Medium-scale traveling ionospheric disturbances (MSTIDs) whose peak-to-peak amplitude was larger than 20 TECU (=1016el/m2) were observed at midlatitude during the geomagnetic storm on 10 November 2004. This amplitude was more than 10 times larger than that of the average MSTID. High-resolution data of the GPS Earth Observation Network (GEONET) clarified the characteristic of the total electron content (TEC) disturbances over Japan on 10 November 2004. The disturbances started around 1000 UT in the central part of Japan. The maximum of TEC temporal change was 7.2 TECU in 30 s. The disturbances had several wave fronts which extended from northwest to southeast and propagated from northeast to southwest. TEC data around Japan revealed that the disturbances were mainly observed from 18°N/S to 34°N/S of the geomagnetic latitude in the both hemispheres. Since those characteristics were similar to those of MSTIDs in spite of the unusual large amplitude, the MSTIDs are referred as “super-MSTIDs” in this paper. TEC variations of the super-MSTIDs were also observed at 460 km altitude by the GRACE satellite. The ion density fluctuations of the super-MSTIDs were observed in situ by the CHAMP and DMSP-F15 satellites, which flew at 360 km and 850 km, respectively. It is found that the plasma density variations of the super-MSTIDs occurred mainly above 360 km altitude. The characteristics that distinguish the event from plasma bubbles are its successive wave fronts, constant northwest-southeast direction along which the wave fronts stretched, and late local time of the occurrence. It is found that the uplift of the ionosphere around sunset excited the super-MSTIDs at midlatitudes. The uplift was attributed to the strong eastward electric field during the geomagnetic storm.
 Various kinds of ionospheric phenomena have been observed during geomagnetic storm periods, such as storm enhanced density, traveling ionospheric disturbance (TID), and plasma bubble. Penetration electric field, disturbance dynamo, neutral wind excited by Joule heating [e.g., Prölss, 1995], and composition change of atmosphere can effect on the electrodynamics of the ionosphere [e.g., Heelis, 2004; Prölss, 1995; Richmond et al., 2003; Fuller-Rowell et al., 1994]. The change of the electrodynamics can cause positive and negative storm in the ionosphere. Modulated electric field and neutral wind during geomagnetic storms could cause TIDs and plasma instabilities, which make ionospheric irregularities like plasma bubbles.
 Medium-scale TID (MSTID) at midlatitudes have been observed in different longitudinal sectors and with several observational instruments in recent years [e.g., Kubota et al., 2000; Saito et al., 2001; Shiokawa et al., 2003; Hernández-Pajares et al., 2006]. MSTID is a wave like structure whose wave length is several hundred kilometers and which propagates at the speed of 100–200 [m/s] [Garcia et al., 2000b; Saito et al., 2002]. Characteristics of MSTIDs at midlatitudes have been revealed with GEONET data, with which two dimensional total electron content (TEC) maps are available in high resolution [Saito et al., 1998]. It is reported that nighttime MSTIDs observed with GEONET data consist of several wave fronts with northwest-southeast direction, and propagate from northeast to southwest. Peak-to-peak amplitude of TEC disturbances of MSTIDs were 1–2 TECU. It was reported that there was no obvious relation between the geomagnetic activity and the activity of MSTID detected by GEONET data [Saito et al., 2001]. There are different results from other longitudes that are not considered, e.g., results from the various World Atmospheric Gravity campaigns in the late 1980s [Williams et al., 1988; Bristow and Greenwald, 1994].
 In this paper, high-resolution data of GEONET were utilized to study MSTID during a geomagnetic storm. To extend the observations, global networks of ground-based GPS receivers such as International GNSS Service (IGS) were used. The coverage of the observational area by ground-based GPS observation was from 90°E to 170°E, and from 45°N to 50°N in the geographic coordinates. Data measured by several satellites were used in order to compare with the ground-based observational data.
 MSTIDs presented in this paper are observed during the geomagnetic storm occurred on 10 November 2004. The storm followed the storm which started on 7 November 2004. Figure 1 shows Dst index from 6 to 11 November. On 10 November, Dst was minimum at 1000 UT and −281 nT. It was observed that MSTID occurred from 0900 UT to 2200 UT in the region between 90°E and 170°E, which is represented by the horizontal bar on 10 November in Figure 1.
 Data of ground-based GPS receivers were used in this study. From ground-based GPS receivers' data, total electron content (TEC) between a GPS satellite and a receiver was derived. Most of the data were obtained by GPS Earth Observation Network (GEONET), which is a dense network of 1,200 receivers over Japan operated by Geographical Survey Institute. The distribution of the receivers is shown in Figure 2. Filled circles represent locations of receivers. The average distance between receivers is about 25 km. Each GPS receiver measures signals about eight GPS satellites in every 30 s. TEC map over Japan with 0.15° × 0.15° resolution is obtained [Saito et al., 1998]. Additional GPS receivers data around the Japanese longitudinal sector were provided by International GNSS Service (IGS), the networks of Scripps Orbit Permanent Array Center (SOPAC), Japan Agency for Marine-Earth Science and Technology (JAMSTEC), and Solar-Terrestrial Environment Laboratory, Nagoya University. The locations of the receivers are shown in Figure 3. The receivers located in the region of 90°E–170°E and 45°S–50°N in the geographic coordinates. Out of 25 receivers in Figure 3, two receivers, which are 0042 and 0100, are GEONET receivers. The list of the station code, name of its network, geographic longitude and latitude, and geomagnetic latitude is shown in Table 1. The sampling rate of each receiver is 30 s.
The station code, the network which provides the data, the geographic longitude and latitude are listed. In the sixth column, ROTI averaged from 1900 LT to 2300 LT at each station are listed.
7.2 × 10−1
4.1 × 10−1
3.1 × 10−1
2.3 × 10−2
3.2 × 10−2
1.8 × 10−1
4.3 × 10−1
4.2 × 10−1
2.3 × 10−1
1.2 × 10−1
2.6 × 10−1
4.3 × 10−1
1.1 × 10−2
8.1 × 10−1
1.4 × 10−2
2.0 × 10−2
2.8 × 10−2
1.1 × 10−1
3.5 × 10−2
3.6 × 10−1
2.6 × 10−2
5.2 × 10−1
1.9 × 10−1
3.4 × 10−1
 Data of in situ measurements by satellites was also used in this study. The names of the satellites, type and the local times of observations, and the altitude used in this study are given in Table 2. Ion density was measured by the Defense Meteorological Satellite Program (DMSP) F15 satellite and the Challenging Minisatellite Recovery and Climate Experiment (CHAMP) satellite. The altitudes of observations by DMSP and CHAMP were about 850 km and 360 km, respectively. Sampling intervals of the measurements by DMSP and CHAMP are 1 and 15 s, respectively. Local time of DMSP F15 observation was about 2100. Local time of CHAMP observation used in this study was from 0205 to 0225. TEC, which was observed by the Gravity Recovery and Climate Experiment (GRACE)-A satellite, was used in the study. TEC observed by the GPS receiver on the GRACE-A satellite is TEC between the satellite and GPS satellites, whose altitude was 460–470 km and 20,200 km altitude, respectively. Sampling interval, and local time of the observation was 10 s and 2340–2350, respectively. Two dimensional map of airglow emission was obtained from the Global Ultraviolet Imager (GUVI) instrument on Thermosphere Ionosphere Mesosphere Energy and Dynamics (TIMED) satellite [Christensen et al., 2003]. The emission, which was 135.6 nm emission, was observed by scanning the field of view on the each pass. Emission of 135.6 nm is that of radiative recombination of the oxygen ion [Hanson, 1969]. The brightness of the emission varies as the product of the electron density and the oxygen ion concentration, which is approximately equivalent to the square of the electron density in the F region. Local time of the airglow emission used in this study was from 0010 to 0215.
