Kokubun (1983) reported the local time variation of normalized amplitude of sudden commencement (SC) with a strong day-night asymmetry (i.e., maximum amplitude near noon and minimum amplitude near midnight) at geosynchronous orbit with 81 SC events. Further careful inspection of Kokubun's local time distribution reveals that the normalized SC amplitudes in the prenoon sector (MLT = 9–12) are larger than those in the postnoon sector (MLT = 12–15). That is, there is a morning-afternoon asymmetry in the normalized SC amplitudes. Until now, however, there are no studies on this SC-associated morning-afternoon asymmetry at geosynchronous orbit. Motivated by this previous observation, we investigate a large data set (422 SC events in total) of geosynchronous SC observations and confirm that the geosynchronous SC amplitudes normalized to SYM-H are larger in the morning sector than in the afternoon sector. This morning-asymmetry is probably caused by the enhancement of partial ring current, which is located in the premidnight sector, due to solar wind dynamic pressure increase. We also examine seasonal variations of the normalized SC amplitude and find that the SC-associated geosynchronous magnetic field perturbations are dependent on seasons of the year. This may be due to the location of the magnetopause current and cross-tail current enhanced during the SC interval with respect to geosynchronous spacecraft position.
 The Earth's magnetosphere is dynamically and continuously disturbed by variations in the solar wind and the interplanetary magnetic field (IMF). The strongest magnetospheric disturbance on a short timescale occurs during the passage of interplanetary (IP) shock accompanied by large changes in the solar wind dynamic pressure. The IP-shock associated geomagnetic disturbance is known as a sudden commencement (SC), which is the sudden increase in the horizontal (H) component on the ground at low latitudes resulting from abrupt and strong steplike changes in solar wind dynamic pressure behind the shock front. At high latitudes on the ground bipolar magnetic field variations occur for step-like increases in the solar wind dynamic pressure. These global signatures on the ground have been successfully explained by a model using the combined effects of fast-mode waves, Alfvén waves, field-aligned currents produced by magnetospheric convection, and ionospheric currents [Araki, 1994].
 The solar wind dynamic pressure strongly controls the geosynchronous magnetic field strength and direction. When a high solar wind dynamic pressure front compresses the magnetosphere, both the magnetopause surface current and the cross-tail current are enhanced [e.g., Wing and Sibeck, 1997]. Thus, SC-associated magnetopause and cross-tail currents contribute to geosynchronous magnetic field perturbations on the dayside and on the nightside, respectively. This leads to a local time asymmetry for the amplitude of SC at geosynchronous orbit with a maximum amplitude near noon and a minimum amplitude near midnight. Such a day-night asymmetry of geosynchronous SC amplitudes has been reported by many authors [e.g., Kokubun, 1983; Kuwashima and Fukunishi, 1985; Kuwashima et al., 1985; Borodkova et al., 2005; Villante and Piersanti, 2008; Wang et al., 2009]. Further careful inspection of the local time dependence of geosynchronous SC amplitudes in those studies reveals that the amplitude in the morning side is larger than that in the afternoon. That is, there is a morning-afternoon asymmetry. However, the published previous works had not been focused on this morning-afternoon asymmetry.
 Since SC amplitude at geosynchronous orbit depends on changes in the magnetopause and/or cross-tail currents associated with changes in solar wind dynamic pressure, it varies considerably in each case because of the scatter of the magnitude of the SC source. In order to remove the variability, the normalization of geosynchronous SC amplitude to the solar wind dynamic pressure [Borodkova et al., 2005; Villante and Piersanti, 2008] or to the averaged SC amplitude at low-latitude ground stations [Kokubun, 1983; Kuwashima and Fukunishi, 1985; Kuwashima et al., 1985] has been used to examine how the magnetosphere responses to solar wind pressure changes. Nevertheless, the normalized amplitude shows a large spread of values. This indicates that the response of the geosynchronous magnetic field is very complicated. Thus it is important to continue the careful analyses of both case and statistical studies, in order to improve our understanding of how the geosynchronous magnetic field responds to sudden solar wind changes.
