physics of Paris. examined to study interhemispheric behavior of ground responses to solar wind pressure sudden impulse events. The geographic and corrected geomagnetic (CGM) coordinates of the stations used in our study are listed in Table 1. Most of the stations are located along the ı magnetic meridians where the magnetic local time (MLT) midnight is UT. [11] The ground magnetometer data are sampled in the temporal resolution of 1 s (IQA, AGO P03, and M88 316), 5 s (PGG), 10 s (SPA), 20 s (STF), and 60 s (KOU) and mean level subtracted. The ground data presented in this study conform to the HEZ coordinate convention, in which H points to magnetic north, E magnetic east, and Z downward. To identify geomagnetic responses to solar wind pressure impulse events observed by the ACE spacecraft, data from the low-latitude station (Kourou, KOU, 9.46 ı MLAT; ı MLON) were first surveyed. It has been suggested that the equatorial regions typically show direct and sensitive responses to sudden impulse events [e.g., Sibeck, 1993, and references therein]. Thus, the signatures observed at KOU (increases in the H component) are used to determine the onset times of the MIEs and to time shift the ACE data presented in this paper after comparing them with the solar wind pressure events. [12] Each MIE presented in this paper has been visually inspected to specify the duration of the events. The event start time is defined as the onset of the responses as observed in the H component at the low-latitude station, KOU. The event end time is when the increasing signature at KOU stops. In fact, there are some events in which response end times cannot be easily determined mainly due to their continuously pulsating nature (quite possibly some of them should be ULF waves caused by the solar wind impulse events). Therefore, in this study, we focus on the first positivenegative or negative-positive impulse signatures. This can also help make the hodogram easy to read by showing the only one rotation of polarization as will be seen in the next section. [13] To examine conjugate aspects of geomagnetic field responses to solar wind pressure impulse events, we compared the amplitude difference in magnetic impulse deflection in the H component ( H North H South ), cross correlation, and ionospheric convection between the conjugate events. Since MIEs often display unique magnetic field variations (typically, a bipolar structure in the H component), cross correlation between the conjugate events can be used to quantify symmetry/asymmetry characteristics. Using 6057 the cross-correlation analysis, arrival time difference (phase delay) of MIEs between conjugate points is also estimated. Although MIEs often associate TCV events, this paper does not cover the traveling vortical structures. As magnetic field perturbations are observed as ionospheric convection on the ground via Hall currents, hodograms are useful in displaying the pattern of the perturbations. A hodogram is a graphical representation of tracing of field vectors over the duration of an event. In this study, the H component and E component of the magnetic field data are plotted in the X Y plane after the magnetic field data are rotated by 90 ı counterclockwise to represent ionospheric convection inferred from the magnetic field perturbations. The solar elevation angle ( S), defined as the angle between the horizon and the center of the Sun s disc, is used to examine whether interhemispheric responses are related with ionospheric conductivity. The angles during each event have been calculated using the National Oceanic and Atmospheric Administration (NOAA) website tool at azel.html. [14] Data from the Super Dual Auroral Radar Network (SuperDARN) covering the conjugate stations in this study are also used to observe ionospheric convection patterns in response to a solar wind pressure sudden impulse event. The SuperDARN data provide 2 min averaged convection vector information derived from Doppler velocities associated with E B plasma drift in the ionosphere. The three stations in the Northern Hemisphere are STO (Stokkseyri), PYK (Pykkvibaer), and GBR (Goose Bay). Their conjugate stations in the Southern Hemisphere are SAN (Sanae), SYS (Syowa South), and SYE (Syowa East). Among the events studied in this paper, only one event coincided with the SuperDARN conjugate coverage. The simultaneous ground magnetometer and SuperDARN observations are presented in section Convection projection maps are also presented here for the conjugate study of convection patterns using the IGRF model for internal sources and Tsyganenko model T01 [Tsyganenko, 2002] for external (magnetospheric) current sources as implemented in Baker et al. [2004]. 3. Observations [15] Seventy solar wind pressure impulse events have been identified from the ACE satellite in the year from 1998 to 2010 when the ground magnetometer chain was in the dayside. In this study, a narrower dayside MLT range (MLT

