The geomagnetic conjugacy of large-scale traveling ionospheric disturbances (LSTIDs) was studied using total electron content (TEC) data derived from GPS networks in Japan and Australia. The number of simultaneous occurrences of LSTIDs within ±1 hour in both hemispheres was 5 out of 20 (21) events in Japan (Australia) when Kp ≥ 5− and 0 out of 15 (10) when Kp ≤ 4+. As for the LSTIDs observed simultaneously in both hemispheres, the propagation velocities of equatorward LSTID were comparable between the two hemispheres, with differences of 10–40%. The crossing times at 30° geomagnetic latitude of the simultaneous LSTIDs over Japan and Australia were also different by several tens of minutes for all five events. These observational results indicate that the LSTIDs observed almost simultaneously in both hemispheres are not connected electromagnetically through the geomagnetic field but are generated by atmospheric gravity waves propagating to the equator independently in the two hemispheres.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Large-scale traveling ionospheric disturbances (LSTIDs) have been investigated for more than 3 decades [e.g., George, 1968; Davis and da Rosa, 1969]. The progress of these studies is summarized in the review papers by Hunsucker  and Hocke and Schlegel . LSTIDs have horizontal scales of more than 1000 km and periods of 30–180 min [Hunsucker, 1982]. They are believed to be ionospheric manifestations of the passage of atmospheric gravity waves that are generated at high latitudes by energy input from the magnetosphere. Studies of the generation and propagation of LSTIDs can clarify part of the energy flow from the magnetosphere to the low-latitude ionosphere.
 The recently developed imaging technique using multipoint GPS networks and/or 630-nm airglow imagers has been applied to study the dynamics of the ionosphere [e.g., Saito et al., 1998, 2002; Afraimovich et al., 2000a, 2000b, 2002]. Global coverage and continuous operation of the GPS networks of the International GPS Service (IGS) can provide global maps of total electron content (TEC) that are used to study LSTIDs [Ho et al., 1996, 1998]. Shiokawa et al. [2002, 2003c] investigated prominent LSTIDs during geomagnetic storms using data from both 630-nm airglow images and the GPS Earth Observation Network (GEONET) in Japan. Time sequences of two-dimensional TEC maps reveal precise spatial structures and temporal evolution of LSTIDs, as well as the damping rates of LSTIDs along the meridian [Tsugawa et al., 2003, 2004].
 In recent years, geomagnetic conjugate observations of traveling ionospheric disturbances (TIDs) using 630-nm airglow imagers have been reported. Otsuka et al.  first reported a symmetric pattern of nighttime medium-scale TIDs (MSTIDs) using 630-nm airglow images obtained at geomagnetic conjugate points in the northern and southern hemispheres. MSTIDs have often been observed at night over Japan, regardless of geomagnetic activity. These nighttime MSTIDs generally have wavelengths of a few hundred kilometers and propagate southwestward at 50–100 m/s [Saito et al., 2001, 2002; Ogawa et al., 2002; Shiokawa et al., 2003a, 2003b]. Shiokawa et al.  reported symmetric structures of MSTIDs and simultaneous LSTID events through 630-nm airglow observations at geomagnetic conjugate points in the northern and southern hemispheres during the third FRONT (F-region Radio and Optical Measurement of Nighttime TID) campaign. They concluded that the symmetric structures of MSTIDs indicate strong electrodynamic coupling between the two hemispheres through the geomagnetic field line. On the other hand, the timing of LSTID occurrence was different by ∼20 min between the two hemispheres.
 In this study, we investigated geomagnetic conjugacy of LSTIDs using 1-year TEC data in Japan from GEONET and in the geomagnetic conjugate region in Australia from IGS. Five LSTID events were observed almost simultaneously in both hemispheres, which is only 14% of all 35 LSTID events in Japan. There were time differences of several tens of minutes and velocity differences of 10–40% between the two hemispheres.
2. TEC Data in Japan and Australia
 TEC data analyzed in this study were derived from the GPS data of the GEONET, a dense and wide-area GPS network in Japan, and of GPS receivers in Australia belonging to the IGS. Each receiver provides carrier phase and pseudo-range measurements in two L-band frequency (f1 = 1575.42 MHz, f2 = 1227.60 MHz) every 30 s.