 The objective of this study is to answer two questions: (1) What causes the morning-afternoon asymmetry of the normalized geosynchronous SC amplitudes? (2) What causes a large scatter of the normalized geosynchronous SC amplitudes? In order to answer question 1, we need an asymmetric current enhancement during magnetospheric compressions. The asymmetric ring current is considered in an empirical model developed by Tsyganenko and Sitnov . The model values are compared with our observations, and we discuss whether our observations can be explained with the asymmetric ring current. In order to answer question 2, we statistically examine seasonal variations of the normalized amplitudes. Previous studies mainly focus on the total magnetic field and/or the compressional component (i.e., Bz in GSM coordinates or H in VDH coordinates), which is parallel to the ambient magnetic field direction, at geosynchronous orbit during SC intervals. Unlike previous studies, we examine the SC-associated geosynchronous magnetic field perturbations in the radial BV, azimuthal BD, and compressional BH components in VDH coordinates and discuss how these field components depend on the magnetic local time and seasons of the year.
 The paper is organized as follows. Section 2 describes the data and event selection. Section 3 presents two geosynchronous SC events observed near noon and midnight, respectively. Section 4 presents statistical results. Discussion is presented in section 5, and the conclusions are presented in section 6.
2. Data Sets and Event Selection
 In order to statistically examine geosynchronous magnetic field variations during SC events, we use geosynchronous magnetic field data measured at GOES 6 and GOES 7 for the period from January 1990 to April 1993 and at GOES 8, GOES 9, GOES 10, GOES 11, and GOES 12 for the period from January 1996 to December 2009. The GOES magnetic field data used in this study are 1-min averages of 0.5-s samples acquired with a fluxgate magnetometer. The 1-min GOES data (GOES 8 to GOES 12) were obtained from coordinated data analysis web (CDAWeb), http://cdaweb.gsfc.nasa.gov/cdaweb. The 3-sec resolution magnetic field data from GOES 6 and GOES 7 are provided from NOAA National Geophysical Data Center, http://goes.ngdc.noaa.gov. The 3-sec GOES 6 and GOES 7 data were resampled at 1-min intervals to make sampling common with GOES 8 to GOES 12.
 The dipole VDH coordinate system is adopted for the magnetic field at geosynchronous orbit. In this coordinate system H is parallel to the dipole axis and is positive northward, V points radially outwards and is perpendicular to the magnetic dipole axis, and D completes a right-hand orthogonal system and is positive eastward. Since the dominant magnetic field component at geosynchronous orbit is the H component and we are interested in magnetic field perturbations with respect to the dipole axis, the dipole VDH coordinate system is appropriate for describing the magnetospheric compression in the geosynchronous region by IP shock. We examine SC-associated magnetic field perturbations in the BV, BD, and BH components at geosynchronous orbit.
 SC is a clear global phenomena, and its onset is nearly simultaneously detected everywhere on the Earth. However, the amplitude of SC at mid/low-latitude ground stations shows a local time dependence [e.g., Araki et al., 2006]. Thus, we use the 1-min SYM-H index [Iyemori and Rao, 1996], which is calculated by averaging the H-component disturbance for six stations at mid/low latitudes, to identify the SC events on the ground. The SYM-H is provided at http://wdc.kugi.kyoto-u.ac.jp. The onset times of SC events have been provided at ftp://ftp.ngdc.noaa.gov/STP/. We selected the SC events defined as a rapid increase in SYM-H values with more than 5 nT (ΔSYM-H) within 10 min (ΔT: the SC risetime). During the period from January 1990 to April 1993 and from January 1996 to December 2009, 422 SC events are selected from the SYM-H data.
 To examine solar wind and IMF conditions for the SC events, we use the solar wind and IMF data from the SWE (∼94 s) [Ogilvie et al., 1995] and MFI (∼46 s) [Lepping et al., 1995] instruments, respectively, on the Wind spacecraft, and from the SWEPAM (∼64 s) [McComas et al., 1998] and the MFE (∼16 s) [Smith et al., 1998] instruments, respectively, on the ACE spacecraft. The solar wind and IMF data have been provided by CDAWeb. We choose the solar wind and IMF conditions corresponding to 169 SC events from 1998 to 2009.