Table 2. Observations From Spacecraft Used in This Studya
Sampling Rate (s)
Names of the spacecraft, target of the observation, altitude of the spacecraft, and local time of the observation are listed.
 TEC data measured at 0042 station in Japan from 0930 UT to 1430 UT are shown in Figure 4. The data were derived for the ray path between the satellite number (Pseudo Random Noise: PRN) 31 and the 0042 ground-based GPS receiver (36.5°N, 140.6°) whose elevation angle was larger than 45°. Dots in Figure 4 represent each data sampled every 30 s. Instrumental biases which were included in TEC data derived from the carrier phase delays and pseudoranges of the GPS signals were removed with a least square fitting method [Otsuka et al., 2002]. The altitude of ionosphere was assumed to be 400 km. TEC data were converted to the vertical TEC with correcting the effect of the radio wave raypath length change inside of the ionosphere. Amplitude and phase scintillations are observed at frequency that is higher than the Fresnel frequency [Pi et al., 1997]. Fresnel frequency for L1 of GPS signal is from 0.08 Hz to 0.36 Hz at midlatitude [Pi et al., 1997]. The sampling rate, 30 s, that is 0.033 Hz, is lower than the Fresnel frequency. Therefore, the scintillation could be hardly observed with the 30 s sampling ground-based GPS measurement. As a rare case, when scintillation affects on the ground-based measurement, the GPS receiver losses lock on the GPS signals because of the sudden changes occur in the carrier phase, which is called a cycle slip. Integer ambiguities in phase measurements, which could be caused by cycle slips, were removed by using the measured phase pseudoranges [Saito et al., 1998]. Since there were no isolated data or gaps in the data of Figure 4, scintillation did not affect significantly on this observation. It is seen that intense TEC fluctuations started around 1015 UT, which is 1915 LT, in Figure 4. TEC varied in the range from 4.5 TECU to 48.1 TECU. The largest change of TEC within 30 s was 7.2 TECU which was observed around 1150 UT. The fluctuations were detected until the end of the observation of this satellite-receiver pair.
 Two-dimensional maps of TEC over Japan at 1210 UT (2110 LT), 1230 UT (2130 LT), and 2230 UT (0730 LT) are shown in Figures 5a, 5b, and 5c, respectively. TEC derived with the same procedure as Figure 4 was projected at the ionospheric height. Spatial resolution of the maps is 0.15° × 0.15°. In Figure 5a, there were some TEC wave fronts which were stretching from northwest to southeast. Between the wave fronts, there were some regions where TEC was larger than 30 TECU and smaller than 30 TECU, which are referred to as “TEC enhanced region” and “TEC depletion region,” respectively. One of the TEC depletion regions which extended from (134°E, 38°N) to (137°E, 34°N) is marked A in Figure 5a. One of the depletion region A' in Figure 5b, which extended from (132°E, 37°N) to (135°E, 33°N), was TEC depletion region A in Figure 5a. It is found that the wave fronts propagated from northeast to southwest. At 2230 UT (0730 LT), wave fronts whose direction was northwest-southeast were also observed, although the peak-to-peak amplitude was smaller than those observed at 1210 UT (2110 LT) and 1230 UT (2130 LT). Characteristics of the structure such as the direction along which the wave fronts stretched and movement of the structure were similar to those of MSTIDs that have been frequently detected by GEONET. Peak-to-peak amplitudes, which were more than 20 TECU, were more than 10 times of those of MSTID that have been reported in previous studies [e.g., Saito et al., 2002; Tsugawa et al., 2006]. Because of the large amplitude, this MSTID-like fluctuation was named “super-MSTIDs” in this paper.
 Characteristics of these “super-MSTIDs” were investigated with two-dimensional TEC data in detail. The spatial variations of TEC along the line represented by the thick line in Figure 2 are shown in Figure 6. The line, which is from (27.8°N, 126.1°E) to (41.2°N, 142.5°E), was almost perpendicular to the wave fronts of the super-MSTIDs in Figure 5. The azimuthal angle of the path was 50 degree clockwise from north. TEC values were averaged along the direction orthogonal to the path in order to smear small-scale structures because small-scale structures are out of this topic. The center of the area where the TEC data were averaged was the path, and the width of the area was 220 km. Averaged TEC is plotted against the longitude of the path in Figure 6. TEC data used in Figure 6 were that from the PRN7 satellite. The solid and dashed lines show the spatial profiles of TEC at 1210 UT (2110 LT) and 1230 UT (2130 LT), respectively. Positive and negative peaks of the TEC variation are labeled A, A', B, C, B', C', and D'. The wavelength at 1210 UT, which is the distance between two positive peaks B and C, which appeared around 134.8°E and 138.0°E, respectively, was 290 km. There were variations in wavelengths at 1230 UT; the distance between B' and C' was about 100 km, and the distance between A' and D' was about 500 km. The range of wavelengths, from 100 to 500 km, covered wavelength observed at other local time.
 Arrows below TEC profile around 136.2°E and 133.7°E in Figure 6 indicate the TEC minima which traveled from northeast to southwest. The TEC minima correspond to the TEC depletions of A and A' in Figures 5a and 5b, respectively. The distance between two minima was 220 km. Therefore the velocity of this minimum along the direction orthogonal to the path was 180 m/s around 1220 UT. The region where the super-MSTIDs appeared was investigated with ground-based GPS receivers which are shown in Figure 3. Time series of TEC from 0700 UT to 2400 UT on 10 November 2004 are shown in Figure 7. The stations were YSSK (142.7°E, 47.0°N), 0042 (140.6°E, 36.5°N), BJFS (115.9°E, 39.6°N), 0100 (127.8° E, 26.1°N), and TOW2 (147.1°E, 19.3°S). The data from all satellites whose elevation angle was larger than 45° are plotted. Stations of 0042 and 0100 are those of GEONET. Stations of YSSK, BJFS, and TOW2 are those of IGS. Instrumental biases of IGS receivers were removed from the TEC data. Vertical lines between 0800 UT and 1030 UT, and dashed vertical lines between 1800 UT and 2200 UT in Figure 7 indicate the sunset and sunrise time, respectively, at the 300 km altitude at each station.
