Variations in ionospheric height associated with the prereversal enhancement (PRE) at two equatorial ionosonde stations separated by 6.34° in longitude were studied for the first time. The stations used were in Chumphon (10.7°N, 99.4°E, 3.3 Mag. Lat.), Thailand and Bac Lieu (9.3°N, 105.7°E, 1.6 Mag. Lat.), Vietnam. Variations in the virtual height of the bottomside of the F region (h′F) at 2.5 MHz observed at these two stations were analyzed for a period in March and April 2006. When the equatorial spread F (ESF) was not observed, h′F variations at the two stations were very similar, with a local time shift corresponding to the longitude separation. However, when ESF was observed, they were often significantly different. Our results show that h′F enhancement, which could be interpreted as being due to the eastward electric field enhancement, is quite localized in longitude when ESF occurs.
 Nighttime equatorial spread F (ESF) is a manifestation of ionospheric irregularities in the F region. Such irregularities cause scintillations of satellite signals, resulting in degradation of satellite communications and, in the worst case, in lock-off of satellite signals. The basic mechanism of ESF, the Rayleigh-Taylor instability associated with the strong prereversal enhancement (PRE) of the eastward electric field after sunset, is well known. The most important unsolved problem related to ESF irregularities is the day-to-day variability of their occurrence. Understanding what controls the day-to-day variability is essential to predicting their occurrence.
 Quasi-periodic occurrence of ESF in the longitudinal direction has been observed by transequatorial HF radio wave propagation [Röttger, 1973; Maruyama and Kawamura, 2006], airborne airglow imager measurements [Weber et al., 1980], satellite measurements [Lin et al., 2005], incoherent scatter radar measurements [Tsunoda, 2005], and coherent backscatter radar measurements [Fukao et al., 2006]. Table 1 summarizes typical longitudinal lengths of the quasi-periodic appearance of ESF. The importance of large-scale horizontal structures of the F region to the day-to-day variability of ESF occurrences has been pointed out by Tsunoda  and Fukao et al. . The large-scale structures reported by previous studies are, however, smaller than the horizontal extent of PRE.
Table 1. Zonal Scale Length Reported in the Literature
Tsunoda  used data from ALTAIR incoherent scatter radar measurements at Kwajalein Atoll [Tsunoda and White, 1981] to show that longitudinal large-scale wave structures (LSWSs) with a typical wavelength of 400 km in the F region always exist when plasma bubbles are generated. He also found that the LSWSs were standing waves in the frame of view of an observer fixed on the earth's surface; i.e., they corotated with the earth and did not propagate with respect to the observer. Recently, Tsunoda and Ecklund  studied upwellings associated with PRE using a VHF backscatter radar and found that there are two velocity components in the local vertical plasma drift, one about the scale size of PRE and the other about that of LSWSs.
 Using the Equatorial Atmosphere Radar in Indonesia, Fukao et al.  showed that multiple plasma bubbles were often generated at distances from each other of 370–1000 km. They presumed that the longitudinal structure originated from atmospheric gravity waves may directly or indirectly modulate the ionosphere.
 The longitudinal variation of the electric field associated with PRE has also been studied by Fejer et al. . They showed that the difference between vertical drift velocities measured at Jicamarca and Huancayo, both in Peru and separated by 170 km in the longitudinal direction, had significant day-to-day variability, though they were, on average, similar to each other. They attributed the difference to the difference in measurement equipment, an incoherent scatter radar at Jicamarca, and an ionosonde at Huancayo.
 All of these previous studies pointed out the importance of longitudinal ionospheric structures, among other mechanisms, with a scale size of several hundred kilometers. However, the mechanism that generates the longitudinal structure is still unclear. For example, Tsunoda  and Fukao et al.  analyzed only a very limited number of examples. Furthermore, the distance between Jicamarca and Huancayo is not large enough to see the effect of structures spanning several hundred kilometers. Fukao et al.  observed irregularities associated with plasma bubbles, but information on the background ionosphere was lacking. Continuous long-term observations are necessary.
 To understand the effects of longitudinal structures more clearly, we conducted continuous ionosonde observations at two Southeast Asian sites separated by a distance of about 700 km. This is the first experiment ever to observe the longitudinal variability in the ionospheric height change associated with PRE.
 We studied the day-to-day variability in ESF occurrences and longitudinal variations of the ionospheric height and examined the importance of longitudinal ionospheric structures.