 Slant TEC, Is, measured along each GPS satellite-receiver path can be derived using the following equation [Mannucci et al., 1999]:
where L1 and L2 are the recorded carrier phases of the signal (converted to distance units), λ1n1 and λ2n2 are integer cycle ambiguities, and br and bs are satellite and receiver instrumental biases terms. Vertical TEC, I, can be obtained from the slant TEC using the following equation:
where χ is the satellite zenith angle at the thin-shell ionosphere located at the F region peak height (300 km as predicted by the IRI model). We did not use the TEC data from large satellite zenith angles (60–90°) to reduce errors due to conversion from slant to vertical TEC and cycle slips. Although we can obtain the relative change of TEC with a precision of about 0.1 TECU (1 TECU = 1016 electrons/m2) using equations (1) and (2), the integer cycle ambiguities and the both biases are unknown. It is reasonable to ignore these unknown values because this study is focused on perturbation components of TEC caused by LSTIDs. The perturbation components of TEC are derived by subtracting 60-min running averages.
 The GEONET consists of about 1000 GPS receivers whose locations are shown by the solid circles in Figure 1. Time sequences of high-resolution TEC maps over Japan derived from the GEONET data have provided a powerful tool to reveal the spatial structure and temporal evolution of MSTIDs [Saito et al., 1998, 2001, 2002; Ogawa et al., 2002] and LSTIDs [Shiokawa et al., 2002, 2003c; Tsugawa et al., 2003, 2004]. In this research, LSTIDs over Japan from January to December 2002 were identified with the technique developed by Tsugawa et al. , who used time sequences of perturbation TEC maps with a 10-min interval. Each map covers the area of 24–48°N latitude and 124–148°E longitude. The pixel size of the maps is 0.15° × 0.15°. The TEC value for each pixel is an average of perturbations for all satellite-receiver paths that cross the pixel at the thin-shell ionosphere.
 The geomagnetic conjugate region of Japan widely overlaps Australia. Although the network of GPS receivers in Australia is very sparse, geomagnetic conjugate points of four Australian GPS stations are located near 136°E longitude, passing through the center of Japan. These stations are Darwin (darw) [12.4°S, 131.9°E], Jabiru (jab1) [12.7°S, 132.9°E], Alice Springs (alic) [23.7°S, 133.9°E], and Ceduna (cedu) [31.9°S, 133.8°E]. Their geomagnetic conjugate points are shown in Figure 1 by the crosses. The three GPS receivers, darw (or jab1), alic, and cedu, have sufficient spacing to detect equatorward propagating LSTIDs. We derived TEC data from these four GPS receivers using equations (1) and (2) and analyzed time series of the perturbation TEC at each GPS receiver to detect southern hemispheric LSTIDs.
3.1. Occurrence of Simultaneous LSTIDs in Both Hemispheres
 Using the time sequences of TEC maps of 1-year GEONET data in 2002, we identified 35 LSTID events over Japan. The LSTIDs were defined as TEC enhancements larger than 0.5 TECU, extending horizontally for distances longer than 1000 km, and propagating over Japan within ∼3 hours. These criteria for detecting LSTIDs were determined by the definition of LSTIDs denoted by Hunsucker . All 35 LSTIDs were propagating equatorward. Out of the 35 events, 20 events were observed when Kp ≥ 5− and 15 events when Kp ≤ 4+. The Kp index represents the disturbance of the geomagnetic field in the subauroral region during each 3-hour period. In this study, we used the Kp values obtained 2 hours before the passage of the LSTIDs over Japan, taking into consideration the time lag between the generation time of LSTIDs in the auroral region (60–80°N) and the observation time in Japan (30–45°N).
 LSTIDs in Australia were also examined using TEC data from the three GPS receivers in Australia for the full year of 2002. We defined the LSTIDs in Australia as TEC enhancements larger than 0.5 TECU and clearly propagating equatorward over Australia within ∼3 hours. We identified 31 LSTID events in Australia (21 events when Kp ≥ 5− and 10 events when Kp ≤ 4+).
 Out of 35 LSTID events in Japan and 31 events in Australia, five events were simultaneously (within ±1 hour) observed in both hemispheres. All five simultaneous LSTIDs were observed during geomagnetically disturbed periods when Kp ≥ 5−. It is noted that there was no simultaneous observation of LSTIDs in both hemispheres when Kp ≤ 4+. Table 1 shows a summary of the occurrences of LSTIDs observed in Japan, Australia, and both hemispheres against the Kp index.
