We present midlatitude medium-scale traveling ionospheric disturbances (MSTIDs) observed with a Super Dual Auroral Radar Network (SuperDARN) HF radar at around 10 MHz in Hokkaido, Japan, in combination with a 630-nm all-sky imager and a GPS network (GEONET) that provides total electron content (TEC) data. MSTIDs propagating southward from high latitudes are detected at first with the HF radar and then with the imager and GEONET. We analyze two MSTID events, one in winter (event 1) and the other in summer (event 2), to find that MSTIDs appear simultaneously, at least, at 55°–25°N. It is shown that nighttime MSTIDs propagate toward the southwest over a horizontal distance of about 4000 km, and daytime MSTIDs do so toward the southeast. Daytime radar echoes are due to ground/sea surface (GS) scatter, while nighttime echoes in event 1 return from 15-m-scale F region field-aligned irregularities (FAIs) and those in event 2 are due to GS scatter. Doppler velocities of the nighttime F region FAI echoes in event 1 are negative (motion away from the radar) within strong echo regions and are positive (motion toward the radar) within weak echo regions. This fact suggests that the strong (weak) echoes return from suppressed (enhanced) airglow/TEC areas, in line with previous observations over central Japan. The nighttime MSTIDs in events 1 and 2 are often accompanied by concurrent coherent echoes from FAIs in sporadic E (Es) layers. The Es echo areas in event 2 rather coincide with suppressed airglow/TEC areas in the F region that are connected with the echo areas along the geomagnetic field, indicating the existence of E and F region coupling at night.
 Medium-scale traveling ionospheric disturbances (MSTIDs) in the F region ionosphere have been believed to be ionospheric manifestations of atmospheric gravity waves (AGWs) caused by geomagnetic activity at high latitudes or launched from the lower atmosphere, since Hines  suggested an important role of AGWs in inducing large-scale (≥100 km) wavy structures in the ionospheric plasma. MSTIDs have horizontal wavelengths of several hundred kilometers, horizontal phase velocities of 100–250 m s−1, and periods of 15–60 min, and have been well studied for long years (see reviews by Hunsucker  and Hocke and Schlegel ). Previous observations at midlatitudes and high latitudes indicate that propagation direction of MSTIDs is controlled by thermospheric neutral wind, geomagnetic activity, etc., and change with local time [e.g., Crowley et al., 1987; Hocke and Schlegel, 1996, and references therein; Shiokawa et al., 2003a; He et al., 2004; Kotake et al., 2007; Tsugawa et al., 2007; Shiokawa et al., 2008].
 With the advent of observational techniques, capable of imaging two-dimensional ionospheric plasma distribution, such as multiple GPS satellites, all-sky airglow imagers, etc., the characteristics of nighttime MSTIDs have become clearer. For example, Mendillo et al.  and Garcia et al.  detected southwestward propagating MSTIDs with phase fronts aligned along NW–SE with a 630-nm imager at Arecibo, Puerto Rico (18.3°N, 66.75°W), and discussed their relation to F region dynamics. Similar MSTIDs having a large spatial extent (≥2500 km) along the Japanese islands were observed simultaneously with five imagers and a GPS network (GEONET: GPS Earth Observation Network) in Japan by Kubota et al.  and Saito et al. . Nighttime MSTID activities over Japan revealed from airglow imager observations had a major peak in summer (May–July) and a minor peak in winter (November–February), and were inactive in equinoxes [Shiokawa et al., 2003a]. Such a seasonal dependence has been confirmed from a statistical analysis of GPS total electron content (TEC) data over Japan by Kotake et al. , who also found daytime MSTIDs to be most active in winter.
 MSTIDs at northern high latitudes have been well studied with the Super Dual Auroral Radar Network (SuperDARN) HF radars [e.g., Bristow and Greenwald, 1996, and references therein]. Bristow et al.  found the probability of medium-scale AGW observations to be highest in winter months and lowest in summer months, which is consistent with MSTID observations using beacon waves from NNSS satellites at southern high latitudes [Ogawa et al., 1987]. In the SuperDARN observations at high latitudes, equatorward propagating MSTIDs are detected as a quasiperiodic fluctuation of the echo power backscattered from the ground or sea surface. The fluctuations are caused by focusing and defocusing of HF ray paths due to electron density perturbations associated with MSTIDs [Samson et al., 1990]. MSTIDs generated at high latitudes may propagate toward midlatitude [e.g., Ishida et al., 2008], and be observed over Japan [e.g., Kubota et al., 2000; Saito et al., 2001]. However, whether MSTIDs generated at high latitudes can propagate to Japan (and farther southward) over a long distance or not is still unclear from the viewpoint of observations.