 At 0042 station, TEC enhancement was observed from 0900 UT. It increased by about 30 TECU. The onset of the enhancement was around the sunset at the 300 km altitude. After the TEC enhancement, TEC fluctuations of the super-MSTIDs which were seen in Figures 4, 5, and 6 appeared around 1000 UT. The fluctuation continued until 2200 UT while peak-to-peak amplitude changed. The peak-to-peak amplitudes were more than 20 TECU from 1100 UT to 1300 UT. The amplitude varied with time and decreased after 1700 UT. It was more than 10 TECU until 1900 UT. Fluctuations of the super-MSTIDs observed continuously from 1030 UT to 2200 UT. TEC variations at YSSK station are shown on the first panel of Figure 7. The location of YSSK is north of 0042 station as shown in Figure 3. There were little TEC variations in YSSK. The north edge of the super-MSTIDs was south of YSSK. On the third panel, TEC variation at BJFS station is represented. The location of BJFS station is west of 0042 station as shown in Figure 3. Latitudes of BJFS and 0042 are almost the same. After the sunset, which was around 1040 UT, TEC increased by about 30 TECU. Before TEC began to decrease, small MSTID-like variation of TEC was seen. It was found that activity of the super-MSTIDs was low at west of 140.6°E. The fourth panel of Figure 7 shows the TEC variations at 0100 station. The location of 0100 station is closer to the equator than the other stations. The geomagnetic latitude of 0100 station was 16.4°N. After the sunset which was about 1000 UT, TEC increased by about 40 TECU. The enhancement was larger than that at 0042 and BJFS stations. Large TEC depletions were seen about two hours after the sunset. The depletions were about 30 TECU. Only one depletion was observed. Time scale of the depletion was more than one hour. Those features of the TEC depletion were different from those of the super-MSTIDs. Spatial structure of the super-MSTIDs had more than two maxima and minima. Time scale of the super-MSTIDs was less than one hour. This variation of TEC observed at 0100 station could be plasma bubble which is often observed in the equatorial region. Four hours after the sunset, MSTID-like fluctuation of TEC was observed. The peak-to-peak amplitude of this fluctuation was less that of 0042. The activity of the super-MSTIDs was smaller near the equatorial region than at the midlatitude region.
 The fifth panel of Figure 7 shows TEC variation at TOW2 station which located in the southern hemisphere. It located almost geomagnetically conjugate point of the 0042 station. TEC enhancement was observed from 0830 UT. The enhancement preceded the sunset which was 0940 UT. Note that the sunset was earlier in the northern hemisphere than in the southern hemisphere at the same longitude. A TEC variation of the super-MSTIDs was observed at 0900 UT which was after the TEC enhancement and before the sunset. The peak-to-peak amplitude was more than 20 TECU. The super-MSTIDs terminated earlier than that in the northern hemisphere. To study the super-MSTIDs' activity, Rate Of TEC change Index (ROTI) was calculated. ROTI is standard deviation of time differential of TEC calculated every five minutes. The time window of standard deviation reflects the spatial scale size of the phenomenon. The spatial scale of the phenomenon which is detected by five minutes ROTI corresponds to the displacement of the pierce point of the GPS radio wave at the ionospheric altitude for five minutes. When the ionospheric altitude is assumed 400 km, velocity of the pierce point of the GPS radio wave at the ionospheric altitude is approximately 60 m/s around the zenith, therefore the spatial scale of the phenomenon is about 20 km [e.g., Pi et al., 1997; Nishioka et al., 2008]. ROTI with time window of five minutes is used in studies for ionospheric irregularities at midlatitudes and low latitudes [e.g., Pi et al., 1997; Beach et al., 1999; Nishioka et al., 2008]. ROTI is an index of activity of ionospheric irregularities such as MSTIDs and plasma bubble. To study the activity of the super-MSTIDs, ROTI was averaged from 1900 to 2300 LT at each ground-based GPS station shown in Figure 2. The averaged ROTIs are shown on the sixth column of Table 1. The value of ROTI is indicated as size of a circle at each station in Figure 8. Size of the circle at each station in Figure 8 is proportional to the value of ROTI. ROTIs of 0042, TOW2 stations where the super-MSTIDs were found in Figure 7 were 0.72 [TECU/min] and 1.1 [TECU/min], respectively. On the other hand, ROTIs of YSSK, BJFS and 0100 stations where the super-MSTIDs were not found in Figure 7 were 0.34 [TECU/min], 0.31 [TECU/min], and 0.41 [TECU/min]. ROTIs where the super-MSTIDs were observed were larger than those where the super-MSTIDs were not observed with ground-based observation. The averaged ROTI can be a proxy of the activity of the super-MSTIDs. Activity of the super-MSTIDs was higher around central part of Japan, which is from 26°N to 42°N in the geographic latitude, than north/south of it. This area corresponds to from 18°N to 34°N in the geomagnetic latitude. The geomagnetically conjugate area of this latitude in the southern hemisphere is from 10°N to 26°N in the geographic latitude, which is coincident with the area where the averaged ROTI had large values of the super-MSTIDs' activity in the southern hemisphere. The western boundary of the region where the super-MSTIDs appeared is expected to be around 120°E. At the stations west of 120°E like LHAS, SIS2, KARR, etc, ROTI was less than 3.5 × 10−1 [TECU/min].
 The variation of TEC between the GRACE-A satellite and the GPS satellite PRN7 from 1336 UT to 1400 UT is shown in Figure 9. The trajectory is shown with the long dash dotted arrow in Figure 8. Zenith angle between the GRACE-A satellite and the GPS satellite was displayed on the bottom. Although the TEC value includes instrumental bias that comes from the receiver and the transmitter, it measures the TEC variation precisely. The data where the GRACE-A satellite flew over between 26°N and 42°N in the geographic coordinates, which are 18°N and 32°N in the geomagnetic coordinates, are shown by the shaded area in Figure 9. Local time of the observation was from 2339 to 2348. Fluctuations were observed at the shaded area. The amplitude of fluctuations was 5–12 TECU. The super-MSTIDs were observed by the ground-based GPS receivers at this local time in Figure 7. Therefore, fluctuations observed over Japan in Figure 9 are believed to be the super-MSTIDs. The amplitude of fluctuations by the ground-based observations were 8–12 TECU. In the southern hemisphere, fluctuations were observed, though there were some data gaps.