Table 1. Occurrence Number of LSTIDs in Japan, Australia, and Both Hemispheres during January–December 2002
Occurrence of LSTIDs
Kp ≤ 4+
Kp ≥ 5–
3.2. Simultaneous LSTIDs in Both Hemispheres
 The crossing times, phase velocities, and periods of the five simultaneous LSTIDs in both hemispheres were investigated in detail in the geomagnetic conjugate regions. We present two simultaneous LSTID events on 24 October 2002 and one event on 23 May 2002. Then, we summarize all five simultaneous events.
3.2.1. Two Events of 24 October 2002
Figures 2a–2s show a time sequence of two-dimensional maps of TEC perturbations every 20 min over Japan at 1400–1840 UT (2300–0340 LT) on 24 October 2002. The values displayed on the maps are deviations from 60-min running averages of TEC. Several consecutive LSTIDs are seen to propagate over Japan toward the equator. The average Kp index during this period was 6−, indicating strong geomagnetic disturbances due to the energy input into the high-latitude ionosphere from the magnetosphere. It is likely that this energy input was the source of the LSTIDs observed propagating equatorward over Japan.
 To investigate the temporal evolution of LSTIDs in detail, we analyzed TEC temporal variations along the horizontal distance represented by the arrow in Figure 1. The axis was defined along 136°E longitude from 52°N to 25°N latitude in order to pass through the center of Japan and around the conjugate points of GPS receivers in Australia. Because the enhancement of TEC due to the equatorward LSTIDs is considered almost uniform in the zonal direction over Japan, the perturbation component of TEC over Japan was averaged in the zonal direction for all longitudes from 124°E to 148°E to increase the spatial resolution of TEC data. This averaging also effectively reduces the TEC variations caused by MSTIDs.
 Recent papers reported that the azimuthal directions of the LSTIDs' propagation are not always coincide with the equatorial direction but have a direction deviation of ∼20° from the equatorward [Afraimovich et al., 2004, 2005; Afraimovich and Voeykov, 2004; Leonovich et al., 2004; Tsugawa et al., 2003, 2004]. In this study, however, the propagation directions of the LSTIDs in both hemispheres were assumed to be equatorward because we could not determine the precise propagation direction of the LSTIDs in the southern hemisphere using the sparse GPS receiver network in Australia.
Figure 3a shows the temporal evolution of TEC perturbations (≤60 min), which were derived from all satellite-receiver paths with zenith angles smaller than 60° and averaged in the zonal direction. There is no TEC data from GEONET stations between 0 km and ∼700 km in the horizontal distance. GEONET stations located south of Japan (≥2000 km) are very sparse and deviate from the axis of horizontal distance at 136°E. In the following analyses, we used the TEC data between 700 km and 2000 km in horizontal distance to increase the accuracy of analysis of the LSTIDs. Four TEC enhancements identified as LSTIDs are clearly seen to travel equatorward over Japan between 1400 and 2000 UT (2300–0500 LT) on 24 October 2002.
Figure 3b shows the TEC perturbations (≤60 min) observed at cedu, alic, and jab1 in Australia from south (top) to north (bottom). Different curves for each station were derived from different satellite-receiver paths whose zenith angles were smaller than 60°. Two LSTIDs propagating northward were detected by the three GPS stations in Australia at 1500–1800 UT (2400–0300 LT). No other clear LSTID was found in Australia during the plotted interval. The two LSTIDs designated by arrows in Figure 3 were observed almost simultaneously over Japan and Australia.
Figures 4a and 4b show temporal variations of the peak locations for the two simultaneous LSTIDs on 24 October 2002. The maximum and minimum of the TEC variations associated with the LSTIDs over Japan are represented by triangles and crosses, respectively. The squares with standard deviations represent the TEC peaks of the simultaneous LSTIDs in Australia, which are projected onto the 136°E axis in the northern hemisphere and averaged for all satellite-receiver paths. The solid and dashed lines are linear least-squares lines fitted to the TEC peaks of the LSTIDs in Japan and Australia, respectively. The horizontal dotted line represents 30° geomagnetic latitude, which corresponds to 37.1°N and 22.4°S geographic latitudes.