 To investigate MSTID characteristics at midlatitudes and possible long distance propagation of MSTIDs, we use a SuperDARN radar in Hokkaido, Japan, capable of viewing a wide area to the northeast of Hokkaido. We also use a 630-nm all-sky airglow imager in Hokkaido and GEONET to monitor southward propagation of MSTIDs that have crossed a field of view of the radar. We analyze two MSTID events, one in winter (hereinafter called “event 1”) and the other in summer (“event 2”), in detail. The radar echoes are due to either ground/sea surface (GS) scatter, as found at high latitudes, or coherent scatter from 15-m-scale field-aligned irregularities (FAIs) in the ionosphere. Importantly, nighttime MSTIDs are often accompanied by concurrent coherent echoes from sporadic E (Es) layers, indicating the existence of E and F region coupling at night in both the events.
 In this paper we use three observation instruments, that is, an HF radar located at Rikubetsu in Hokkaido (43.53°N, 143.61°E; 36.46°N magnetic; L value 1.55), a 630-nm all-sky airglow imager at Rikubetsu, and an extremely dense GEONET in Japan. The HF radar is one of the SuperDARN HF radars [Greenwald et al., 1995]. The all-sky imager is capable of imaging nighttime F region plasma distribution mainly at altitudes of 250–300 km below the F layer peak [e.g., Mendillo et al., 1997; Garcia et al., 2000; Ogawa et al., 2002]. GEONET of about 1200 receivers with an average distance between two receivers of about 25 km provides data of total electron content (GPS-TEC) between the GPS altitude (20,200 km) and the ground every 30 s [e.g., Saito et al., 2001].
Figure 1 shows a field of view (FOV) of the SuperDARN Hokkaido radar in geographic coordinates. The FOV is covered with 16 narrow beams (beam numbers 0, 1, 2, …, 15; each two-way beam width between 2.5° at 20 MHz and 6° at 8 MHz) over an azimuth sector of 52°. The beams have maximum sensitivity at elevation angles of 15°–35° (depending on radar frequency) suitable for the detection of E and F region FAIs. Expected E region coherent echo area at 90–120 km altitudes is marked in Figure 1 as “E region echoes” (see Figure 10 in detail). F region FAI and GS scatter echoes return from farther ranges beyond the E region echo area. The radar beam was sequentially scanned from beam 0 to beam 15 with a step of 3.24° in azimuth, a scan repeat time of ∼120 s (or 60 s), and a range resolution of 45 km. The first range gate was set to 180 km. The Rikubetsu all-sky imager, equipped with a thinned and back-illuminated cooled-CCD camera with 512 × 512 pixels, took airglow images every 5.5 min with an exposure time of 165 s [e.g., Shiokawa et al., 2009]. The imager FOV with a diameter of about 1100 km at 250 km altitude for a zenith angle of 65° is also shown in Figure 1.
3.1. Observations on 22–23 January 2007 (Event 1)
 Since the start of operation of the Hokkaido radar in December 2006, the most prominent phenomenon observed with the radar has been MSTIDs. As an example of MSTIDs in winter (event 1), Figure 2 displays range-time-intensity (RTI) plot of the radar echo power observed at around 11 MHz on beam 5 (see Figure 1) on 22–23 January 2007. Note Japan Standard Time (JST) = UT + 9 h. In both the daytime (0700–1600 JST) and nighttime (2300–0300 JST) the echo regions appearing repeatedly with a period of about 60 min at farther ranges move toward the radar site with time, and disappear at ranges of about 900 and 600 km, respectively. Such echo striations are believed to be caused by the MSTID propagation in the F region [e.g., Samson et al., 1990]. The echo disappearance at the indicated ranges does not always mean that MSTIDs decay there, but are due to the oblique radar beam covering the limited elevation angles (see above). The Kp indices during 2100–0900 UT were 0, 1+, 2, 0+ and those during 0900–2100 UT were 1, 0+, 0+, 0. Coherent echoes from Es layers also appear intermittently at around 300 km range in the daytime and nighttime.