4 Figure 1. A solar wind dynamic pressure impulse event and its responses at geosynchronous orbit and on the ground, observed on 27 January (a) Solar wind dynamic pressure (Pd) and (b) interplanetary magnetic field data from the ACE satellite. The solar wind pressure impulse event is marked by the vertical lines i and ii. (c) GOES 8 and (d) GOES 10 magnetic field data. (e) Ground magnetometer data from the midlatitude station, Kourou (KOU). (f and g) Ground magnetometer data from the high-latitude conjugate pair, Kangerlussuaq (STF) and AGO-P03, along with ionospheric convection patterns (polarization hodograms) on the right-hand side. The duration of the ground response is indicated by the vertical lines i and iii. Note that the ACE data are time shifted by 57 min to be compared with the GOES and ground data. The CGM latitudes and longitudes of each ground station are shown below the station codes. noon 3 h) is chosen to exclude a more complex case in which the dayside magnetospheric current system interacts with the nightside current system. Among the 70 events, a total of 24 ground events showed ground responses (MIE) from the conjugate points. Section 3.1 presents three example events. The first and the second example events occurred under similar conditions both of them being observed at near-local noon in association with a similar solar wind pressure impulse structure (approximately 17 npa increase within a couple of minutes). However, the first example displays poor conjugacy while the second one shows welldefined conjugate responses. The third example, displaying poor conjugacy, is presented with SuperDARN observations of ionospheric convection patterns at conjugate locations Case Study Example Event 1 [16] Figure 1 displays a solar wind dynamic pressure impulse event on 27 January 2000 and its responses in the inner magnetosphere detected by the GOES satellites and 6058 on the ground observed by the magnetometer network. The ACE data, which are time shifted by 57 min in this graph, show the sudden increase (15 npa) within 1 min (indicated by the vertical lines i and ii) in solar wind dynamic pressure (Figure 1a). This corresponds to dpd/dt =0.25(nPa/s). The IMF Bz was northward during the event (Figure 1b). GOES 8, which was located at 09:47 MLT at onset, detected an increasing magnetic field in the Hp component (Figure 1c) showing an inner magnetospheric response to the impulse event on dayside while no clear signature is found by GOES 10 (Figure 1d) which was located at 05:46 MLT at onset. The low-latitude station, Kourou (KOU), which is used to determine the onset time of the ground response to the impulse event, measured a clear increase in the magnetic field in the H component (Figure 1e), starting at 14:53 UT (12:23 MLT), indicated by the vertical line i. Such a response in dayside equatorial regions has been commonly observed [Araki, 1994]. [17] The ground responses at a high-latitude conjugate pair, Kangerlussuaq (STF) and AGO-P03, are also shown

5 Figure 2. The changes in the ionospheric convection in both hemispheres around the stations used in this study for the event from 14:53 UT to 15:03 UT on 27 January The solid dots and the arrows represent the starting points of the data and the polarization sense, respectively. The station locations, based on the CGM coordinate system, are marked by the open triangles. in Figures 1f and 1g, respectively, along with the changes in the ionospheric convection (hodogram) on the right-hand side. The time range of the hodograms are indicated by the vertical lines i and iii. This event occurred when S =4 ı at STF and S =21 ı at P03, implying that the ionospheric conductivity over P03 might be higher. The interhemispheric response appears to be quite different in terms of response amplitude. During the event in the time range from 14:53 UT to 15:03 UT (indicated by the vertical lines i and iii), the difference in the amplitudes of the disturbances ( H) is 116 nt and the correlation coefficient between the two conjugate events is 0.5 with the event at P03 delayed by 217 s. The ionospheric convection patterns show left-hand rotation in both hemispheres. [18] The changes in the ionospheric convection in both hemispheres around the stations used in this study are shown in Figure 2. Although there were fewer conjugate points in the Southern Hemisphere, it appears that the convection patterns reveal similarity in both hemispheres. The stations located near 30 ı 50 ı MLON in both hemisphere display left-hand rotation in the ionospheric convection. The 6059 conjugate stations near 15 ı MLON (IQA and SPA), on the other hand, observed the opposite rotation of convection (right-hand) Example Event 2 [19] Similar to the previous example event, a sudden increase in solar wind dynamic pressure (19 npa within 2min(dPd/dt =0.16) indicated by the vertical lines i and ii) was observed from the ACE data on 18 March 2002 as shown in Figure 3a. The ACE magnetic field data indicated that the IMF Bz was transiting from near 0 to southward during the impulse event (Figure 3b). The onset of its responses in the inner magnetosphere and on the ground is also clearly observed by the GOES 8 and GOES 10 satellites and the magnetometer network. The GOES 8 satellite, which was located at 08:15 MLT at onset, measured an increasing magnetic field in the Hp component (Figure 3c). The response observed by the GOES-10 satellite, located in the pre-dawn sector (04:09 MLT at onset), does not appear to be as welldefined as that of GOES 8 (Figure 3d). The midlatitude station, Kourou (KOU), which is used to determine the onset time of the ground response to the event, also observed