3. Geosynchronous SC Events Near Noon and Midnight
Figures 1a–1d show the solar wind velocity, density, dynamic pressure, IMF strength (|B|), and IMF Bz in GSM (geocentric solar magnetospheric) coordinates, respectively, observed by ACE from 1600 to 1800 UT on 31 October 2000. During the 2-hour interval, ACE was located near GSE (x, y, z) ∼ (219.9, 7.2, −22.6) RE. Around 1629 UT ACE observed the passage of an interplanetary shock (i.e., sudden increases in the solar wind speed, density, dynamic pressure, and the IMF strength). Through the shock front, the velocity changed from 340 to 430 km s−1, the solar wind density from ∼5 to ∼20 cm−3, resulting in an increase in the solar wind dynamic pressure from ∼1 to ∼6 nPa, and the IMF strength from 5 to 12 nT. Just before the shock, the IMF Bz was near zero, and it suddenly rotated southward around 1629 UT.
Figures 2a and 2b show the magnetospheric response at geosynchronous orbit and on the ground, respectively, to the interplanetary shock observed by ACE around 1629 UT. A sudden increase in the SYM-H started at 1714 UT, and its amplitude (ΔSYM-H) increased 26 nT within 5 min (ΔT), which is the time interval between the onset of a sudden increase and the maximum of SYM-H. This SYM-H increase is classified as a SC. During this SC event, GOES 8 was located near local noon (∼12.4 LT) and observed a sudden increase in the BH component about 1-min earlier, at 1713 UT. The 1-min time delay between the geostationary satellite and the Earth surface requires a propagation speed of the compressional front of ∼600 km/s. This value is much smaller than the averaged equatorial Alfvén velocity of ∼1000 km/s, which gives ∼38-s time delay over 6 RE, estimated by Takahashi and Anderson . It should be noted that GOES and SYM-H data at 1-min resolution have an uncertainty of ±30-second. Thus, ΔT for GOES in the dayside starts 1-min earlier than ΔT for SYM-H or has the same start time as ΔT for SYM-H. The time interval between the initial response (1713 UT) and the maximum of BH (1717 UT) at GOES is 4 min. That is, the initial response time to the IP shock at GOES is not the same as the risetime (ΔT = 5 min) of SYM-H. The risetime, called SC interval, is determined by time for an IP shock to sweep geoeffective distance along the magnetopause for ground detection of the magnetospheric compression [Araki et al., 2004]. In our study we are interested in geosynchronous magnetic field perturbations associated with SC events, which are identified from the ground SYM-H data. Thus we use ΔT (i.e., the SC interval) determined by ground SYM-H data and examine the geosynchronous magnetic field variation (ΔB) during the SC interval.
Figure 3 shows the SC event observed on February 28, 2002, when GOES 8 was located near midnight (23.6 LT). The SC onset was identified at 0450 UT with a marked increase in SYM-H (ΔSYM-H = 32 nT) within 7 min. After 0457 UT, SYM-H gradually increases until 0504 UT. In our study we select the interval while SYM-H exhibits a steep increase as a risetime (ΔT) of SYM-H. At 0453 UT GOES 8 observed a sudden decrease in the BH component with an amplitude of −13 nT within 7 min. This sharp decrease near midnight geosynchronous orbit can be explained by the enhancement of the cross-tail current [e.g., Kokubun, 1983; Wang et al., 2009]. The sudden decrease in BH at GOES occurred 3 min later than the SC onset. Since this 3-min time lag is much longer than the propagation time delay of fast mode waves from ground to geosynchronous orbit, we suggest that the time delay comes from the propagation motion of solar wind discontinuity moving tailward in the magnetosheath around the flank magnetopause [Kawano et al., 1992].
4. Statistical Analysis
 In this section we present the results of the statistical characteristics of geosynchronous magnetic field perturbations (ΔBV, ΔBD, ΔBH) in VDH coordinates normalized to ΔSYM-H for the 422 SC events selected using the procedure described above. The normalization is introduced with the intention of removing data scatter, which may be caused by the scatter of the magnitude of the source of SC events. The same method was used in previous SC studies [Kokubun, 1983; Kuwashima and Fukunishi, 1985].