Figure 10 shows the ion density between 40°S and 50°N, and 96.7°E and 171.7°SE in the geographic coordinates at 850 km altitude observed by the DMSP-F15 satellite. Figure 10 (top, middle, and bottom) show the data of 1005–1031 UT, 1147–1213 UT, and 1329–1355 UT, respectively. The trajectories of these passes are shown with long solid arrows in Figure 8. The locations of DMSP-F15 at the start time on each pass, 1005, 1147, and 1329 UT, are represented at the end of the arrows in Figure 10. Data when the satellite flew over Japan (26°N–42°N and 18°N–34°N, in the geographic and geomagnetic coordinates, respectively) are represented by the shaded areas in Figure 10 (top and middle). In the shaded areas on the first and the second passes, where the super-MSTIDs were detected by the ground-based observations, ion density fluctuations were seen in the DMSP-F15 data. The region where ion density fluctuations were observed is indicated with the rectangular box in Figure 10 (top). The scale of the rectangular box in Figure 10 in the manuscript is about 600 km. The scales of ion density fluctuations were less than 200 km. The length orthogonal to the wave fronts that corresponds to the track-along length 200 km is about 70 km. This 70 km is smaller than the wavelength of the super-MSTIDs observed with the ground-based GPS measurements, which was 100–500 km. The difference of wavelength between the DMSP and ground-based GPS measurements could be caused by the difference of observational altitudes.
 The ion density of depletion region was 0.5–1.5 × 105 [el/cm3], while ambient density was about 2.0 × 105 [el/cm3]. The ratio of the depletion density to the ambient density was from 25 to 75%. In the southern hemisphere, fluctuations were observed between 40°S and 20°S in the geomagnetic latitude. The region where ion density fluctuation was observed by the DMSP-F15 satellite corresponded to the region where TEC disturbance of the super-MSTIDs was observed with ground-based GPS receivers. The ion density of depletion region in the southern hemisphere was 2.1–2.4 × 104 [el/cm3] with ambient density 3.0–4.2 × 105 [el/cm3]. The ratio of the depletion to the background density was 50–80%. Several depletions were observed in the lower latitudes than 20°N in the magnetic latitude in the northern and southern hemispheres. The density in the depletions was 1.1 × 103 [el/cm] while the background density was 1.2 × 105 [el/cm3]. The ratio of depletion to the ambient density was more than 99%. This depletions corresponded to plasma bubbles which were seen in 0100 station in Figure 7. In the lower latitude than the latitudes where plasma bubbles were observed, there was a region whose ion density was low, which has been called “equatorial hole” [e.g., Greenspan et al., 1991]. The density of the hole was less than 1% of those out of the hole. This hole would be caused by the uplift of the ionosphere by the strong eastward electric field [e.g., Greenspan et al., 1991].
 In the second pass from 1147 UT to 1213 UT, shown in Figure 10 (middle), fluctuations of MSTID were observed by the DMSP-F15 satellite in the shaded area. The density of depleted region was 1.0–3.0 × 105 [el/cm3]. The ratio of the depletion to the background was about 15–50%, which was smaller than that on the previous pass. In the southern hemisphere of the second pass, fluctuation of the super-MSTIDs and depletion of plasma bubbles were observed. The equatorial hole was not observed on the second pass. It indicates that eastward electric field was not so strong as on the previous pass. Fluctuations whose scale size and depletion ratio were almost the same as those of MSTIDs was observed around the geomagnetic equator. This fluctuation might be related to plasma bubbles. In the third pass from 1329 to 1355 UT, the fluctuations observed between 30°S and 40°S in the geomagnetic latitude could be MSTID. MSTIDs were not observed in the northern hemisphere while they were observed in the southern hemisphere. The difference between two hemispheres could arise from the longitudinal different. It would be the longitudinal effect, that is, the longitude of the pass in the southern hemisphere was east of that in the northern hemisphere.
Figure 11 shows the ion density measured by CHAMP between 1627 and 1649 UT. The trajectory of the satellite is represented by the long dashed arrow in Figure 8. The local time of the observation was 0205–0226. Data when the CHAMP satellite flew between 26°N and 42°N in the geographic coordinates, which are 18°N and 34°N in the geomagnetic coordinates, are shown by shaded area in Figure 11. It is seen that fluctuations of ion density were observed in the shaded region. The amplitude of the fluctuations was 2.0–4.0 × 104 [el/cm3], which corresponds to 10%–20% of the background density. According to the ground-based observation displayed in Figure 7, the super-MSTIDs were observed at this local time. Therefore, fluctuations observed over Japan in Figure 11 are expected to be plasma density fluctuations inside of the super-MSTIDs. Similar fluctuations were observed in the southern hemisphere. Equatorial hole observed by the DMSP-F15 was also observed by CHAMP observations. The latitude where the equatorial hole was seen was higher in the CHAMP observations than that from DMSP. The difference in latitude where equatorial hole was observed between DMSP data and CHAMP data would be caused by the difference of their observational altitudes.
 Strong uplift of the ionosphere was observed with ionosonde data in Japan. The first through fourth panels of Figure 12 show h'F and foF2 at Kokubuniji (35.7°N, 139.5°E in the geographic coordinate). The solid and dashed lines represent h'F and foF2, respectively. The data was hourly manual read data. Horizontal axis represents UT from 0500 to 2400 UT on 10 November 2004. The vertical solid lines around 0900 UT, and the dotted-dashed lines represent the sunset and the sunrise at 300 km altitude, respectively. Gray circles indicate time when spread F occurred. At Kokubunji, h'F increased from 180 to 380 km from 0800 UT (1700 LT) to 0900 UT (1800 LT). FoF2 increased after this uplift. Spread F occurred after 1000 UT (1900 LT). Upward vertical drift velocity is derived from the h'F data and plotted in the second panel of Figure 12. Vz was downward around 0730 UT. It increased and reached to 55 [m/s] at 0830 UT (1730 LT). Change of Vz can be caused by E × B drift, plasma drift due to neutral wind, and plasma diffusion.
 On the third and fourth panels of Figure 12, ionosonde data and Vz data derived from the ionosonde data at Yamagawa (31.2°N, 130.6°E in the geographic coordinate), are displayed respectively. The plotting format of the third and fourth panels is same as that of the first and second panels, respectively. At Yamagawa, the increase of h'F was from 220 to 380 km from 0800 UT (1700 LT) to 1100 UT (2000 LT). Increase of foF2 was observed after 0900 UT (1800 LT). Spread F occurred after 1200 UT (2100 LT). Vz was downward around 0730 UT (1630 LT). It increased and reached to 30 [m/s] at 0830 UT (1730 LT). The uplift of the ionosphere was also observed with Doppler shift data of HF radio wave measured at Oarai (36.2°N, 140.4°E in the geographic latitude). The experimental radio wave is transmitted from Chofu (CHF) (35.4°N, 139.3°E in the geographic coordinates). Doppler shift reflects the change of the altitude of the reflection point. When the reflection points move upward and downward, Doppler shift becomes negative and positive, respectively. The fifth panel of Figure 12 shows the upward drift velocity of the reflection point at Oarai. It was found that after 0730 UT (1630 LT), the velocity began to increase and reached 70 [m/s] at 0900 UT (1800 LT). After 1000 UT (1900 LT), data were scattered because the ionosphere was disturbed and many reflection points appeared.