 The horizontal propagation velocities VH [m/s] of LSTIDs were derived from the gradient of the least-squares fitted lines in Figure 4. The velocities VH of the LSTIDs in Japan and Australia were 386 m/s and 314 m/s for the first LSTID event, and 411 m/s and 598 m/s for the second LSTID event, respectively. The periods T [min] of LSTIDs were determined from the average time difference between the maximum and minimum values at each horizontal distance. The periods T of the LSTIDs in Japan and Australia were 70 min and 57 min for the first LSTID event and 77 min and 76 min for the second LSTID event, respectively. The velocities and periods for the second LSTID in both Japan and Australia were larger than for the first one. The differences in VH between Japan and Australia were 72 m/s and 187 m/s for the first and second LSTIDs, which are 18% and 45% of the velocities of the LSTIDs over Japan, respectively. The periods T of the first LSTID were also different by 18% between Japan and Australia, although those of the second LSTID had little difference.
 The first LSTID crossed the 30° geomagnetic latitude at 1559 UT (0059 LT) in the northern hemisphere and 1549 UT (0049 LT) in the southern hemisphere. The second LSTID crossed at 1715 UT (0215 LT) in the northern hemisphere and 1653 UT (0153 LT) in the southern hemisphere. The crossing time of the first and second LSTIDs of 30° geomagnetic latitude were different by 10 min and 22 min, respectively, between the two hemispheres. Both LSTIDs came earlier over Australia than over Japan in the events on 24 October 2002.
Figure 5 shows the perturbations of TEC averaged within regions A (40–44°N, 136–140°E) and B (30–34°N, 130–134°E), which are displayed by boxes A and B in Figure 1. The solid and dashed curves represent TEC perturbations from GEONET receivers in Japan and from conjugate points of IGS receivers in Australia, respectively. The TEC perturbation data are averages of all the TEC data within regions A and B. Only the satellite-receiver paths with zenith angles smaller than 60° are used. Arrows in Figure 5 indicate the two LSTIDs observed almost simultaneously in Japan and Australia, as shown in Figure 3. Figures 5a and 5b clearly reveal that the TEC perturbations due to the LSTIDs in Japan are delayed by several tens of minutes from those at the conjugate regions of Australia. The periods of each cycle of TEC perturbations were also different between Japan and Australia.
 The peak-to-peak amplitudes of LSTIDs have little difference between Japan and Australia as shown in Figure 5. The amplitude of both simultaneous LSTIDs decreased as they traveled equatorward from region A to B. It is necessary to compare their amplitudes with absolute TEC in order to properly investigate the damping of the LSTIDs. However, absolute TEC over Australia has a large ambiguity compared with that over Japan because of the sparseness of GPS receivers. The damping of LSTIDs over Japan has been discussed in detail by Tsugawa et al. [2003, 2004].
3.2.2. Event of 23 May 2002
Figure 6 shows the temporal evolution of TEC perturbations in Japan and Australia at 1100–1600 UT (2000–0100 LT) on 23 May 2002, in the same format as Figure 3. TEC data from darw was used as the lowest-latitude station instead of jab1 because of a lack of data. The average Kp index during this period was 7−, indicating that strong geomagnetic disturbances continued. One LSTID, an intense TEC depression after a small TEC enhancement, is seen to propagate equatorward over Japan at 1230–1400 UT (2130–2300 LT), as represented by the arrow in Figure 6a. During this period, three TEC enhancements propagating equatorward were seen in Australia, as shown in Figure 6b. The second LSTID, shown by arrows, was nearly simultaneous with the LSTID in Japan.
Figure 7 shows temporal variations of the peak locations for the simultaneous LSTIDs in the same format as Figure 4. The horizontal phase velocities VH and periods T of the LSTIDs in Japan and Australia were 750 m/s and 1198 m/s, and 68 min and 52 min, respectively. The LSTIDs crossed the 30° geomagnetic latitude at 1235 UT (2135 LT) in the northern hemisphere and 1301 UT (2201 LT) in the southern hemisphere. The LSTID in Japan came 26 min earlier than in Australia.
 The perturbation component of TEC averaged within regions A and B during the simultaneous event on 23 May 2002, is presented in Figure 8 in the same format as Figure 5. The TEC perturbations corresponding to the simultaneous LSTIDs are shown by an arrow. The LSTID in Japan arrived earlier than that in Australia. The peak-to-peak amplitude of the LSTID in Australia was smaller than that in Japan. The overall TEC variations are quite dissimilar between the conjugate regions.