 We used data from two ionosonde stations near the magnetic equator at Chumphon (10.7°N, 99.4°E, 3.3° Mag. Lat.), Thailand and Bac Lieu (9.3°N, 105.7°E, 1.7° Mag. Lat.), Vietnam, which are a part of Southeast Asia Low-latitude Ionospheric Network (SEALION) [Maruyama et al., 2007]. Chumphon and Bac Lieu are separated by 6.34° in longitude (738 km at the altitude of 300 km). The local time difference between the two stations is 25.4 min.
 The observation parameters are summarized in Table 2. The ionosondes transmit radio waves sweeping the frequency range from two to 30 MHz and receive echoes from the ionosphere every five minutes. The bottomside plasma density profile can be determined by these echoes.
Table 2. Observation Parameters of Ionosondes
Frequency Modulated–Continuous Wave (FM-CW) with Pseudo-Random Tx/Rx Switching
Peak Tx power
Average Tx power
100 kHz s−1
Sweep repetition period
 Simultaneous observation at Chumphon and Bac Lieu started on 5 December 2005. Since then, it has been continuously conducted with only minor interruptions for maintenance.
3. Analysis and Results
 As was shown by Bittencourt and Abdu , the virtual height of the bottom of the equatorial F region (h′F) after sunset provides information on the dynamic processes in the ionosphere. The h′F after sunset is close to the real height and its measurements are used as a measure of the vertical motion of the ionospheric F region. The h′F at 2.5 MHz was manually scaled every five minutes. ESF appearance was also visually identified every five minutes. Range-type ESF that indicates existence of small-scale irregularities was considered in our analysis. Unusual F region traces were classified into strong ESF, weak ESF, and multi-trace echo. Strong ESF refers to a completely diffuse F region trace in which the retardation curve near the critical frequency is hardly recognizable. Weak ESF refers to a diffuse F region trace in which the retardation curve can be recognizable. Multi-trace echo refers to two or more discrete F region traces.
 ESF occurrences in this longitude sector are highest in equinox seasons [Maruyama and Matuura, 1984; Burke et al., 2004; Otsuka et al., 2006]. We analyzed ionograms obtained at Chumphon and Bac Lieu from 10 March to 10 April 2006. Figure 1 shows examples of h′F variations and ESF occurrences. On 22 March 2006 (Figure 1a), weak PRE was observed at both Chumphon and Bac Lieu, and no ESF was observed at either site. The h′F variations at both stations were very similar and displayed a delay of 20–30 min, which corresponds to the local time difference between the two stations. On 27 March 2006 (Figure 1b), the h′F variations and ESF at the two stations were surprisingly different. Strong PRE and ESF were observed at Chumphon, while almost no PRE was observed at Bac Lieu. The difference in maximum h′F between the two stations was about 100 km. No strong ESF associated with PRE was observed at Bac Lieu. The strong ESF observed after 15 UT at Bac Lieu associated with a secondary h′F peak may be the one observed at Chumphon at 13 UT, if it is assumed that ESF irregularities drift eastward at an average velocity of 100 m s−1, which is a typically observed value [e.g., Fukao et al., 2006]. In contrast, on 28 March 2006 (Figure 1c), strong ESF was observed at Bac Lieu, while only weak ESF was observed at Chumphon. In this case, the maximum h′F at Bac Lieu was about 70 km higher than at Chumphon. ESF was observed at the two sites after 14 UT associated with secondary h′F peaks. The ESF could also be generated far away from these sites and drifted over the sites. Since magnetic activity was quiet on these two nights, it is unlikely that the secondary peaks were caused by the penetrating magnetospheric electric field, and it is not very clear that the ESF was triggered on-site.
Figure 2 shows the maximum h′F associated with PRE and strong ESF occurrences at Chumphon and Bac Lieu between 19–21 LT for the period 10 March to 10 April 2006. It can be seen that large differences in the maximum h′F between the two stations were associated with strong ESF events. The analysis took into account the strong ESF that was observed between 19–21 LT, because ESF occurring at later local times is supposed to be generated far away from the observing site and flow into the field of view. In contrast, when the maximum h′F was low at both stations, no strong ESF was observed, and the h′F variations at the two stations were very similar to each other. The mean difference of the PRE peak time in UT at the two stations were 14 min, which is close to the local time difference as typically seen in Figure 1a. Sometimes ESF was not observed even when the maximum h′F at both stations were as high as when ESF was observed. The explanation could be that other factors, such as transequatorial meridional wind [Maruyama and Matuura, 1984; Maruyama, 1988; Saito and Maruyama, 2006], or solar activity variation [Jayachandran et al., 1993], might have worked to suppress the ESF development.