4.1. Local Time Dependence
Figures 4a, 4c, and 4e show the local time distribution of SC-associated geosynchronous magnetic field amplitude in V, D, and H components normalized to ΔSYM-H for 422 SC events, respectively. Note that the number of data points plotted in the figures is much larger than that of the SC events identified from SYM-H. This is due to a simultaneous observation by two GOES spacecraft for 345 SC events. In Figures 4b, 4d, and 4f the solid dots are the medians within each 1-hour local time bin, and the vertical bars connect the lower and upper quartiles.
 The normalized ΔBV amplitude in Figure 4a does not show a clear local time dependence, and there is a considerable spread of the data points. The median values in Figure 4b are near the zero in the 0400–2100 MLT sector. Comparing the upper and lower quartiles for each median, the degree of the data scatter is larger near midnight than near noon in ΔBV. Unlike the normalized ΔBV amplitude, the majority of the normalized ΔBD amplitude data points in Figure 4c fall into a narrow range of amplitude: mostly within the range between 0.0 to 0.3 in the 0000–1200 MLT sector and within the range between 0.15 and −0.2 in the 1200–2400 MLT sector. That is, there is a local time dependence, which is slightly asymmetric with respect to noon. Although local time variation of the normalized ΔBD amplitude is small, the median values in Figure 4d indicate a clear bipolar signature, i.e., positive (eastward) in the 0000–1200 MLT sector with a peak near 0600 MLT and negative (westward) in the 1200–2400 MLT sector with a peak near 1800 MLT. Note that this bipolar signature in the BD component is nearly consistent with the diurnal variation of the geosynchronous magnetic field in the azimuthal component [Kuwashima and Fukunishi, 1985]. Figure 4e shows a clear local time dependence of the normalized ΔBH amplitude: a maximum near local noon and a minimum near local midnight. This local time dependence is very similar to that in previous studies [Kokubun, 1983; Kuwashima and Fukunishi, 1985; Borodkova et al., 2005; Villante and Piersanti, 2008; Wang et al., 2009].
 A major feature to be noted in Figure 4e is the fact that the normalized ΔBH amplitudes in the prenoon sector (0900–1200 MLT) are larger than those in the postnoon sector (1200–1500 MLT). To present a statistical comparison of normalized ΔBH amplitudes in the prenoon sector and in the postnoon sector, the hourly medians in the 1200–2400 MLT sector are mirrored with respect to noon and plotted with the gray solid line in Figure 4f. We find that there is a strong morning-afternoon asymmetry. That is, the medians in the 0600–1100 MLT sector are larger than those in the 1300–1800 MLT sector. In our study we use medians instead of arithmetic means because the former is insensitive to a small number of data points that have extremely large magnitude. Note that the mean value of the normalized ΔBH amplitude for 9–12 MLT (12–15 MLT) is 1.29 (1.08), and a 95% confidence interval for the mean is ±0.08 (±0.07).
 In sections below we will examine if the morning-afternoon asymmetry observed at geosynchronous orbit is due to the diurnal universal time variation of the SYM-H or external sources. We also examine what causes a large scatter of the normalized SC amplitudes at geosynchronous orbit.
4.2. Universal Time Variation of ΔSYM-H
Cliver et al.  reported that geomagnetic disturbances have universal time (UT) variations and that the hourly time derivative of the Dst index has a hole at around 1300–1800 UT in summer (May, Jun, and July) (see Plate 3 in their study). In our study SC-associated geosynchronous magnetic field perturbations are normalized to ΔSYM-H. Thus, we need to examine if the morning-afternoon asymmetry of ΔBH/ΔSYM-H is due to the UT variation of ΔSYM-H.
Figures 5a and 5b show the UT variations of ΔSYM-H of the SC events observed at geosynchronous spacecraft near geographic east longitude 225° (GOES 9, GOES 10, and GOES 11) and 285° (GOES 8 and GOES 12), respectively. We note that geosynchronous spacecraft near geographic east longitude 225° were located near the equator (MLAT = ∼4°–6°) and near geographic east longitude 285° were located slightly off the equator (MLAT = ∼8°–11°). In Figure 5, the individual events are shown by small dots, the large dots are the medians within each 1-hour UT bin, and the vertical bars connect the lower and upper quartiles. Δ corresponding to ΔSYM-H is plotted in Figures 5c and 5d. Since the amplitude of the magnetospheric response is linearly related to the change of the square root of the solar wind dynamic pressure (Δ ) [e.g., Russell et al., 1992], the UT variation of ΔSYM-H is similar to that of Δ .