3.3. Summary of the Simultaneous LSTID Events
Table 2 summarizes the wave parameters of all five simultaneous LSTID events in Japan and Australia, including the date of event, crossing time at geomagnetic latitude of 30°, horizontal phase velocity, period, delay of LSTID in Japan to that in Australia, and averaged Kp index. The interplanetary magnetic field (IMF) in GSM coordinates measured by the ACE satellite is also given. The IMF data was time-shifted to X = −10RE (Earth radii) [Østgaard et al., 2004] and averaged over 0–60 min before when LSTIDs over Japan were assumed to be generated at 65° geomagnetic latitude.
Table 2. Summary of Wave Parameters of All the Simultaneous LSTID Events in Japan and Australia
Event in 2002
Crossing Time, UT (LT)
IMF-Bx, By, Bz, nT
[−5, 10, 5]
[−2, 3, −18]
24 Oct (1)
[−8, 7, −2]
24 Oct (2)
[−7, 3, −4]
[−12, 25, 10]
 In the events of 23 May and 27 November the phase velocities of LSTIDs were large in both Japan and Australia. On the contrary, when the phase velocity was small in one hemisphere, that in the other hemisphere was also small. The phase velocities of LSTIDs were different by 10–40% between the two hemispheres for each event.
 These velocity differences are sufficiently large so that we believe the differences are valid even if their propagation directions deviated by ∼20° from the equatorward. The crossing times of LSTIDs at 30° geomagnetic latitude were also different by several tens of minutes between the two hemispheres. The periods of LSTIDs were different by 20–35% between Japan and Australia for 23 May, 4 September, and the first of the 24 October events, while they were almost the same for the other two events. The averaged IMF orientation was Bx < 0 and By > 0 for all the events. IMF Bz was not always negative, even during geomagnetic storms. The strengths of the IMF components were very different in these five events.
 As summarized in Table 2, the propagation velocities and crossing times of 30° geomagnetic latitude for the simultaneous LSTIDs were different by 10–40% and several tens of minutes between the two hemispheres, respectively. These characteristics are different from the geomagnetic conjugate features of MSTIDs, the temporal and spatial symmetric patterns along the geomagnetic field line [Otsuka et al., 2004; Shiokawa et al., 2005]. The crossing time differences of several tens of minutes are comparable to that reported by Shiokawa et al.  using 630-nm airglow imagers. From these results, we conclude that the simultaneous LSTIDs in both hemispheres are not connected electromagnetically through the geomagnetic field line but are independently generated by atmospheric gravity waves propagating equatorward.
 As shown in Table 1, 20 LSTID events in Japan and 21 events in Australia were observed during the geomagnetically disturbed period (Kp ≥ 5−), and 15 LSTID events in Japan and 10 events in Australia were observed even during the quiet period (Kp ≤ 4+). However, only five simultaneous LSTID events were observed in both hemispheres within ±1 hour when Kp ≥ 5−, and no conjugate LSTID event was identified when Kp ≤ 4+. Although there is a possibility that small amplitudes of the quiet-time LSTIDs could not be identified by a sparse GPS receiver network in Australia [Tsugawa et al., 2004], the simultaneous occurrence of LSTIDs in the two hemispheres was not frequent, even during geomagnetic storms.
 This small number of simultaneous LSTIDs in the two hemispheres has three possible reasons: (1) a different damping rate of LSTIDs propagating from high latitudes to midlatitudes, (2) a difference in the propagation velocities of LSTIDs in the two hemispheres, and (3) hemispheric asymmetries of auroral energy input at high latitudes.
 Background ion density is one of key parameters that control the damping rate of LSTIDs. We checked absolute TEC, which is derived from GEONET using a technique of Otsuka et al.  at the GEONET-TEC Web site (http://stegps.kugi.kyoto-u.ac.jp/). The difference in absolute TEC between summer and winter in Japan was generally less than 10 TECU. The difference would be much smaller at the equinox. These small differences in background TEC could not make a large asymmetry in the damping of LSTIDs [Tsugawa et al., 2003, 2004].