 There may be longitudinal differences in the ionospheric height even when the maximum h′F at both stations are similar, because the local times of h′F peaks at Chumphon and Bac Lieu are sometimes different. Figure 3 plots the h′F at Bac Lieu versus h′F at Chumphon for the same local time between 18 and 22 LT when neither strong nor weak ESF was observed (10 days) and when strong ESF was observed (12 days). The correlation coefficient between h′F at Chumphon and Bac Lieu was 0.71 when no ESF was observed. In contrast, the correlation coefficient was 0.39 when strong ESF was observed at either or both of the stations. This result clearly shows that h′F variations at Chumphon and Bac Lieu are quite different when strong ESF occurs.
4. Discussion and Summary
 We have presented the results from observations by longitudinally separated (738 km at the altitude of 300 km) ionosonde stations at Chumphon and Bac Lieu (at the magnetic equator). The h′F variations associated with PRE at these two stations were very similar when PRE was not very strong and ESF was not observed. In contrast, h′F variations at the two stations were often very different when ESF was observed at one or both stations. Sometimes they were similar to each other, when PRE was strong and ESF was observed at the two stations. 1 April 2006 was such a case. This could be attributed to a longitudinal structure of which wavelength is equal to the distance between the two sites, although we need more stations along the magnetic equator to clarify this hypothesis. With the same magnetic declination angle and the magnetic equator offset from the geographic equator, theoretical studies show that the local time variation of the ionospheric height associated with PRE is equivalent to the longitudinal variation [Batista et al., 1986; Fejer, 1997]. The typical PRE duration is two hours around the local time of sunset [Woodman, 1970]. A local time of two hours corresponds to the longitudinal extent of 30° which is about 3000 km at the equator. In Southeast Asia, the magnetic declination angle and the magnetic equator offset from the geographic equator are almost equal. Therefore, when the PRE structure is described in the sun-fixed coordinates with a longitudinal scale of about 3000 km (∼2 hours in local time), the strong variability of h′F at the longitudinal distance of 700 km is not expected. Our observations, however, show that the view of the smooth large-scale (∼3000 km) PRE structure is only valid when ESF activity is low. The temporal variation of h′F around sunset over one station cannot be mapped onto the other station even though it is separated by 700 km.
 The relationship between the h′F at the two stations is also highly variable. That is, sometimes one is higher than the other, and sometimes they are the same height. This indicates that the scale length of the longitudinal structure associated with ESF is variable. Although it is not possible to distinguish whether the longitudinal structure is propagating or standing, because we have only two observing stations, our results show that the ionospheric height and hence the plasma density are structured with a scale smaller than that of sun-fixed PRE, and this conclusion is consistent with the suggestion by Tsunoda .
 An important finding in this study is that the longitudinal structure much smaller than that of PRE appears to be closely related to ESF occurrences on a day-to-day basis, and supports previous studies [Tsunoda, 2005; Fukao et al., 2006]. It is not, however, clear how important this longitudinal structuring of the ionosphere is compared to other mechanisms proposed to explain the day-to-day variability. Our results show that h′F enhancement, which could be interpreted as being due to the eastward electric field enhancement, is quite localized in longitude when ESF occurs. This could be resulted from collisional shear instability [Hysell and Kudeki, 2004] or spatial resonance of atmospheric gravity waves [Kelley et al., 1981; Hysell et al., 1990]. However, we do not have enough information to speculate about the mechanism responsible for forming the longitudinal structure in the ionosphere. More data on the precise scale size, periodicity, and propagation of the structure is necessary to investigate its nature. A longitudinal chain of dense ionospheric observation with longitudinal resolution smaller than the scale size of the ionospheric longitudinal structure would be desirable to monitor ESF occurrence and predict its propagation.
 The ionosonde stations at Chumphon, and Bac Lieu are operated under agreements between NICT, Japan, and King Mongkut's Institute of Technology Ladkrabang (KMITL), Thailand and the Vietnamese Academy of Science and Technology (VAST), respectively.