Figure 13 shows emission of 135.6 nm airglow observed by the GUVI instrument on the TIMED satellite in the region of 70°E–200°E and 40°S–50°N in the geographic coordinate. The intensity of this 135.6 nm emission is proportional to the square of the electron density. The color bar shows the brightness with logarithmic scale in Rayleigh. Four full passes and parts of two passes of the observation in this area on 10 November 2004 are displayed. The boundaries of the passes are represented with thick solid lines in Figure 13. The TIMED satellite flew from south to north, whose UT and local time are shown below and above each pass. Two bands of enhanced plasma density of Equatorial Ionospheric Anomalies (EIAs) were observed at low latitudes in both hemispheres. The northern one was seen around 15°N–40°N. The southern one was seen around 20°S. The eastward electric field transports plasma from the equatorial region to higher altitudes and latitudes, which is called a fountain effect. In the pass from 1621 UT to 1646 UT, EIA was observed in the region of 30°N–40°N and 130°E–145°E in the geographic latitude. The area is almost consistent with the region where the super-MSTIDs were observed by ground-based GPS receivers and satellites. The magnetic latitude of the area was 22°N–32°N, which was higher than latitude where EIAs usually appear. The fact that the EIA appeared at higher latitude than usual indicates that strong eastward electric field existed in the equatorial area of this longitude before the GUVI observation. In the southern hemisphere in the pass from 1621 to 1646 UT, EIA was much weaker and appeared at lower latitude than that in the northern hemisphere. This asymmetry could be caused by meridional neutral wind on the dip equator. In the pass from 1758 to 1823 UT, EIA was weaker and appeared at low latitude than that in the pass from 1621 to 1646 UT. It indicates that eastward electric field at the equatorial region was smaller between 90°E and 120°E than between 120°E and 150°E.
 Distribution of mass density of neutral atmosphere was measured by the GRACE-A satellite. Figure 14 shows time series of the mass density when the satellite flew from 60°N to 60°S on the pass shown in Figure 8. The geomagnetic and geographic latitudes and the UT are shown on the bottom. The area of 26°N–42°N in the geographic coordinate, which is 18°N–34°N in the geomagnetic coordinates, is shaded. From 1359 to 1405 UT, variations of mass density were observed. The variations could be caused by the Traveling Atmospheric Disturbance (TAD). On the other hands, in the shaded region, where the super-MSTID were observed as shown in Figure 10, there was little fluctuations in mass density.
4.1. Super-MSTIDs Versus Plasma Bubbles
 TEC variations whose amplitude is around 20 TECU are often found associating with plasma bubbles [Nishioka et al., 2008]. An “equatorial hole” seen in Figure 10 suggests the presence of a strong eastward electric field in the equatorial region, that caused a large uplift the ionospheric plasma over the dip equator. Since the uplift could develop strong equatorial anomalies at much higher magnetic latitude than normal, plasma bubbles could appear at the midlatitude. It has been reported that plasma bubbles were observed at midlatitudes under the geomagnetic disturbed condition [e.g., Mendillo et al., 1997; Foster and Rich, 1998]. Therefore, the density disturbance observed with the satellite could be associated with plasma bubbles.
 However, the TEC disturbance observed by GEONET and the satellites had similar characteristics to those of MSTIDs. The direction of wave fronts, the direction of propagation, and it wavelength were consistent with those of nighttime MSTIDs, which have been investigated with GEONET data and 630 nm airglow [Satio et al., 2001; Sahai et al., 2001; Ogawa et al., 2002]. Location where the variations were observed was consistent with that of MSTIDs. Figures 7 and 8 indicate that Super-MSITD had large amplitude only at midlatitudes, which were between 18°N/S and 34°N/S in the geomagnetic coordinates. In Figure 8, the largest values of ROTI were observed at 0042 station and TOW2 station in the northern and southern hemispheres, which are at 27.5°N and 27.2°S in the geomagnetic coordinate, respectively. At geomagnetically low latitude, ROTI was smaller than those of midlatitude stations, 0042 and TOW2. It was reported that there is a possible low-latitude limit of MSTID [Shiokawa et al., 2002]. They found with their airglow measurement that the activity of MSTID was weakened below 18°N in the geomagnetic latitudes.
 Dense network data of GEONET revealed that several characteristics of the event were similar to those of MSTIDs. Characteristics that distinguish the event from plasma bubbles are spatial structure, direction of wave fronts, and local time of the occurrence.
4.1.1. Spatial Structure
 Wave fronts of MSTID appears successively [e.g., Saito et al., 2001]. On the other hand, depletions of plasma bubble appears with an interval of several hundred kilometers [e.g., Fukao et al., 2006]. Since wave fronts of the TEC disturbance on 10 November 2004 appeared successively as seen in Figure 6, the disturbance on 10 November 2004 was identified as MSTIDs.
 TEC variations along a line orthogonal to wave fronts of the TEC disturbance on 10 November 2004 had more than two positive and negative peaks in its structure, as seen in Figure 6. TEC variations along a line orthogonal to wave fronts of the TEC disturbance of MSTIDs, which were previously reported, had the similar characteristics to that of Figure 6. Figure 15 shows variations of TEC perturbation components along a line orthogonal to wave fronts of the MSTIDs, which appeared on 22 May 1998 [Saito et al., 2001]. Since the amplitude of the fluctuations was small compared to the background TEC, TEC variations of only perturbation components were plotted in Figure 15 in the same format as Figure 6. TEC perturbation components were derived by subtracting 60 minutes' averaged TEC from the background TEC. It was found that the variation along the line orthogonal to the wave fronts of MSTIDs had more than three positive and negative peaks, which was similar to that of the event.
 There have been several reports about zonal spacing where plasma bubble appeared [e.g., Tsunoda, 2005; Fukao et al., 2006]. Tsunoda  showed that plasma bubbles were generated with a zonal typical spacing of 400 km in the F region using data of ALTAIR incoherent scatter radar. It has been reported that multiple plasma bubbles were often generated at distances from each other of 370–1000 km [Fukao et al., 2006]. Figure 16 shows a two-dimensional TEC map obtained with GEONET data, at 1210 UT (2110 LT) on 7 April 2002. The plotting format is the same as that of Figure 5. TEC depletion regions were observed in the regions marked a and b, which were around (30°N, 130°E) and around (32°N,134°E), respectively, in the geographic coordinates. The TEC depletion regions were those of plasma bubbles. It was reported that plasma bubbles were observed with ground-based OI-630 nm all-sky imager at Shigaraki (34.8°N, 136.1°E) from 1140 UT (2040 LT) to 1240 UT (2140 LT) on this day [Ogawa et al., 2005]. According to their observation, a plasma bubble was observed around (32°N, 134°E) in the geographic coordinate, around 1200 UT (2100 LT), which was the simultaneous observation of plasma bubble b in Figure 16 (left). Zonal spatial structure of plasma bubble was plotted in the same format as that of Figure 6. Figure 16 (right) shows TEC variations, observed GEONET and PRN 17, along the line from (28.5°N, 126E°) to (34.5°N, 145.5°E) in the Figure 16 (left). Two TEC depletion regions are seen around 130°E and 134°E, which are shaded in Figure 16 (right). The depletion regions marked a and b are those seen in Figure 16 (left). The zonal distance between depletion a and b were about 400 km at the ionospheric altitude. The interval of plasma bubbles was consistent with the previous studies [e.g., Tsunoda, 2005; Fukao et al., 2006].