6 Figure 3. A solar wind dynamic pressure impulse event and its responses at geosynchronous orbit and on the ground, observed on 18 March (a) Solar wind dynamic pressure (Pd) and (b) interplanetary magnetic field data from the ACE satellite. The solar wind pressure impulse event is marked by the vertical lines i and ii. (c) GOES 8 and (d) GOES 10 magnetic field data. (e) Ground magnetometer data from the midlatitude station, Kourou (KOU). (f and g) Ground magnetometer data from the high-latitude conjugate pair, Pangnirtung (PGG) and South Pole (SPA) along with ionospheric convection patterns (polarization hodograms) on the right-hand side. The duration of the ground response is indicated by the vertical lines i and iii. Note that the ACE data is time shifted by 48 min to be compared with the GOES and ground data. The CGM latitudes and longitudes of each ground station are shown below the station codes. the well-defined increasing signature starting at 13:23 UT (10:53 MLT) in the H component (Figure 3e). It is shown that the signatures at GOES 8 and the high-latitude stations appear to begin a few minutes earlier than the onset time determined by the event observed at KOU. This timing issue might be due to either the arrival time difference between at the geosynchronous orbit and the ground, which were in different local times, or the low time resolution of the instruments (GOES: 60 s; KOU: 60 s) compared to the other instruments (PGG: 5 s; SPA: 10 s) so the precise timing could not be made. [20] Figures 3f and 3g show the ground responses at the high-latitude conjugate pair, Pangnirtung (PGG) and South Pole (SPA), respectively. A cross-correlation analysis estimates the correlation coefficient of 0.96 between the two conjugate responses from 13:23 UT to 13:30 UT (indicated by the vertical lines i and iii) with the event at SPA delayed by 15 s. The changes in the ionospheric convection (hodogram) shown in the right-hand side show well-defined right-hand rotation in both hemispheres. This event occurred when S =15 ı at PGG and S =2 ı at SPA, implying that the ionospheric conductivity over PGG might be higher. The difference in the amplitudes of the disturbance ( H) appears to be small (59 nt) compared to the total H component variations of 500 nt. [21] Figure 4 shows the changes in the ionospheric convection in both hemispheres around the stations used in this study for the event presented in this section, displaying rather well-defined conjugacy in the ionospheric convection. For example, both IQA and PGG, and their conjugate stations, M and SPA, observed righthand circular polarization. In addition, convection reversals 6060

7 Figure 4. The changes in the ionospheric convection in both hemispheres around the stations used in this study for the event from 13:23 UT to 13:30 UT on 18 March The solid dots and the arrows represent the starting points of the data and the polarization sense, respectively. The station locations, based on the CGM coordinate system, are marked by the open triangles. (from right-hand to left-hand polarization and vice versa) are clearly shown from the stations near 30 ı 40 ı MLON in both hemispheres. However, the latitudes where the reversal occurred appear to be different; in the Northern Hemisphere, the reversal was observed in the region between GDH (MLAT = 75.3 ı )andupn(mlat = 79.0 ı ) while in the Southern Hemisphere, between M (MLAT = 71.2 ı ) and M (MLAT = 73.4 ı ). The convection reversal event suggests a signature of TCV as observed by Clauer [2003]. Murr et al. [2002] reported polarization reversal of a TCV event, occurring at a slightly lower latitude near the station STF (72.9 ı MLAT) Example Event 3 [22] A rather slow increase of a solar wind dynamic pressure was observed from the ACE satellite on 3 April 2004 as shown in Figure 5a. The dynamic pressure increase rate was 0.02 npa/s. The ACE data are time shifted by 61 min in this graph. The IMF Bz was transiting from southward to near 0 during this event (Figure 5b). The responses from the GOES 12 satellite and on the ground are clear as shown in Figures 5d and 5e, showing increases in the 6061 Hp component and H component, respectively. No clear signature is found from the GOES 10 magnetic field data during the event (Figure 5c). GOES 10 was in the predawn sector (04:59 MLT) and GOES 12 was in the morning sector (09:02 MLT) at onset. The midlatitude station, Kourou (KOU), measured the ground onset of the impulse event at 14:10 UT (11:40 MLT, the vertical line i) as shown in Figure 5e. [23] This impulse event was observed by the high-latitude conjugate pairs, Iqaluit (IQA) and BAS M88 316, as shown in Figures 5f and 5g. Polarization hodograms, amplitude differences ( H), and cross correlation between the conjugate stations were examined for the time range from 14:10 UT to 14:20 UT (between the vertical lines i and iii). The polarization hodogram of the IQA data displays right-hand rotation while that of the M data does not appear to be well defined. The difference in the amplitudes of the disturbance ( H) is74 nt and the correlation coefficient between the two conjugate responses from 14:10 UT to 14:20 UT is 0.46 with the event at M delayed by 85 s. S was 26 ı at IQA and 4 ı at M during this event,