 To see the relations among power, Doppler velocity, and spectral width of the echoes, we chose somewhat arbitrarily the radar beams, range gates, and time intervals for which the echo powers were higher than 10 dB, beyond which both the velocity and width were obtained with higher reliability. Figure 3 shows scatterplots of the Doppler velocity and spectral width versus radar echo power on some beams at some range gates for the F region echoes. The velocity sign is positive (negative) for motion toward (away from) the radar. In the daytime (0415–0515 UT; Figure 3a) the echoes at range gates 23–25 (1215–1350 km) on beams 4–13 have Doppler velocities between −22 and +30 m s−1 (mostly between −15 and +10 m s−1) and spectral widths less than 35 m s−1 (mostly less than 20 m s−1). On the other hand, in the nighttime (1532–1612 UT; Figure 3b) the echoes at range gates 16–18 (900–1035 km) on beams 0–11 have Doppler velocities between −127 and +65 m s−1 (mostly between −120 and +60 m s−1) and spectral widths less than 99 m s−1, whose values are far larger than those in Figure 3a. We think that the daytime and nighttime echoes are due to GS scatter and 15-m-scale F region FAIs, respectively. Similar GS scatter echoes detected in both the daytime and nighttime with the TIGER SuperDARN radar in Tasmania, Australia (43.4°S, 147.2°E) were statistically analyzed by He et al. , who identified GS scatter echoes as echoes with low Doppler velocities (typically between −50 and +50 m s−1) and narrow spectral widths (typically less than 20 m s−1).
 The above mentioned high Doppler velocities and wide spectral widths of the nighttime F region FAI echoes are consistent with the values (between −120 and +200 m s−1 and less than 100 m s−1, respectively) from 46.5-MHz middle and upper atmosphere (MU) radar observations of summer nighttime 3.2-m-scale F region FAIs to the north of Shigaraki (34.9°N, 136.1°E; 25.0°N geomagnetic; marked in Figure 10) in central Japan [e.g., Fukao et al., 1991].
Figure 4 shows two-dimensional maps of the radar echo power and GPS-TEC perturbation amplitude (in TECU; one TECU = 1016 electrons m−2) over Japan at 0030 UT (0930 JST; Figure 4a) and 1604 UT (0104 JST; Figure 4b) in event 1. Six and eight GPS satellites were available to construct the TEC map in Figures 4a and 4b, respectively. Note that relative change of TEC is obtained theoretically with an accuracy of 0.01–0.02 TECU [e.g., Tsugawa et al., 2007]. The TEC perturbation amplitude at each time is defined as deviation from 1-h running average of TEC. Observed, slant TEC values were converted in vertical ones by assuming the ionospheric plasma to exist between 250 and 450 km altitudes. The vertical values thus determined are mapped on geographical coordinates by assuming the ionosphere to be a thin layer located at 300 km altitude [e.g., Saito et al., 2001]. GPS-TEC data with satellite elevation angles larger than 30° were used for the analysis.
 In Figure 4 the areas with strong F region echo power (higher than about 12 dB) and positive TEC amplitude are marked by the rectangles. As mentioned above, the daytime radar echoes from farther ranges, shown in Figure 4a, are due to GS scatter. To know approximate geographical location of ionospheric reflection point of the GS scatter echo, we assume an reflection altitude of the radar wave to be 250 km and a slant range from the radar to the reflection point to be half of the observed echo range: these assumptions may not always be valid [Samson et al., 1990; Bristow and Greenwald, 1995], but seem enough to calculate the approximate reflection point [He et al., 2004]. The reflection points thus determined are plotted in Figure 4a. In Figure 4a, though not so clear, the TEC map exhibits MSTID structures (i.e., quasiperiodic fluctuations of TEC) with phase fronts aligned roughly along ENE–WSW and a wavelength of about 600 km, and the radar echo map does similar structures due to the MSTIDs but with shorter wavelength (it is difficult to estimate exact wavelength from the radar echo map because the ionospheric reflection points plotted are calculated under some assumptions). By considering the 600 km wavelength in the TEC map and the 60 min period of the MSTIDs (Figure 1), the phase velocity toward SSE is estimated to be about 170 m s−1. As mentioned above, the nighttime radar echoes in Figure 4b are due to F region FAIs. Contrary to the daytime case, MSTIDs in the echo map exhibit phase fronts aligned roughly along NNW–SSE with a wavelength of 300 km, indicating the MSTID propagation toward WSW at about 80 m s−1. Such MSTID structures are also discernible in the TEC map.