Figures 5e and 5f show the UT variations of the ΔBH amplitude at geosynchronous spacecraft near geographic east longitude 225° and 285°, respectively, normalized to Δ . The vertical dashed lines indicate the local noon of geosynchronous spacecraft. Figures 5g and 5h show the UT variations of the ΔBH amplitude at geosynchronous orbit near geographic longitude 225° and 285°, respectively, normalized to ΔSYM-H. The trends of the UT variation of ΔBH/Δ and ΔBH/ΔSYM-H are nearly identical. As shown in Figures 5e to 5h, the normalized ΔBH amplitude has a maximum value near local noon. Thus, the ΔBH normalized to ΔSYM-H shown in Figure 4e is not due to the UT variations of geomagnetic disturbances. We note that Villante and Piersanti  observed the morning-afternoon asymmetry of the SC-associated geosynchronous magnetic field amplitude normalized to Δ , which is consistent with our observations.
4.3. Morning-Afternoon Asymmetry and Solar Wind Conditions
 The magnetospheric compression on the dayside results from the increase in the magnetopause surface current associated with solar wind dynamic pressure enhancement. Therefore, the amplitude of the magnetic field perturbations associated with SC depends on the change in solar wind dynamic pressure. If the morning-afternoon asymmetry of the normalized ΔBH amplitude is directly related to solar wind dynamic pressure changes, the small amplitude ratio when GOES spacecraft was in the afternoon sector may be due to small changes in solar wind dynamic pressure. In Figure 6 we examine the relationship between Δ corresponding to the dayside SC events and the normalized ΔBH amplitudes. Although there is a considerable spread of Δ data points in each local time bin, the median values have no clear local time dependence when GOES moves from 0600 to 1800 MLT. These results suggest that the morning-afternoon asymmetry of the normalized ΔBH amplitude is not due to the solar wind dynamic pressure changes.
 The magnetospheric compression starts at the point where IP shocks first strike the magnetopause, and the amplitude of the magnetospheric magnetic field enhancement is greatest just inside at that point. If the front of the IP shock lies along the Parker spiral orientation (i.e., the shock normal is directed earthward and dawnward), it should strike first on the postnoon side of the magnetopause. Consequently, the amplitude of the magnetic field enhancement is larger in the postnoon sector than in the prenoon sector because the postnoon sector is closer to the source region. If the shock front is aligned with orthospiral direction, the magnetic field enhancement is larger in the prenoon sector than in the postnoon sector.
 In order to examine the relationship between the morning-afternoon asymmetry of the normalized ΔBH amplitudes and the shock front direction, we selected 149 shock normal vectors associated with our SC events from the published archival data in the study of the IP shock and geosynchronous magnetic field variations by Wang et al. . Figure 7 shows the distribution of the angle between the Sun–Earth line (−X GSM) and normal vector to the front of the IP shock. The positive (negative) angle indicates that the shock front faces dawnward (duskward). There is a peak at the positive angle of 0°–10°. About 80% (117 of 149 IP shock events) of the events had angles within ±30°. This indicates that most of IP shock fronts in Figure 7 hit the magnetopause in the 1000–1400 MLT sector. Out of 117 IP shock events 71 events had a positive angle and 46 events had a negative angle. That is, the shock front angles on the magnetopause are biased toward the positive angle. Thus large amplitudes of the normalized ΔBH could be expected in the postnoon sector. In Figures 8a and 8b we plot the normalized ΔBH amplitudes corresponding to 117 IP shock events and their median values in 3-hour local time bin. It is found that the morning-afternoon asymmetry in Figure 8b is similar to that in Figure 4f. This indicates that the morning-afternoon asymmetry in our study is not controlled by the shock front angle.