 If the propagation velocities of LSTIDs are significantly different between the two hemispheres, we may miss the simultaneous LSTIDs by the criterion of less than 1-hour difference. The horizontal group velocity of atmospheric gravity waves Vgx depends on their wave parameters and the background neutral temperature T0, as given by the following equation [Hines, 1960]:
where ω is the frequency, kx is the horizontal wave number of gravity waves, γ is the ratio of specific heats, g is gravitational acceleration, and R is the gas constant. We estimated the dependence of group velocity Vgx on T0, ω, and kx using equation (3). For a large-scale gravity wave with 2000 km horizontal wavelength and 70 min period, Vgx varies ∼10% when ω and kx changes 10%. On the other hand, Vgx varies less than ∼0.01% when T0 changes 10%. This estimate indicates that the period and wavelength of atmospheric gravity waves are much more important in the determination of the propagation velocity than the background temperature of the neutral atmosphere. The period and wavelength would be primarily determined by the source condition in the auroral zone. Thus the difference in auroral input between the two hemispheres may produce different propagation velocities of LSTIDs and contribute to the low occurrence of simultaneous LSTIDs in both hemispheres.
 As for the source location of LSTIDs, Østgaard et al.  reported, using satellite auroral imagers, that IMF By and Bz components control the asymmetric location of substorm onset and auroral features in the conjugate hemispheres. Their global auroral map in the two hemispheres revealed that the auroral location is usually not symmetric, and the longitudinal displacement between the two hemispheres can be as much as ∼1500 km. We examined whether such a longitudinal displacement of source region could cause the large time lag or the asymmetry of the LSTIDs between the two hemispheres. Figure 9 shows the time lag of LSTIDs between the two hemispheres at 30° geomagnetic latitude as a function of longitudinal displacement of the source regions of LSTIDs. It is assumed that the source is localized at 65° geomagnetic latitude and that LSTIDs propagate at a constant velocity in both hemispheres along great circle path from the source to the observation point. The solid, dashed, and dotted curves represent the time lag for the velocity of 400, 600, and 800 m/s, respectively. This calculation clarified that the 1500 km longitudinal displacement of the source region could cause the time lag of 12–24 min for the LSTIDs propagating at 400–800 m/s. Although these time lags are comparable to those of the observed simultaneous LSTIDs, such longitudinal asymmetry of the source regions would not be responsible for the larger time lag than ±1 hour, the criteria of the simultaneous LSTIDs. The time lags of the five simultaneous LSTID events in fact seems not to be independent of the orientations and strengths of IMF By and Bz.
 The asymmetry of auroral power between the summer and winter hemispheres was reported by Shue et al. . Lu et al.  simulated the response of the ionosphere during the geomagnetic storm event of 10 January 1997, using the thermosphere-ionosphere-electrodynamics general circulation model with observed auroral zone data as inputs. Their simulation results revealed several features of LSTIDs propagating to the equator after the energy inputs into the northern hemisphere. However, the southern hemispheric LSTIDs did not have one-to-one correspondence between the energy input and LSTID generation at high latitudes and could not be identified at midlatitudes. The result of their simulation seems to be consistent with the present study.
 The geomagnetic conjugacy of LSTIDs was investigated using TEC data from GPS receiver networks in Japan and Australia. The observational results are summarized as follows:
 1. The number of simultaneous occurrences of LSTIDs within ±1 hour in both hemispheres was 5 events out of 20 events in Japan and 21 events in Australia when Kp ≥ 5−, and 0 event out of 15 events in Japan and 10 events in Australia when Kp ≤ 4+.
 2. As for the simultaneous LSTIDs in both hemispheres, the phase velocity is comparable between the two hemispheres, with differences of 10–40%.
 3. The crossing times at 30° geomagnetic latitude of the simultaneous LSTIDs were different by several tens of minutes.
 These observational results indicate that the LSTIDs are not connected electromagnetically through the geomagnetic field between the two hemispheres but are generated by atmospheric gravity waves, which propagate to the equator independently. The asymmetry of LSTID appearance at midlatitudes in the two hemispheres might be caused by asymmetry of the energy input in the auroral zone rather than the background condition of the ionosphere and the atmosphere.
 We acknowledge the Geographical Survey Institute (GEONET) of Japan and the International GPS Service for providing the GPS data. The ACE data was obtained from the CDAWeb (http://cdaweb.gsfc.nasa.gov/). This work is supported by a Grant-in-Aid for Scientific Research (13573006) and by the Ministry of Education, Culture, Sports, Science and Technology, Japan (Dynamics of the Sun-Earth-Life Interactive System, G-4, the 21st Century COE Program). T.T. was supported by Research Aid of the Inoue Foundation for Science and is supported by a grant of Research Fellowships of the Japan Society for the Promotion of Science (JSPS) for Young Scientists.
 Arthur Richmond thanks E. L. Afraimovich and William Bristow for their assistance in evaluating this paper.