 To know detailed spatial relationship between the echo power and Doppler velocity at night, Figures 5a and 5b show RTI and range-time-velocity plots, respectively, on beam 5. The strong echo regions in Figure 5a are traced by the oblique grey lines that are also overlaid in Figure 5b. Again, the velocity sign is positive (negative) for motion toward (away from) the radar. It is observed that after 1430 UT the strong (weak) echo regions coincide well with the negative (positive) velocity ones. To demonstrate such a relationship more clearly, Figures 5c and 5d display a two-dimensional map of the echo power and Doppler velocity, respectively, at 1612 UT. The strong echo areas in Figure 5c are marked by the rectangles that are also overlaid in Figure 5d. Again, the strong (weak) echo areas coincide with the negative (positive) velocity areas.
Figure 4b is enlarged in Figure 6 to highlight the E region echoes. Figures 6a and 6b display a map of the echo power and Doppler velocity, respectively. The data with echo power higher than 9 dB are plotted. The echoes returned mainly from altitudes of 90–120 km (see Figure 10). The demarcations between plus and minus velocities are indicated by the thick solid lines in Figure 6b. Clearly, the velocities are negative (positive) in the northeastern (southwestern) part of the strong echo area. In Figure 6c the 630-nm airglow intensity deviation map at the assumed altitude of 250 km at 1604 UT is mapped down to 100 km altitude along the geomagnetic field, and is overlaid on the Doppler velocity and GPS-TEC maps. The airglow intensity deviation is defined as (I − IB)/IB (in %) where I is the measured intensity and IB is the 1-h running average of I. It can be observed that the E region velocities have northward (southward) component on the northeastern (southwestern) part of the enhanced airglow region, and that the demarcation lines shown in Figure 6b align along the enhanced airglow region. Such coincidences are also discernible even if the assumed emission altitude is changed from 250 km to 300 km, though the airglow map at 100 km altitude in Figure 6c shifts northward by a small amount, i.e., about 29 km (0.26°).
3.2. Observations on 11 June 2007 (Event 2)
 As an example of MSTIDs in summer (event 2), Figure 7 displays RTI plot of the radar echo power observed at 9.1 MHz on beam 5 on 11 June 2007. The Kp indices on this day were 1, 0, 0+, 0+, 0+, 1, 1−, 1, 5−. The echo patterns are largely different from those on 22–23 January in Figure 2. At night (0900–1700 UT; 1800–0200 JST) the ranges of some echoes exceed the maximum observation range of 3510 km. As will be described later, the nighttime echoes beyond about 600 km range and the daytime echoes beyond about 400 km are not due to F region FAIs but GS scatter. Though not so clear as the case in Figure 2, the southward propagating MSTID signatures are more or less discernible all the day. The nighttime Es echoes at ranges of 200–500 km during 1100–1600 UT (2000–0100 UT) are strong, and are related to the MSTID propagation, as will be discussed later. Such nighttime Es echoes in summer have been also observed over central Japan with an HF radar [Tanaka and Venkateswaran, 1982a, 1982b] and the MU radar [e.g., Yamamoto et al., 1991].
Figure 8 presents two-dimensional maps of the radar echo power and GPS-TEC perturbation amplitude at 1201 UT (2101 JST). Six GPS satellites were available to construct the TEC map. Since the F region echoes have low Doppler velocities and narrow spectral widths (see Figure 9b), the echoes are regarded as GS scatter echoes. Therefore, like Figure 4a, the ionospheric reflection points of the GS scatter echoes are plotted in Figure 8. It can be observed that the MSTIDs propagate southwestward, in line with the propagation in Figure 4b.