 In terms of spatial structure, the TEC disturbance appeared on 10 November 2004 was the similar to MSTIDs, not plasma bubbles.
4.1.2. Direction of Wave Fronts
 Wave fronts of MSTIDs stretch from northwest to southeast, and the direction is constant [e.g., Saito et al., 2002; Shiokawa et al., 2003]. On the other hand, plasma bubble alignments, which are caused by velocity shear of plasma drift on the dip equator, change according to local time because the shear varies with local time [Martinis et al., 2003]. Since the wave fronts of the TEC disturbance appeared on 10 November 2004, stretched in the northwest-southeast direction from 1130 UT (2030 LT) to 2230 UT (0730 LT), the event was identified as MSTID.
 The wave fronts of the TEC disturbance studied in this paper stretched from northwest to southeast at 1210 UT (2110 LT), 1230 UT (2130 LT), and 2230 UT (0730 LT) as seen in Figure 5a, 5b, and 5c, respectively. It was found that the wave fronts of the TEC disturbance stretched from northwest to southeast and it was almost constant from 1130 UT (2030 LT) to 2230 UT (0730 LT), when the disturbance observed. The direction did not change from the time when the wave fronts were clearly observed from 1130 UT (2030 LT).
 Keograms obtained with GEONET-TEC data and the ground-based OI-630 nm all-sky imager at Shigaraki indicated that the stretch direction of MSTIDs' wave fronts was from northwest to southeast and it was almost constant during the presence of the MSTIDs [Saito et al., 2001, 2002; Shiokawa et al., 2003]. Constant northwest-southeast of stretched direction of wave fronts is characteristics of MSTIDs observed at nighttime at midlatitudes.
 On the other hand, direction along which plasma bubbles stretch changes according to local time. The directions along where plasma bubbles stretch reflects velocity shear of plasma drift. The change of the alignment direction is caused by latitudinal difference of zonal drift velocities [Martinis et al., 2003]. It was observed that plasma bubbles slanted westward in the northern hemisphere because of the latitudinal difference of zonal drift velocities [e.g., Zalesak et al., 1982; Kelley et al., 2003]. Zonal drift velocity during geomagnetically disturbed day also has latitudinal difference [Taylor et al., 1997; Fejer and Emmert, 2003]. The plasma bubble alignments could be modulated by the latitudinal difference of zonal drift velocities during geomagnetically disturbed days.
 The stretched direction of wave fronts of the TEC disturbance in this paper was northwest-southeast and almost constant during the night. Constant direction of the wave fronts is one of the characteristics of MSTIDs. Therefore, we believe that the disturbance on 10 November 2004 must be MSTIDs.
4.1.3. Local Time of the Disturbance
 Plasma bubbles hardly occur after 0730 LT [e.g., Kakad et al., 2007]. On the other hand, the TEC disturbances appeared after 0730 LT.
 The TEC disturbances were observed continuously from 1000 UT (1900 LT) to 2230 UT (0730 LT) at 0042 station (36.5°N, 140.6°E) as seen in Figure 7. Wave fronts can be seen from Figure 5c, which was 2230 UT (0730 LT). The local time, when the disturbance observed after 0730 LT at 0042 station was generated, must be later than 0730 LT, because the disturbance was propagated from the east of the station, where the local time was later than 0730 LT.
 During geomagnetically active periods, plasma bubbles can be generated after midnight which are probably affected by electric perturbations associated with the magnetic activity [Bhattacharyya et al., 2002; Kakad et al., 2007; Tulasi Ram et al., 2008]. Kakad et al.  studied local time when plasma bubbles were generated with ground-based scintillation measurement. According to their study, plasma bubbles are mostly generated around midnight hours during disturbed days. They used a proxy of freshness of plasma bubbles from their scintillation data. The probability that the index was larger than the threshold they set in their paper was less than 5% at 0630 LT while it was more than 20% at midnight hours. The probability after 0730 LT was zero.
 The growth rate of plasma bubble is proportional to ΣPF/(ΣPF + ΣPE) where ΣPF and ΣPE are field-aligned integrated conductivities in the F and E regions, respectively [Sultan, 1996]. Plasma bubbles appear during the nighttime when ΣPE becomes quite small compared to ΣPF. After sunrise, ΣPE becomes larger than ΣPF. Electric field generated by the F region dynamo is shorted out by large ΣPE, and then the growth rate becomes small. Therefore, plasma bubbles are not generated after sunrise.
 The TEC disturbance on 10 November 2004 could not be plasma bubbles because of the late local time of the disturbance.
4.2. Altitude of the Presence of the Disturbance
 Observational data of the super-MSTIDs by satellites can help to study the vertical structure of the super-MSTIDs. At 1030, 1340, and 1630 UT, when the DMSP satellite, the GRACE-A satellite, and the CHAMP satellite flew over the Japanese sector, the peak-to-peak amplitude of the super-MSTIDs were about 12–28, 8–12, and 3–6 TECU, respectively. The TEC amplitude of the super-MSTIDs observed with the ground-based receivers, ΔTECGPS, and the corresponding background TEC, TEC0 are listed on Table 3. Amplitudes of TEC disturbances observed by the GRACE satellite, which is referred as ΔTECGRACE, are listed on the fourth column of Table 3. By comparing ΔTECGRACE and ΔTECGPS, 42–100% of ΔTECGRACE/ΔTECGPS indicates that more than half of the TEC variations observed by the ground-based GPS receivers was contributed by TEC variations observed above 460 km altitude. The altitude of the super-MSTIDs was higher than ionospheric F region peak which is usually observed.
ΔTEC, TEC0, and R are the TEC amplitude of super-MSTID, background TEC, and the ratio of amplitude of super-MSTID to the background, respectively. The subscripts “GPS,” “DMSP,” “GRACE,” and “CHAMP” indicate measurement with ground-based GPS receivers and the DMSP, GRACE, and CHAMP satellites, respectively.