8 Figure 5. A solar wind dynamic pressure impulse event and its responses at geosynchronous orbit and on the ground, observed on 3 April (a) Solar wind dynamic pressure (Pd) and (b) interplanetary magnetic field data from the ACE satellite. The solar wind pressure impulse event is marked by the vertical lines i and ii. (c) GOES-10 and (d) GOES-12 magnetic field data. (e) Ground magnetometer data from the midlatitude station, Kourou (KOU). (f and g) Ground magnetometer data from the high-latitude conjugate pair, Iqaluit (IQA) and BAS M along with ionospheric convection patterns (polarization hodograms) on the right-hand side. The duration of the ground response is indicated by the vertical lines i and iii. Note that the ACE data is time shifted by 61 min to be compared with the GOES and ground data. The CGM latitudes and longitudes of each ground station are shown below the station codes. suggesting that the ionospheric conductivity over IQA might be higher. [24] The changes in the ionospheric convection in both hemispheres around the stations used in this study are shown in Figure 6. It appears that both stations located around 70 to 80 ı MLAT and their conjugate stations located around 70 to 80 ı MLAT display right-hand (or quasi right-hand) rotation in their convection patterns. The convection patterns at higher latitudes in the Northern Hemisphere show more linear polarization. [25] SuperDARN data were available in both the highlatitude conjugate locations that encompass the ground stations for the event presented in this section. Ionospheric plasma convection (E B) velocity vectors derived from line-of-sight SuperDARN measurements in both hemispheres and a mapped projection of the convection vectors using a magnetic field model are shown in Figure 7. Figures 7a 7d represent 2 min averaged convection patterns during four time periods: one before and three after the onset time (14:10 UT; 11:40 MLT). The first column in each panel displays the convection vectors measured in the Northern Hemisphere, covering the west and east coasts of Greenland and northeastern Canada. The second column shows the convection vectors measured from the locations conjugate to the northern counterparts. The convection vectors shown in the third column in each panel are obtained by projecting the northern data onto the conjugate locations in the Southern Hemisphere determined by the IGRF model for internal sources and the Tsyganenko T01 model for external (magnetospheric) current sources. The panels in the second and third columns are plotted such that they are seen through the Earth from the North Pole. [26] The ionospheric plasma convection patterns shown in Figure 7 reveal their temporal changes around the onset time and interhemispheric differences. In the Northern Hemisphere, the convection patterns around the local time of interest (11:40 MLT) near 75 ı MLAT undergo changing directions and increasing velocities before and after the onset of the MIE. More points with westward convection appear to be seen in the Northern Hemisphere after the onset 6062

9 Figure 6. The changes in the ionospheric convection in both hemispheres around the stations used in this study for the event from 14:10 UT to 14:20 UT on 3 April The solid dots and the arrows represent the starting points of the data and the polarization sense, respectively. The station locations, based on the CGM coordinate system, are marked by the open triangles. while such tendency is not clearly observed in the Southern Hemisphere. It is also observed that the convection reversal shown at near 75 ı MLAT and 13 MLT in the Southern Hemisphere was not observed at the same latitudes in the Northern Hemisphere (Panel b). Unfortunately, there are very few overlapping conjugate point observations near the local time in the Southern Hemisphere. In both the Northern Hemisphere and the projection map, westward convection is shown dominantly near 75 ı MLAT and between 12 and 13 MLT after the onset. It is also found that the data from the Southern Hemisphere (second column) and the mapped projection (third column) do not appear to agree with each other between 14 and 16 MLT at around 75 ı. The maximum ground deflection occurred at the high-latitude stations at around 14:15 UT as shown in Figures 5f and 5g. However, no remarkable difference in SuperDARN data is observed around 14:15 UT (Figure 7d) compared to the previous time period (Figure 7c). At this time, IQA (CGM latitude of 72.5 ı ) was at local time of 10:09 MLT while M (CGM latitude of 72.1 ı ) was at 10:45 MLT Statistical Study [27] Interhemispheric asymmetry between the conjugate pairs was examined statistically using the 24 events observed in our study. First, the relationship between solar wind dynamic pressure events and their ground responses is examined as shown in Figure 8, in which the ground responses ( H) are well correlated with the solar wind dynamic pressure increases ( Pd). H for both KOU and the high-latitude conjugate stations is computed from the peak-to-peak difference in the H component over the entire duration of each event. As shown previously in the example events, the low-latitude station, KOU, observed monotonic increases in the H component in response to solar wind impulse events (Figure 8a). Although somewhat weak, H component geomagnetic field changes ( H) at highlatitude conjugate points appear to be linearly correlated with solar wind dynamic pressure changes ( Pd) as shown in Figure 8b. Given the linear fits for both the northern and southern events shown in Figure 8b, the slight difference in responses between the hemispheres might be either due to