Figure 9 shows scatterplots of the Doppler velocity and spectral width versus radar echo power of the nighttime E and F region echoes. Like Figure 3, we chose somewhat arbitrarily the radar beams, range gates, and time intervals for which the echo powers were higher than 10 dB. The Es echoes (Figure 9a) have Doppler velocities between −53 and +57 m s−1 and spectral widths less than 103 m s−1 (mostly less than 90 m s−1) at range gates 2–3 (270–360 km) on beams 2–13. Such high Doppler velocities and wide spectral widths suggest that the echoes are due to coherent scatter from Es FAIs but not reflection from Es clouds. On the other hand, the F region echoes (Figure 9b) have Doppler velocities between −44 and +63 m s−1 (mostly between −30 and +25 m s−1) and spectral widths less than 60 m s−1 (mostly less than 50 m s−1) at range gates 19–21 (1035–1125 km) on beams 4–13, in line with those in Figure 3a, indicating that the echoes are due to GS scatter. The above mentioned Doppler velocities and spectral widths of the nighttime E region FAI echoes are consistent with past MU radar results (between −100 and +100 m s−1 and less than 150 m s−1, respectively) [e.g., Ogawa et al., 1995].
Figure 8 is enlarged in Figure 10 where contrary to Figure 8, the radar echo ranges are observed ones but not the ionospheric reflection points. Overlaid in Figure 10 are contour lines of perpendicularity between the radar wave vector and the geomagnetic field vector (IGRF 2005; epoch = 0000 UT on 1 January 2007) at altitudes (h shown in the inset) of 90, 100, 110, 120, and 150 km. No wave refraction during propagation is assumed to obtain the lines. The radar echoes at 270–500 km ranges return mainly from the E region at altitudes of 90–120 km. However, the echo areas on beams 12–15 are deviated from the contour lines and are located at ranges shorter than expected, maybe because the radar waves are refracted downward during propagation due to enhanced electron density in Es (note that the observed echo ranges in Figure 10 are calculated by assuming no wave refraction). An oblique straight line connects point 1 (38°N, 136°E; distance = 0 km) and point 2 (46°N, 150°E; about 1450 km), and is almost parallel to the directions of beams 13–15 and almost perpendicular to the MSTID phase fronts on the TEC map. We will show hereinafter a “keogram” along points 1 and 2, that is, time variation of the radar echo power, Doppler velocity, GPS-TEC perturbation amplitude or 630-nm airglow intensity deviation along the straight line.
Figure 11 displays keograms of the airglow deviation (Figure 11a) and GPS-TEC perturbation (Figure 11b) on the night of 11 June. Clear southwestward propagating MSTIDs with a phase velocity of about 120 m s−1, periods of 30–50 min, and wavelengths of 200–350 km are discernible, in particular, in Figure 11a. The bright airglow regions in Figure 11a are traced by the oblique lines that are also overlaid in Figure 11b, where the bright airglow regions coincide well with the enhanced TEC regions: again, we note the assumptions of the airglow altitude to be 250 km and the ionosphere to be a thin layer located at 300 km altitude. Such a spatial relationship was also found in an MSTID event over Shigaraki [Ogawa et al., 2002].
Figure 12a shows keograms of the radar echo power on beam 15 and the airglow intensity deviation along point 1 to point 2 in Figure 10. The 1000 km point in distance on the y axis corresponds to a radar range of about 200 km on beam 15. Note that the radar range is the observed range but not the ionospheric reflection point. As mentioned above, the F region radar echoes at ranges beyond about 600 km are due to GS scatter, and the echoes at 200–600 km ranges are due to Es FAIs. Clearly, the Es echo regions move with time in harmony with the movement of the enhanced and suppressed airglow regions, and appear within and near the suppressed airglow regions (therefore, suppressed TEC regions; see Figure 11) with periods of the MSTIDs. The Es echo and suppressed airglow regions indicated by the red lines with numbers 1–4 and number 5 have a trace velocity of 120 and 90 m s−1, respectively. The Es echo regions can be connected with the MSTID echo regions that are indicated by the black lines with numbers 1–5 whose trace velocities are 240 m s−1 for numbers 1–4 and 180 m s−1 for number 5, which are twice the trace velocities of the red lines. This fact is understandable because an ionospheric reflection point of GS scatter echo is considered to move toward the radar with a velocity of half of the movement of GS scatter point. Thus, the movement of MSTIDs detected with the radar and imager (therefore, GEONET) is very consistent with that of the Es echo regions.