 The ratios of ΔTECGPS to the background TEC, which is referred as RGPS, are also listed in the fifth column of Table 3. Ratios to amplitude of the ion density disturbance to its background observed by the DMSP and CHAMP satellites are summarized as RDMSP and RCHAMP, in the sixth and seventh column of Table 3, respectively. At 850 km altitude, ion density was measured by the DMSP satellite. There were depletion regions of ion density in the latitude range of ±15°–±35°, as shown in Figure 10 (top), where the super-MSTIDs were observed with the ground-based observation. Track-along length of region a, which was seen in Figure 10, was 600 km. It corresponds to 200 km orthogonal to the wavelength of the super-MSTIDs observed with GEONET data as seen Figure 5. The location of the ion density fluctuations was close to the location where the super-MSTIDs were observed with GEONET observation as seen in Figure 8. Therefore, this fluctuation could be associated to the super-MSTIDs. Ratio of the fluctuation to the ambient ion density, which was described as RDMSP was from 25% to 75%. On the other hand, RGPS at that time was 40–92%. The depletion rate observed with the ground-based GPS receiver was comparable that of the DMSP satellite. On the pass from 1147 to 1213 UT shown in Figure 10 (middle), the amplitude of the fluctuations was 1.0–3.0 × 104 [el/cm3] with the background of about 7.0 × 104 [el/cm3], that is, the ratio was 15–50%. The ratio decreased with time and westward movement of the satellite, which is consistent with the ground-based GPS observation (Figure 8). RCMAMP, which represents the ratio of the density disturbance amplitude at 360 km altitude was 10–20% while R GPS was 40–60%. RGPS > RCHAMP and RGPS ∼ RDMSP indicate that the super-MSTIDs mainly existed above 360 km altitude and extended to 850 km.
 The altitude where TEC variations of the super-MSTIDs appeared was different from that where TEC depletions of plasma bubbles appeared. Plasma bubbles were observed with DMSP ion density observation on 10 November 2004. In Figure 10 (top), deep depletion of ion density was seen at lower latitude than 15°N/S in the geomagnetic latitude. This depletion was caused by plasma bubble according to the ground-based observation and its latitude. The ion density of depletion region was about 1.0 × 103 [el/cm3], while the background ion density was about 1.0 × 105 [el/cm3]. The ratio of depletion to the background ion density observed by the DMSP satellite was about 99%. The depletion ratio measured at the 0100 ground-based GPS receiver was 25–75%. The ratio of depletion of plasma bubble measured with the DMSP-F15 was larger than that with the ground-based GPS receivers. In the case of the super-MSTIDs, the ratio with the DMSP measurement and ground-based GPS measurements were comparable. It means that TEC depletions of plasma bubble were contributed by TEC depletions around 850 km altitude while TEC variations of the super-MSTIDs were not mainly contributed by TEC variations around 850 km.
4.3. Generation Mechanism of the Super-MSTID
 Perkins instability is the most reasonable mechanism to generate MSTID. The linear growth rate of the Perkins instability is expressed as
where E0, B, D, H are the background electric field, the geomagnetic field, the dip angle of the geomagnetic field, and the scale height of the neutral atmosphere. α and θ are the angles of the direction of the wave vector, and the electric field from the east, respectively. The growth rate γ is maximum when α = θ/2 and
The growth rate γ is large when the direction of E0 is east and north.
 The eastward electric field E0 was given by the following form:
〈≠in〉 is the height-integrated ion-neutral collision frequency, that is,
when equation (3) was used. Equation (5) indicates that the growth rate is proportional to sin2D. The dip angles for the central Japan (the 0042 station) and southern part of Japan (the 0100 station) were 50.0° and 37.0°, respectively. Values of sin2D for the 0042 and 0100 stations were 0.59 and 0.36, respectively. Higher growth rate at the 0042 station than at 0100 station was consistent with the observational result, that is, the activity of super-MSTID was higher in the central part of Japan than in the southern part of Japan, as shown in Figures 7 and 8.
 Strong eastward electric field could contributed to γ. Figure 10 shows that the ionosphere was uplifted in both hemispheres. Uplift of ionosphere in the wide range in the latitude indicates the uplift by the eastward electric field. Strong uplift of the ionosphere was also observed with ionosonde data in Japan as seen in Figure 12. The uplift can be caused by E × B drift, plasma drift due to neutral wind, and plasma diffusion. Eastward electric field is favorable for Perkins instability. When we assume that only E × B drift contributed to the uplift of the ionosphere, eastward electric fields for Vz = 55 [m/s] at Kokubunji and Vz = 70[m/s] at Oarai are estimated to be 2.4 [mV/m] and 2.8 [mV/m], respectively, using the relation of vz = (E × B)/B2. The altitude of the ionosphere is assumed 400 km. The growth rate estimated with eastward electric fields 2.8 [mV/m], which was estimated from the HFD data at Oarai, was 1.6 × 10−4 [s−1]. The growth rates are very small although it was larger than those reported in geomagnetically quiet periods [e.g., Shiokawa et al., 2003] and comparable to those reported in geomagnetically disturbed periods [Miller et al., 1997; Garcia et al., 2000a]. The small growth rate of the Perkins instability has been a problem as a mechanism of MSTIDs [e.g., Garcia et al., 2000a, 2000b; Shiokawa et al., 2003].
 Both of direct penetration electric field and disturbance dynamo could contributed to the eastward electric field [e.g., Fejer and Scherliess, 1997; Maruyama et al., 2005]. Eastward electric fields associated to the geomagnetic storm could contribute to the growth rate of the Perkins instability. AE index varied between 1000 and 2000 nT when the MSTIDs were observed. It has been reported that AE index can be a proxy of penetration electric fields at low latitudes [e.g., Fejer and Scherliess, 1997]. Eastward penetration electric field associated to increase of AE index could contributed to the generation of the MSTIDs. Disturbance dynamo electric field could contribute to an eastward field at night at midlatitude [Blanc and Richmond, 1980]. Eastward electric field could also be caused by the equatorward neutral wind. In Figure 13, enhanced EIA was seen only in the northern hemisphere. There would be equatorward neutral wind in the Japanese sector. The southward neutral wind could be a cause of eastward electric field.
 In a previous study, well-defined bands where the F region was alternatively high and low were observed with incoherent scatter radar at Arecibo, Puerto Rico [Behnke, 1979]. The transition between adjoining regions was very sharp in their observation. In another study, similar events were reported with airglow observation at Arecibo during high geomagnetic activity [Garcia et al., 2000a; Kelley et al., 2000]. A structure, measured as emissions of 630.0 nm, extended from northwest to southeast, with the wavelength 280 km, from 2100 to 2300 LT to the midnight. The intensity of the emission, which comes from about 250 km altitude, varied from 60 Rayleigh to 120 Rayleigh. The structure traveled toward the southwest at the speed of 63 [m/s]. Characteristics of the event, that is, the direction along the wave front stretched and the direction of propagation, and the geomagnetic condition when the structure appeared, were similar to those of the super-MSTIDs.