10 Figure 7. Two minute averaged convection velocity vectors obtained from Super Dual Auroral Radar Network (SuperDARN) before and after the onset time (14:10 UT; 11:40 MLT) for the event shown in section 3.1.3). (a) The first and second columns present convection patterns observed in the Northern and Southern Hemispheres and the third column convection vectors in the Southern Hemisphere mapped from the northern convection using the Tsyganenko model for the time periods 14:08 14:10 UT. (b) Same as in Figure 7a except for the time period 14:10 14:12 UT. (c) Same as in Figure 7a except for the time period 14:12 14:14 UT. (d) Same as in Figure 7a except for the time period 14:14 14:16 UT. Note that the maps are represented in the geomagnetic coordinate system: The meridional lines are in MLT. The panels in the second and third columns are plotted such that they are seen through the Earth from the North Pole. 6064

11 Figure 8. Geomagnetic field response ( H) to solar wind dynamic pressure increase ( Pd) at (a) the low-latitude station KOU and (b) the high-latitude conjugate stations for the entire set of events presented in this study. weaker magnetic fields in the south or merely due to the small number of the events. [28] In addition to solar wind dynamic pressure, we examined ground response asymmetry in relation to IMF changes. IMF components, measured by the ACE satellite, are calculated by averaging the IMF components for two time ranges, i.e., before (from 5 min before to the onset of the impulse) and after (from the onset of the impulse to 5 min after) impulse. The change of the averaged IMF values around the onset (before and after the impulse) is divided by the duration of the solar wind pressure impulse event to estimate the rate of IMF change (db/dt). Figure 9 demonstrates cross correlations and amplitude differences in the H component ( H North H South ) of the ground responses in relation to IMF By and Bz changes over the duration of the solar wind pressure increase (db/dt). As shown in the figure, it is found that neither the cross-correlation coefficients nor the amplitude differences in the H component show a clear correlation with the IMF By and Bz changes. However, although not very evident, it appears that smaller amplitude differences between the conjugate points are observed when the IMF By and Bz change rate is small (Figures 9c and 9d). The ground response parameters, crosscorrelation coefficients and amplitude differences, are also compared with the IMF By, IMFBz, and clock angle (the angle between IMF By and Bz) before and after impulses separately, also showing no clear correlations (figures not shown here). [29] Figure 10 shows ground magnetic field H-component variations from selected events. The onset times, which are determined by the H component responses at the lowlatitude station, KOU, as described in sections 2 and 3, are indicated by the first dotted vertical lines. The solid lines Figure 9. Asymmetry in geomagnetic field response to solar wind pressure impulse events: (a and b) cross coefficient and (c and d) amplitude difference, H North H South in relation to the IMF By and Bz changes over the duration of the solar wind pressure increase (db/dt). 6065

12 Figure 10. Magnetic field variations in the H components of the conjugate ground stations from selected events. The onset times are indicated by the first dotted vertical lines. The panels are arranged in order of the level of cross correlation: (a c) highly (r 0.9), (d f) moderately (0.7 r <0.9), and (g i) poorly (r <0.7) correlated events. Labeled in each plot are correlation coefficient (r) and arrival time difference in seconds (t d ) over the course of each event (marked with the two dotted vertical lines). represent data from the northern stations and the dashed lines their conjugate stations in the Southern Hemisphere. The events reveal that asymmetry is found not only in their response patterns (e.g., pulsation forms) but also in their amplitudes and arrival time differences. Figure 10 is arranged in order of the level of cross correlation: highly (r 0.9) (Figures 10a 10c), moderately (0.7 r <0.9) (Figures 10d 10f) and poorly (r <0.7) (Figures 10g to 10i) correlated events. We point out that Figures 10g and 10i are the first and third example events, respectively. [30] Figure 11 shows the statistical distribution of crosscorrelation coefficients (Figure 11a), arrival time differences (Figure 11b), and amplitude differences in the H component between the conjugate points for the entire set of events presented in this study (Figure 11c). The cross-correlation coefficients are quite largely