 It has been reported that geosynchronous magnetopause crossings are caused by magnetospheric compression due to solar wind dynamic pressure enhancement and by erosion of the geomagnetic field in the presence of southward IMF Bz [Rufenach et al., 1989]. Dmitriev et al.  suggested that magnetopause erosion operates more intensively in the prenoon sector, indicating a morning-afternoon magnetopause asymmetry during extreme solar wind conditions. In order to check whether or not there is the relationship between the magnetopause erosion and the morning-afternoon asymmetry of the normalized ΔBH amplitudes in Figure 4e, the local time distributions of the normalized ΔBH amplitudes are plotted in Figure 9a for northward IMF Bz and Figure 9b for southward IMF Bz. The IMF Bz corresponding to each SC event was determined with the value after the solar wind pressure jump. The local time dependence of amplitude ratio for the northward IMF Bz is similar to that for the southward IMF Bz, and the morning-afternoon asymmetry is evident for both IMF conditions. We therefore conclude that the morning-afternoon asymmetry is not associated with the IMF Bz condition.
4.4. Seasonal Variation
 Since SC-associated geosynchronous magnetic field perturbations are produced by changes in the magnetopause surface current on the dayside and in the cross-tail current on the nightside due to solar wind dynamic pressure changes [Kokubun, 1983], seasonal variations can be expected for the data presented in the dipole VDH coordinates if the current sources deviate from the magnetic equatorial plane and their deviation depends on the dipole tilt angle. We observed a strong seasonal dependence in the normalized ΔBV amplitude.
Figure 10 shows the normalized ΔBV amplitude for the four seasons. The data set was separated into a spring equinoctial season (February, March, April), a summer solstice season (May, Jun, July), a fall equinoctial season (August, September, October), and a winter solstice season (November, December, January). The solstices and equinoxes are the midpoints of these seasons. The data points in the equinoctial seasons are scattered around zero. However, most of data points are distributed below (above) zero in summer (winter). That is, there is a significant difference between summer and winter. We will discuss in the next section what causes the summer and winter dependence in the normalized ΔBV.
 In this section, we will first discuss a possible source for the morning-afternoon asymmetry of the SC-associated geosynchronous magnetic field variations by comparing our observations with geomagnetic field model. Second we will discuss the relationship between the SC-associated geosynchronous magnetic field perturbation and the displacement of the magnetopause and cross-tail currents with respect to the geosynchronous spacecraft position.
5.1. Partial Ring Current and Associated Geosynchronous Magnetic Field
 One of main results from our statistical analysis is the observation of the morning-afternoon asymmetry of the SC-associated geosynchronous magnetic field perturbations. That is, the SC amplitudes at geosynchronous orbit are larger in the morning sector than in the afternoon sector. The morning-afternoon asymmetry of geosynchronous magnetic field strength can be found from an empirical model developed by Tsyganenko and Sitnov , referred to as the TS05 model, including the partial ring current (PRC). The model includes the solar wind dynamic pressure (Psw) and Dst index as input parameters along with other parameters that specify the state of the solar wind and the magnetosphere. In order to examine the effect of the PRC, we compare the H-component geosynchronous magnetic fields obtained by considering all currents (solid curve) and by excluding the PRC (dashed curve) in the TS05 model in Figure 11. The input parameters were set to 1 nPa for Psw, −50 nT for Dst, and zero for IMF By and Bz. Using these input parameters, the PRC is located in the dusk sector, and the local time position of the PRC's center is around 1900 MLT (i.e., ∼77° magnetic longitude angle measured from the midnight meridian duskward) [Tsyganenko, 2002]. Figure 11 shows that the PRC is the most significant source of the asymmetric magnetic field at geosynchronous orbit. In other words other currents (magnetopause current, symmetric ring current, cross-tail current, and field-aligned current) in the TS05 model contribute to the symmetric magnetic field.