Figures 12b and 12c display keograms of the Doppler velocity and airglow intensity deviation on beams 15 and 10, respectively. In Figure 12b most of the E region velocities are positive (FAI motion toward the radar). In Figure 12c, however, the velocities in the earliest striation are positive (negative) in the left side (right side), suggesting the existence of antiparallel FAI motion in and around the suppressed airglow region: the antiparallel motion is also discernible on beams 8–13 (not shown). Such antiparallel motions are also recognized in the striations during 1400–1510 UT. The velocities are mostly positive in other striations, similar to those in Figure 12b. One of the reasons why the antiparallel FAI motions are observed on beam 10 but not on beam 15 is supposed to be the difference in the beam direction and/or due to spatiotemporal change in Doppler velocities of the drifting Es FAIs.
 Simultaneous observations of summer nighttime Es echoes with the MU radar and MSTIDs with a 630-nm all-sky imager [Otsuka et al., 2007] or with GEONET [Saito et al., 2007] have revealed detailed spatial relationships between the Es echo regions and MSTID structures, showing that E region plasma drifts southeastward (northwestward) in the airglow or TEC enhancement (suppression) due to MSTIDs. These observations are not always consistent with our results showing that the northwestward plasma drifts also exist in some suppressed airglow/TEC region. We do not know exact reasons of this partial inconsistency, but one possible reason is that the HF radar waves were sometime refracted downward during propagation due to strong Es layers, as exemplified in Figure 10. When strong refraction occurs, detailed comparison between the radar and airglow maps may be meaningless: however, such a refraction effect is neglected in the MU radar cases mentioned above because of the high radar frequency of 46.5 MHz.
 We assumed an emission altitude of the airglow shown in Figure 12 to be 250 km. If this is not the case, the above described spatial relations among the echo region, Doppler velocity variation, and airglow region are not always true. However, estimating real emission altitudes and thickness of the airglow layer from the current airglow observations is impossible.
4. Summary and Discussion
 We have case-studied two MSTID events, one in winter (23 January 2007; event 1) and the other in summer (11 June 2007; event 2), observed with the SuperDARN Hokkaido radar, 630-nm all-sky imager, and GEONET under geomagnetically quiet conditions (mostly 0 ≤ Kp ≤ 1). These MSTIDs were accompanied by concurrent coherent echoes from Es FAIs. It is noted that MSTIDs at high latitudes appear even under quiet conditions [e.g., Ogawa et al., 1987; Crowley et al., 1987; Samson et al., 1990]. The results can be summarized as follows:
 1. In event 1 in January, daytime MSTIDs simultaneously detected with the radar and GEONET propagate toward SSE at about 170 m s−1 , and have a period of about 60 min and a wavelength of about 600 km with phase fronts aligned along ENE–WSW. Low Doppler velocities (mostly between −15 and +10 m s−1) and narrow spectral widths (mostly less than 20 m s−1) of radar echoes suggest that the daytime MSTID echoes are due to GS scatter.
 2. In event 1, characteristics of nighttime echoes are largely different from those of the daytime echoes. Nighttime MSTIDs propagate toward WSW at about 80 m s−1, and have a period of about 60 min and a wavelength of about 300 km with phase fronts aligned along NNW–SSE. High Doppler velocities (mostly between −120 and +60 m s−1) and wide spectral widths (less than 99 m s−1) of radar echoes suggest that the echoes are due to 15-m-scale F region FAIs. Negative (positive) Doppler velocity regions clearly coincide with strong (weak) echo regions (Figure 5). A comparison between E region Doppler velocity and 630-nm airglow maps indicates that velocities have northward (southward) component on the northeastern (southwestern) part of the enhanced airglow region (Figure 6c).
 3. In event 2 in June, nighttime MSTIDs propagate southwestward at about 120 m s−1 with periods of 30–50 min and wavelengths of 220–350 km. Bright airglow regions well correspond to enhanced TEC regions, suggesting that enhanced electron density in association with MSTIDs also contributes to the TEC enhancement. Contrary to the nighttime F region echoes in event 1 (Figure 4b), the nighttime MSTID echoes in event 2 are due to GS scatter, as suggested by low Doppler velocities (mostly between −30 and +25 m s−1) and narrow spectral widths (mostly less than 50 m s−1), which are comparable to those of the daytime echoes in event 1 (Figure 3a), and by Figure 12a.