 There are several characteristics of MSTIDs that cannot be explained by Perkins instability. The speed and direction of its propagation is one of them. According to a statistic study with airglow measurements at midlatitude by Garcia et al. [2000b], the velocity of MSTIDs varied from 50 [m/s] to 170 [m/s]. The speed 180 [m/s] is close to those velocities but larger than those velocities. Several process that can explain the propagation have been suggested like coupling between E and F regions [e.g., Otsuka et al., 2008; Cosgrove, 2007]. Further study is necessary to understand the propagation process.
 The wavelength of 100–500 km of the super-MSTIDs cannot be explained by the Perkins instability. The growth rate of the instability has no preference for scale size [Perkins, 1973]. According to a statistic study with airglow measurements by Garcia et al. [2000b], the wavelength varied from 50 to 500 km. The wavelength of MSTIDs may be determined by the gravity waves, which can be a seed of the instability [e.g., Kelley and Fukao, 1991; Shiokawa et al., 2003]. The wavelength may be related to coupling process between the E and F regions. Length of Es layer was similar to the wavelength of MSTIDs [Cosgrove, 2007]. Further studies are necessary for the wavelength.
 There have been several reports that TIDs could be caused by atmospheric gravity waves. The gravity waves could be induced by the energy input from the magnetosphere into the auroral zone [Williams et al., 1988; Bristow and Greenwald, 1994]. It have been reported that large-scale traveling ionospheric disturbances (LSTIDs) were observed during geomagnetically disturbed periods [e.g., Shiokawa et al., 2002; Tsugawa et al., 2003]. The characteristics of LSTIDs, which were studied with GEONET [Tsugawa et al., 2004], were different from those of the event on 10 November 2004. The mean wavelength of LSTIDs was about 2,000 km while that of the event was 100–500 km. Wave fronts of LSTIDs stretch in the west-south direction while that of the event in this paper stretched in the northwest-southeast direction. Because of the difference of the characteristics, the event cannot be LSTIDs.
 There were no obvious correlation between geomagnetic activity and occurrence of MSTIDs [Saito et al., 2001; Tsugawa et al., 2007]. Figure 14 also shows that TAD did not appeared at the region where the super-MSTIDs were observed around 1340 UT. Although the role of gravity wave might be important for the MSTIDs' generation [e.g., Hines, 1960; Hooke, 1968], the correlation between generation of the nighttime MSTIDs and activity of gravity waves is not found. Shiokawa et al.  performed model calculation of MSTIDs generated by gravity waves and by an oscillating electric field and found that gravity waves could not directly cause the MSTIDs while polarization electric field could produce MSTIDs. It is also reported that the polarization electric field was observed in the nighttime MSTIDs structures [Shiokawa et al., 2003], which suggests that polarization electric field plays an important role in the generation of MSTIDs [e.g., Saito et al., 1995]. Since the super-MSTIDs appeared during the geomagnetically disturbed period, there is a possibility that gravity waves might contribute to the appearance of the super-MSTIDs. However, the relationship between the activity of gravity waves and the source of the super-MSTIDs is not clear.
 Gravity waves can propagate in a waveguide mode around 110 km altitude, as suggested by Richmond . If gravity waves caused MSTIDs, MSTID would appear in the bottom side of the F region. It is also reported that the variation of the electron density of MSTIDs took place mainly in the bottom side of the ionospheric F region [Saito et al., 2001]. On 10 November 2004, F region was higher altitude than usual. Figure 12 indicates that h'F at Kokubunji was higher than 350 km until 1200 UT. Since plasma cannot be produced after sunset, electron density in the bottom side could be low and the altitude of the bottom of F region could be higher than usual (e.g., >250 km). Then, there could be small perturbation of the super-MSTIDs at lower altitude (e.g., <250 km). Further study is necessary to clarify the relationship between gravity wave and MSTIDs.
 Intensified MSTIDs whose peak-to-peak amplitude was larger than 20TECU were observed at midlatitude during the geomagnetic storm on 10 November 2004. Characteristics of the MSTIDs were clarified was with high-resolution data of GEONET. The disturbance started around 1000 UT, followed by a TEC enhancement after the sunset in the central part of Japan. Peak-to-peak amplitude of the disturbance was larger than 20TECU around 1200 UT. TEC map indicates that the disturbance had several wave fronts which extended from northwest to southeast, and propagated from northeast to southwest. The velocity of the propagation was about 180 m/s. Wave length of the structure was from 100 to 500 km. The TEC disturbances were observed around the geomagnetic conjugate point of Japan. The activity of disturbances was high at midlatitude than at low latitude in the both hemispheres. These characteristics were similar to those of MSTID except for the large amplitude, which was 10 times larger than usual MSTIDs. The disturbance was named as “super-MSTIDs” in this paper because of the large amplitude of the MSTID-like disturbance. TEC variations of the super-MSTIDs were also observed at 460 km altitude by the GRACE satellite. The amplitude of the TEC variations that were TEC between 460 km altitude and the GPS orbit had comparable amplitude with that observed by the ground-based GPS receivers. The ion density fluctuations of the super-MSTIDs were observed in situ by the CHAMP and DMSP-F15 satellites, which flew at 360 km and 850 km altitude, respectively. It is found that the plasma density variations of the super-MSTIDs occurred mainly above 360 km altitude.
 The characteristics that distinguish the event from plasma bubbles are its successive wave fronts, constant northwest-southeast direction along which the wave fronts stretched, and late local time of the occurrence. Large amplitude of the super-MSTIDs could be caused by uplift of the ionosphere. Uplift of the ionosphere can be suggested by the Equatorial holes of ion density profile measured by the DMSP satellite and high brightness of 135.6 nm emission measured by the TIMED satellite. Strong eastward electric field would cause the uplift of the ionosphere.
 Ground-based GPS data GEONET was provided by Geographical Survey Institute, Japan. Other ground-based GPS network data were provided by International GNSS Service, the networks of Scripps Orbit Permanent Array Center, Japan Agency for Marine-Earth Science and Technology, and Solar-Terrestrial Environment Laboratory, Nagoya University. Data of the GRACE and CHAMP satellites were provided by the GeoForschungsZentrum, Potsdam, Germany. Data of the DMSP satellite were provided by C. Huan and F. Rich at Air Force Research Laboratory. The ionosonde data of Kokubunji and Yamakawa were provided by the National Institute of Information and Communications Technology, Japan. The HFD data were provided by Sugadaira Space Radio Observatory, the University of Electro-Communications. TIMED/GUVI data, which is supported from the NASA MO and DA program, were provided by Applied Physics Laboratory, Johns Hopkins University.
 Zuyin Pu thanks Alan Rodger, Shunrong Zhang, and another reviewer for their assistance in evaluating this paper.