 We next examine whether the SC-associated morning-afternoon asymmetry observed at geosynchronous orbit is due to the PRC. For this purpose we use the TS05 model and compute the ratio between the TS05 model magnetic field perturbation at GOES position and TS05 model perturbation at local noon at L = 3 caused by solar wind dynamic pressure changes. Of the 422 SC events, the solar wind and IMF conditions are available for 169 SC events. Comparison of the TS05 model and observation for 169 SC events is shown in Figure 12. In each panel the small dots are individual events, and the large dots are medians in each 3-hour local time bin. The vertical bars connect the lower and upper quartiles. Following the median values, we find that the TS05 model median in 0600–0900 MLT is larger than that in 1500–1800 MLT. That is, there is a clear morning-afternoon asymmetry in the TS05 model. This agrees with the observation. However, the model median in 0900–1200 MLT is smaller than that in 1200–1500 MLT. This is opposite to the observations.
 To explain why the disagreement between TS05 model and observation occurs for the medians from 0900–1500 MLT, we examine the effect of the PRC for solar wind dynamic pressure variations in the TS05 model. The model fields at geosynchronous orbit for Psw = 1 nPa, 2 nPa, 5 nPa, and 10 nPa, respectively, are plotted as a function of local time in Figure 13. Note that the averages of Psw at the start time and at the end time of SC events in our study are about 2 nPa and 9 nPa, respectively. All other input parameters were set to the same values as those used in Figure 11. The solid dot in each curve indicates the maximum amplitude in the TS05 model BH component. Close comparison of the Psw change from Psw = 1 nPa to 10 nPa reveals that the location of the peak amplitude moves toward local noon. This result can be explained by the fact that the relative contribution of the PRC decreases as Psw increases in the TS05 model. That is, as Psw grows in magnitude, the dayside symmetric magnetic field, which is produced by the symmetric magnetopause current, increases because the magnetopause current is strongly affected by Psw. The contribution of the PRC in the TS05 model, however, remains constant because the PRC is not affected by Psw. Thus, the TS05 model provides more symmetric distribution near noon than observations under the high Psw.
 The PRC is closed by the Region 2 currents, generated if the gradients of the magnetic pressure and thermal pressure are not aligned [e.g., Wolf et al., 2007]. Recently, Nakano et al.  observed that the Region 2 current intensity increases as Psw increases. This indicates that PRC intensity is also dependent on Psw. When an IP shock passes over the Earth's magnetosphere, the degree of the day-night asymmetry of the magnetospheric magnetic field increases because the dayside magnetic field is increased by the enhanced magnetopause current and the nightside magnetic field is decreased by the enhanced tail current. This leads to more sunward deviation of the magnetic pressure gradient, and the large intensity of the PRC and Region 2 currents are expected during SC event. Thus, we suggest that the TS05 model underestimates the effect of the PRC under the high Psw.
5.2. SC-Associated Geosynchronous Magnetic Field Perturbations
 Perturbations in the D component (i.e., the azimuthal component) at geosynchronous orbit during the passage of IP shock show the local time dependence (see Figures 4c and 4d). The medians of the normalized ΔBD amplitude clearly show a bipolar signature with a positive peak near dawn and a negative peak near dusk. This bipolar signature (positive in the 0000–1200 MLT sector and negative in the 1200–2400 MLT sector) indicates that the magnetospheric field configuration is stretched antisunward during SC event. Figure 14 schematically shows the geosynchronous magnetic field perturbation in the D component at 0600 MLT and 1800 MLT. Note that D is positive eastward. In the figure we consider two snapshots of the field configuration before and after SC event. During the passage of IP shock the magnetospheric magnetic field lines on the morning and evening sides are distorted and swept back tailward [Nagano and Araki, 1984]. This azimuthal magnetic field deflection may be due to the magnetospheric compression associated with IP shock. At 0600 MLT ΔBD is positive, and at 1800 MLT ΔBD is negative as shown in Figure 14. The amplitude of the SC-associated ΔBD perturbation is at a maximum at 0600 MLT and at a minimum at 1800 MLT.