 4. In event 2, nighttime echoes from FAIs in Es layers have Doppler velocities between −53 and +57 m s−1 and spectral widths mostly less than 90 m s−1. A comparison between radar echo and 630-nm airglow intensity keograms indicates that movement of MSTIDs is very consistent with that of Es echo regions, and that Es echoes appear within and near suppressed airglow (suppressed TEC) regions with periods of MSTIDs. Most of the E region velocities on beam 15 (the easternmost beam) are positive (i.e., FAI motion toward the radar), and velocities on beams 8–13 exhibit antiparallel motions in and near the suppressed airglow region. Again note that for the comparison between the Doppler velocity and airglow regions, we assumed an airglow emission altitude of 250 km.
4.1. Long-Distance Propagation of MSTIDs
 From a case study Kubota et al.  demonstrated that nighttime MSTIDs appeared simultaneously over Japan with a spatial extent of more than 2500 km, and that MSTID structures propagated southwestward over a horizontal distance of more than 1000 km. This paper supports these results, and moreover presents evidences that both nighttime and daytime MSTIDs in events 1 and 2 appear at the same time, at least, from 55°N to 25°N, as shown in Figures 4 and 8. We point out that nighttime MSTIDs propagated over a horizontal distance of about 4000 km, as inferred from Figures 2, 5, 7, and 11 and GPS-TEC maps. We do not know where these MSTIDs were generated at high latitudes to the north of Hokkaido. Auroral sources might not be responsible for the MSTID generation because of very low geomagnetic activities [e.g., Waldock and Jones, 1987]. Using the SuperDARN radars at auroral latitudes, Bristow et al.  showed that MSTIDs propagated over 2000–2500 km from their sources. Observations with the Hokkaido radar and two SuperDARN radars in Alaska have suggested a possibility of long-distance propagation (more than 3000 km) of daytime MSTIDs from auroral to midlatitudes [Ishida et al., 2008]. Ogawa et al.  have clearly demonstrated that nighttime MSTIDs that had traversed over Japan propagated southwestward beyond Taiwan at around 24°N over 3000 km. Meanwhile, we note that the propagation direction of nighttime MSTIDs detected with the Hokkaido radar is not always southwestward. In some cases the propagation direction changes from southwestward to northeastward before midnight: see an example reported by Shiokawa et al. , who have pointed out some role of F layer altitude decrease or poleward thermospheric wind in changing the direction.
4.2. F and Es FAIs Associated With MSTIDs
 The MU radar at Shigaraki has often detected echoes from nighttime 3.2-m-scale F region FAIs in summer [e.g., Fukao et al., 1991]. However, the Hokkaido HF radar capable of detecting echoes from 15-m-scale F region FAIs observed nighttime echoes in event 1 in January (the above item 2), but not in event 2 in June (item 3). In association with nighttime MSTID events on other days in the summer of 2007, F region FAI echoes were rarely observed with the Hokkaido radar. One of the reasons may be that in summer the HF radar wave cannot penetrate into the deep F region where 15-m-scale FAIs exist, but is reflected downward in the bottomside F region to cause GS scatter. Another possibility is that FAI echoes are masked by GS scatter echoes with power stronger than FAI echo power. On the other hand, FAI echoes from Es layers were detected in the daytime and nighttime in both events 1 and 2.
Saito et al.  made the first simultaneous MU radar and GEONET observations of MSTIDs in summer night to discuss a spatial relationship between FAI region and wavy TEC structure in the F region. From simultaneous all-sky imager and DMSP satellite observations of summer nighttime MSTIDs over Shigaraki, Shiokawa et al. [2003b] found that ions within dark (bright) airglow region that aligned along NW–SE drifted northwestward (southeastward) due to polarization electric field pointing to NE (SW). Using the MU radar and GEONET, Saito et al.  have shown that FAI patches causing radar echoes associated with summer nighttime MSTIDs drifted northwestward (southeastward) in suppressed (enhanced) TEC region that aligned along NW–SE, while the echo regions themselves moved southwestward in harmony with southwestward propagation of MSTIDs. An important finding from our nighttime observations at 45°–60°N in event 1 is that negative (positive) Doppler velocity regions clearly coincide with strong (weak) echo regions (item 2).