distributed ranging between 0.23 and 0.96 (Figure 11a), with the arrival time differences ranging 200 s (in which case the northern hemispheric response leads) to 200 s (in which case the southern hemispheric response leads). Seventy-five percent of the events show that the arrival time differences are less than 60 s (Figure 11b). The arrival time difference appears to be rather centered around 0 s while the cross-correlation coefficients are widely distributed as mentioned earlier. The stations in the Northern Hemisphere usually recorded stronger H than those in the Southern Hemisphere did during the events (Figure 11c) although the differences in H show no clear dependence with other factors such as local time, season, ionospheric conductivity (estimated by solar elevation angle, S), and geomagnetic activity (Dst and Kp indices) (figures not shown). Figure 12 shows arrival time differences in the H components between the conjugate points as a function of local time (Figure 12a) and cross-correlation coefficient (Figure 12b). It appears that the arrival time differences have local time dependence the Southern Hemisphere tends to respond earlier in the prenoon sector whereas in the afternoon sector, the Northern Hemisphere leads. It also appears that the arrival time differences are smaller in the moderately and highly correlated events (r 0.7). 6066

13 [31] The changes in the ionospheric convection at the three conjugate pairs are represented as polarization hodograms for selected events in Figure 13. The hodograms are created from the data obtained during the events (7 to 15 min after the onset of each event). Among the total 24 events examined in this study, 13 events displayed similar polarization sense between the conjugate stations whereas the rest (11 events) showed mismatches in polarization sense. It is also found that the convection patterns do not appear to be correlated with solar wind IMF orientations, local times, and interhemispheric waveform correlation and arrival time difference (figures not included). KIM ET AL.: CONJUGATE MAGNETIC IMPULSE EVENTS Figure 12. (a) Local time dependence of arrival time differences in the H components between the conjugate points. (b) Relationship between arrival time difference and crosscorrelation coefficient in the H-component. Figure 11. Statistical distribution of (a) cross-correlation coefficient, (b) arrival time difference, and (c) amplitude difference in the H components between the conjugate points for the entire set of events presented in this study. 4. Discussion [32] This study presents geomagnetic responses at conjugate points in association with solar wind dynamic pressure impulse events. As described in section 1, there is a correlation between solar wind dynamic pressure impulse events and MIEs. This is also shown in Figure 8. Korotova et al. [2011] suggested that from their THEMIS and ground observations, a widespread solar wind pressure impulse instead of a localized flux transfer event (FTE) is the cause of the MIE in their study. It is interesting, however, that in our study, only 24 events have been identified as responses to the 70 solar wind pressure impulse events. Sibeck [1993] explained that the lack of ground response might be because of events in the magnetosphere are (1) weak that the signals were not detected on the ground or (2) temporally not abrupt enough to produce necessary FACs required to generate strong transient features on the ground. It is also worthwhile to note that there have also been studies suggesting that solar wind dynamic pressure impulses could be merely one of the drivers responsible for MIEs on the ground. Lanzerotti et al. [1991] found that changes in solar wind dynamic pressure were associated with only 25% of MIEs. Similarly, the study by Sitar et al. [1996] also showed that changes in solar wind pressure and the characteristics of the geomagnetic response are not correlated. A statistical study by Kataoka et al. [2003] found that 50% of the MIEs were associated 6067

14 Figure 13. The changes in the ionospheric convection at the conjugate stations for selected events. The panels are arranged in the same order as shown in Figure 10. The horizontal components (H and E) of the magnetic field data are plotted in the X Y plane. The vertical component of the local magnetic fields is out of the figure. The time periods marked in each plot indicate the temporal ranges of the hodograms. with hot flow anomaly (HFA) while 30% and 20% of the MIEs were related with bursty reconnection and pressure pulses, respectively. [33] Sibeck [1993] found that variations in the IMF orientation control the fraction of the solar wind dynamic pressure applied to the magnetosphere and consequently the magnetospheric magnetic field strength. Kataoka et al. [2002] observed MIEs accompanied by TCVs at interhemispheric conjugate points and attributed the signatures to interplanetary tangential discontinuities exhibiting a rapid turning of the IMF and abrupt solar wind dynamic pressure changes. On the other hand, Sitar et al. [1996] observed that neither the amplitude nor the direction of IMF Bz plays a role in determining the amplitude of the ground signatures. A similar idea is also suggested