 Statistical results of the normalized ΔBV amplitudes plotted in Figure 10 show a clear difference between summer and winter. We suggest that these solstice seasonal variations of the normalized ΔBV amplitude are due to a displacement of SC-associated currents with respect to the GOES location. In summer GOES spacecraft on the dayside is located south of the SC-associated magnetopause surface current and on the nightside north of the SC-associated cross-tail current. In winter the spacecraft position with respect to the SC-associated magnetopause current (cross-tail current) is north (south) on the dayside (nightside). Figure 15 schematically illustrates SC-associated geosynchronous magnetic field perturbations at the summer solstice. At local noon a geosynchronous satellite is below the SC-associated magnetopause current (JMP) and at local midnight the satellite is above the SC-associated cross-tail current (JCT). The magnetic field perturbations caused by JMP and JCT give rise to a negative ΔBV near local noon and midnight, respectively. In Figure 10b, the data scatter of ΔBV is larger near midnight than near noon. This indicates that the location of JCT with respect to a geosynchronous satellite near midnight is more variable than that of JMP with respect to a geosynchronous satellite near noon. The situation at winter solstice is just the opposite. Thus, we have a positive ΔBV as shown in Figure 10b. We note that the location of JCT in Figure 15 cannot explain a positive ΔBH near midnight shown in Figure 4e. One qualitative explanation for the positive ΔBH near midnight can be given in terms of the radial location of the SC-associated JCT relative to geostationary orbit. Assuming that JCT lies below and earthward of geostationary orbit in the geomagnetic field configuration plotted in Figure 15, then we have a negative ΔBV and positive ΔBH near midnight.
 In order to examine the contribution to ΔBV from different current systems, we compare the ΔBV at midnight geomagnetic equator estimated from the magnetopause current (solid lines) and the cross-tail current (dashed lines) by using the TS05 model for solar wind dynamic pressure change from 2 nPa to 8 nPa at winter solstice and summer solstice in Figure 16. The solid circles indicate the ΔBV at geosynchronous orbit. At winter (summer) solstice, ΔBV is positive (negative). The polarity of ΔBV from the TS05 model is consistent with our observations. Comparing the amplitudes of ΔBV estimated from the magnetopause and cross-tail currents at geosynchronous orbit, it is found that the contribution to ΔBV from the magnetopause current is about 17–20% of the cross-tail current. Thus, ΔBV at midnight geosynchronous orbit is primarily controlled by the cross-tail current.
 We have performed a statistical analysis of the geosynchronous magnetic field variation normalized to SYM-H when SC is observed on the ground. The normalized amplitude in the H component at geosynchronous orbit clearly shows the morning-afternoon asymmetry. That is, the amplitudes in the morning sector are larger than those in the afternoon sector. We suggest that this morning-afternoon asymmetry is due to the enhancement of the PRC under the high Psw condition. A similar morning-afternoon asymmetry appears in TS05 magnetic field model, including the PRC. We have compared our observations and SC-associated TS05 model field variations. The medians in 0600–0900 MLT and 1500–1800 MLT from the model and observation show a similar trend but differed in the 0900–1500 MLT sector. The reason for this discrepancy is due to the fact that the symmetric magnetic field associated with the magnetopause current, strongly affected by Psw, increases for SC intervals. However, the PRC is not affected as much as the magnetopause current by Psw in the TS05 model. Thus, our observations suggest that the contribution of the PRC at geosynchronous orbit is much larger than that expected from TS05 model while the IP shock passes over the Earth's magnetosphere.
 Although the SC-associated geosynchronous magnetic field was normalized to the amplitude of SC from SYM-H to remove data scatter, the normalized geosynchronous fields show a large data scatter. We examined what caused such a large scatter. It is found that the amplitude of the normalized geosynchronous magnetic field variations in the V component strongly depends on the seasonal variations. Such seasonal variations are due to a displacement of SC-associated currents with respect to the geosynchronous spacecraft position.
 The GOES magnetic field data were provided from coordinated data analysis web (CDAWeb), http://cdaweb.gsfc.nasa.gov/cdaweb and from NOAA National Geophysical Data Center, http://goes.ngdc.noaa.gov. The solar wind and magnetic field data of ACE and/or WIND were obtained from the NASA CDAWeb site. This work was supported by the WCU program through NRF funded by MEST of Korea (R31-10016) and also supported by MEST, Space Core Technology Development Program (2012M1A3A3A02033285). Work of K.-H. Kim was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2010–0007393).
 Robert Lysak thanks the reviewers for their assistance in evaluating this paper.