 Considering the above mentioned previous observations at 35°–38°N over Shigaraki, we suppose that our strong (weak) radar echoes due to strong (weak) FAIs associated with MSTIDs in event 1 (Figure 5) returned from suppressed (enhanced) 630-nm airglow/TEC regions, as shown in Figure 13a. In Figure 13a the negative (positive) Doppler velocity regions coincide with the strong (weak) FAI regions, and the Ep × Bo plasma drift is driven by the polarization electric field, Ep, induced by neutral wind and spatial undulation of the F region electron density, as has been discussed by Otsuka et al.  and Saito et al. , who also presented schematic pictures similar to Figure 13a. According to the recent work of Otsuka et al. , the northeastward Ep causing the northwestward plasma drift (i.e., negative Doppler velocity) in the suppressed airglow region strengthens effective eastward background electric field E + (U × Bo), where E is the background electric field and U is the neutral wind, to induce strong E × B instability (strong FAIs) in the bottomside of the F region. On the other hand, the effective eastward background electric field weakens in the enhanced airglow region to induce weak instability (weak FAIs). We think that the schematic picture in Figure 13a is also suitable to the summer night MSTID events over Shigaraki. In fact, using the MU radar, Fukao et al.  clearly demonstrated that intense and most turbulent echoes returned from high negative Doppler velocity regions, while rather weak echoes with positive Doppler velocities did from the edges of the intense echo regions.
4.3. E and F Region Coupling
 We have found that nighttime MSTIDs in events 1 and 2 are accompanied by concurrent coherent echoes from Es layers (items 2 and 4), suggesting the existence of electrical coupling between the E and F regions along the geomagnetic field. Figure 12 demonstrates that the nighttime Es echoes in event 2 are stronger in the suppressed airglow (therefore, TEC) areas in the F region. Such a spatial relationship is schematically illustrated in Figure 13b, where the plasma drifts causing positive Doppler velocities are supposed to be mainly toward SE (Figures 6c and 12b). The northwestward drifts adjacent to the southeastward drifts may coexist within and around the suppressed airglow regions (Figures 6c and 12c). Such kind of electrical coupling between the E and F regions at midlatitudes have been extensively studied, as mentioned in Introduction. A schematic picture similar to Figure 13b has been presented by Otsuka et al. , who made simultaneous observations of summer nighttime Es echoes with the MU radar and MSTIDs with a 630-nm all-sky imager, as mentioned in detail in section 3.2.
 The Hokkaido radar used a range resolution of 45 km with a horizontal beam width of about 5.4° (at 10 MHz) to observe Es echoes. These values are rather poor, compared with the MU radar case (600 m and 2.3°), and may not be enough to explore detailed physical processes of E and F region coupling. We have been occasionally operating the Hokkaido radar with a 15-km resolution since 13 June 2007 to collect MSTID examples with better spatial resolution. Anyway, we point out the existence of coupling at around 45°N (38°N magnetic), in addition to over Shigaraki (∼35°N; 25°N geomagnetic).
 HF radar, 630-nm all-sky imager, and GEONET observations of MSTIDs under geomagnetically quiet conditions revealed that MSTIDs in events 1 and 2 appeared at the same time, at least, at 55°–25°N. Nighttime MSTIDs propagated southwestward over a horizontal distance of about 4000 km, while daytime MSTIDs did southeastward. Nighttime MSTIDs in both events 1 and 2 detected with the radar were due to GS scatter and FAIs in the F region, respectively. We point out that strong (weak) nighttime F region echoes in event 1 returned from suppressed (enhanced) airglow/TEC areas. Nighttime MSTIDs were often accompanied by concurrent coherent echoes from FAIs in Es layers. In event 2, strong Es echo areas rather coincide with suppressed airglow/TEC areas in the F region that are connected with the echo regions along the geomagnetic field. This fact indicates the existence of E and F region coupling at night. Daytime MSTIDs detected with the radar were exclusively caused by GS scatter, in line with past MSTID observations at high latitudes.
 In the future, detailed analyses are required to examine possible E and F region coupling during the daytime. Though the origins of MSTIDs at midlatitudes are unknown, our MSTIDs might be generated at high latitudes. To pursue this issue and to explore a possibility of AGWs launched from the lower atmosphere to the north of Hokkaido, an extension of the observation area toward higher latitudes beyond the Hokkaido radar FOV is necessary in the future.
 Original GPS data and TEC data were supplied by the Geographical Survey Institute of Japan and Kyoto University, respectively. This work was supported by Grant-in-Aid for Scientific Research of the Ministry of Education, Culture, Sports, Science and Technology of Japan (19340141, 18403011, 20244080).
 Zuyin Pu thanks Alan Rodger and Michael Kelley for their assistance in evaluating this paper.