in Figure 9, which indicates that IMF orientation has little effect on asymmetry in the ground responses represented as cross correlation and amplitude difference in the H components at the conjugate points except that it appears that the amplitude difference is smaller with small IMF By and Bz change rates. We also investigated interhemispheric asymmetry under different IMF By, IMF tilt, and clock angle conditions and found no clear correlation, similar to the observations by Sitar et al. [1996]. [34] Unlike previous studies [e.g., Lanzerotti et al., 1991; Kataoka et al., 2001; Murr et al., 2002] showing good interhemispheric conjugacy in their observations of MIEs, our observation results reveal quite notable asymmetry in geomagnetic response between the conjugate points. The wide range of cross-correlation coefficient, arrival time difference, and amplitude difference in the H components between the conjugate stations are clearly shown in Figures 10 and 11. This is perhaps partly because we carried out conjugate observations under various conditions (IMF orientations, seasons, and ionospheric conductivity) and investigated the more detailed conjugate response features (amplitude difference, cross correlation, and convection patterns) than previous studies have covered. Figure 12a displays a slight local time dependence of arrival time differences in the H component signatures between the conjugate points. However, the physical implication is not clearly explained because IMF By (both before and after the impulse events), which is known to create asymmetry in the dawn-dusk direction, showed little effect on asymmetry between the conjugate points. No local time dependence of the other parameters (e.g., cross correlation and amplitude difference) was observed from our statistics. Similarly, no strong local time dependence of the magnitude of the MIEs on the ground was found in the study by Lanzerotti et al. [1991]. [35] It should be noted that the pairs of the stations in this study are nominally conjugate (in terms of their magnetic latitudes and longitudes) based on the IGRF/CGM model. Therefore, we additionally examined the conjugacy of each station pair location during each impulse event using the Tsyganenko T01 model. As the T01 model requires solar wind parameters (IMF and dynamic pressure) and Dst index, conjugate points may vary depending on such conditions. Magnetic field tracing between the conjugate locations using the model failed only for 5 of the 24 events, implying that the station pairs were in the open field line region during those events. For the rest, most of the station pairs maintained conjugacy during each event although deviations in magnetic longitude appear larger than those in magnetic latitude. The average percentage errors of magnetic latitudes and longitudes at conjugate locations based on the Tsyganenko model are 1.3% and 8.2%, respectively. However, we also found that the interhemispheric asymmetry (amplitude difference and cross correlation) is not necessarily related with whether the field lines were open or closed, which is determined by the Tsyganenko model. In other words, of the five events where the field lines were found to be open, some events showed symmetry in ground responses between the conjugate locations. It is known that neither IGRF nor the Tsyganenko model can provide accurate information about conjugacy especially when the magnetosphere is in 6068

15 changing state during impulse events, which cannot be precisely stated or measured. The events showing high correlation and small difference in timing suggest that the interhemispheric stations were on nearly conjugate field lines, while where they differ, the pairs might be on open field lines or on the field line connecting to the magnetopause. As an example, Siscoe et al. [2001] described how field lines connect and the magnetosphere can be twisted up and thus field lines end up at different MLTs in opposite hemispheres. Figure 12b clearly demonstrates the new finding that the nominally conjugate pairs are sometimes on field lines that go somewhere else but at other times, the conjugacy is confirmed. [36] From multi-instrument observations of a TCV event associated with the FAC system, Murr et al. [2002] suggested that the overall nature of the TCV event at the conjugate points is similar, including an upward and then downward FAC pair, the amplitude of the perturbations, the propagation speed, and the spatial scale. However, from the magnetometer and riometer signatures, they found a time delay between the two hemispheres (IQA-SPA pair) with the event at SPA being delayed by 2 min while the perturbations were similar in shape and amplitude. They explained that the delay might be due to the difference in the length of the field line (using the Tsyganenko model) from the magnetospheric driver to each conjugate location (IQA and SPA). Their observation of the time delay can be explained by Alfvén waves propagating along the field lines of such differences. The observations of arrival time difference between conjugate points address an important question as to whether such conjugate asymmetry has seasonal dependence (dipole tilt angle variation). We also examined the relationship between seasons and the ground responses (i.e., cross correlation, arrival time difference, and amplitude difference) and found no clear correlation (figure not shown). [37] Of course, consideration of seasonal dependence also includes the ionospheric conductivity difference (sunlit conditions) between hemispheres. Our study observed no clear dependence of the ionospheric conductivity (figure not shown). The event shown in Figure 1 is a good example in which the amplitude of the response in the Northern Hemisphere is remarkably larger although the ionospheric conductivity over the station in the Southern Hemisphere might be much higher because of the higher S. While some studies have shown that ionospheric conductivity plays a role in determining the magnitude of ground response [e.g., Le et al., 1993; Kikuchi et al., 2001], others found no effect on ground signatures. Conjugate observations by Lanzerotti et al. [1991] and Murr et al. [2002] showed that the amplitude of the magnetic perturbations was similar at the conjugate locations regardless of the ionospheric sunlit conditions. Murr et al. [2002] argued that this might indicate that TCVs are driven by a current source in the magnetosphere. Fillingim et al. [2011] found the difference between the maximum magnetic disturbance in the Northern Hemisphere and the auroral emission in the Southern Hemisphere during the TCV events, attributing it to a decoupling of these processes possibly due to the difference in the ionospheric conductivity between the hemispheres. [38] Ideally, the same signatures of ionospheric convection (or polarization) would be expected at conjugate points if FACs, generated from the equator, are directed with respect to both ionospheres assuming the ionospheric conditions have little effect on determining ground response features. Similarities in polarization sense at conjugate points indicate that ground magnetic perturbations are produced by FACs flowing into or out of the ionospheres in both hemispheres [e.g., Kataoka et al., 2001]. However, there are studies that reported asymmetrical conjugate patterns in ionospheric convection. For example, Lanzerotti et al. [1991] found among the 60 conjugate events, 51 events were of the opposite polarization sense. Our study found that of the 24 conjugate events, only 13 events showed similar polarization sense. In addition, no local time dependence of the ellipticity of the MIEs was found from our observations. Figure 13 also shows asymmetry in polarization patterns. Polarization differences between the conjugate points might suggest the difference in mapping of the field lines associated with the perturbations between the points. The study of a TCV event by Murr et al. [2002], however, argued that the skewed flow vortices are not solely due to mapping but are possibly a reflection of the driving magnetospheric flows. [39] The SuperDARN results shown in Figure 7 also clearly display asymmetry in convection patterns at conjugate points. Although the southern hemispheric observation points are much less than the those of the Northern Hemisphere, one can see how differently the convection patterns evolve after the onset of the MIEs in both hemispheres. This might indicate that the current systems responsible for the MIEs were not established at the same magnetic latitudes in both hemispheres. In addition, the mapped projections are quite different from the actual measurements. Again, this could support the idea of the interhemispheric magnetic field mapping issue using the models such as the IGRF and Tsyganenko models as mentioned earlier. It should be noted that it is also possible that the discrepancy could be due to the fact that some of the points in both hemispheres do not overlap. [40] A number of studies have shown that solar wind pressure perturbation at the magnetospheric boundary creates FACs flowing into the high-latitude ionospheres [e.g., Farrugia et al., 1989; Southwood and Kivelson, 1990; Lysak and Lee, 1992; Araki, 1994; Sibeck et al., 2003; Kikuchi et al., 2001]. With the FACs flowing from the magnetospheric equatorial region to both hemispheres, good interhemispheric conjugacy is expected assuming the dipole tilt angle and ionospheric conductivity play little role in conjugate responses. Apparently, our study suggests this is not the case. Lanzerotti et al. [1991] pointed out that conjugacy cannot be expected to be maintained during periods of high geomagnetic activity. However, we found no correlation with Kp and Dst indices. [41] The results shown in our study might suggest that asymmetric ground responses to solar wind pressure impulse events are attributed to a local small-scale source more dominantly than the large-scale magnetospheric counterpart. In other words, although both conjugate points respond to an impulse event via an interhemispheric FAC system, a local current system such as Hall currents (perhaps due to different magnetic field mapping between conjugate points) can create asymmetry. Using the formulation in Glassmeier [1987] and Lanzerotti et al. [1991], the field-aligned currents (J k ) are estimated by J k e kha B G / 0 l where k h is the horizontal scale of the current system, a the height of 6069