First observations of large-scale wave structure (LSWS) and the subsequent development of equatorial spread F (ESF), using total electron content (TEC) derived from the ground based reception of beacon signals from the CERTO (Coherent Electromagnetic Radio Tomography) radio beacon on board C/NOFS (Communications/Navigation Outage Forecasting System) satellite, are presented. Selected examples of TEC variations, using measurements made during January 2009 from Bac Lieu, Vietnam (9.2°N, 105.6°E geographic, 1.7°N magnetic dip latitude) are presented to illustrate two key findings: (1) LSWS appears to play a more important role in the development of ESF than the post-sunset rise (PSSR) of the F-layer, and (2) LSWS can appear well before E region sunset. Other findings, that LSWS does not have significant zonal drift in the initial stages of growth, and can have zonal wavelengths of several hundred kilometers, corroborate earlier reports.
 Equatorial spread F (ESF) is a generic name, which refers to the presence of a wide spectrum of field-aligned irregularities in the equatorial nighttime F-region that can have spatial scales from 100 km to 10 cm. The importance of ambient ionospheric and thermospheric conditions such as vertical plasma drifts, plasma density gradients, zonal, meridional and vertical neutral winds, initial seed perturbations etc., in the initiation and non-linear development of ESF and its dynamics has been well-recognized [e.g., Sekar and Raghavarao, 1987; Sekar and Kelley, 1998]. The linear growth rate of generalized Rayleigh-Taylor (R-T) instability, which directly and indirectly depends on all these factors, is often used as an indicator of when and where ESF can occur, if there are seed perturbations present [Sultan, 1996]. But, in any season and solar epoch, under seemingly identical ionospheric conditions, ESF might occur on one night and might be absent on another, and this enigmatic day-to-day variability still poses a challenge to the complete understanding of this phenomenon. For instance, though the occurrence of ESF is climatologically related to the post-sunset rise (PSSR) of the F-layer - which is often regarded as the most important pre-requisite for ESF- it can appear to be unrelated to ESF when examined on a day-to-day basis [Hysell and Burcham, 2002]. In an alternative approach, Tsunoda  reported that the presence of a large-scale wave structure (LSWS) in the bottom side F-layer actually dictates the subsequent development of ESF. It was reported that the first appearance of LSWS could be just before the E region sunset. The LSWS can be identified as a quasi-periodic modulation in the altitude of isoelectron density contours in the bottomside F-region, superimposed on a mean slope that increases in altitude from west to east; the latter is consistent with the PSSR [Tsunoda and White, 1981]. The zonal wavelength of LSWS was determined to be ∼400 km, and this corroborated with the earlier observations of Röttger  that ESF occurs quasi-periodically in the longitudinal direction. It was pointed out that unlike PSSR the presence of LSWS is correlated with the occurrence of ESF virtually on a day-to-day basis [Tsunoda, 1981; Tsunoda and White, 1981; Singh et al., 1997; Tsunoda, 2005]. The observations of LSWS were mostly based on measurements by ALTAIR, located at Kwajalein Atoll, (9.4°N, 167.5°E, 4.3° dip lat) [Tsunoda, 1981; Tsunoda and White, 1981; Tsunoda, 2005], in-situ probes [Singh et al., 1997], and those of satellite to ground total electron content (TEC) [Tsunoda and Towle, 1979] using a satellite in a low inclination orbit. Abdu et al.  showed that “satellite” traces preceded range type ESF in the ionograms over Fortaleza. Recently, such “satellite” traces in ionograms [Tsunoda, 2008] and multi-reflected echoes in the ionograms [Tsunoda, 2009], both obtained from Kwajalein Atoll, were identified as signatures of LSWS. In another interesting study, Saito and Maruyama  used virtual heights of the bottom side F-layer at two locations separated by 6.3° in longitude to infer the presence of LSWS. The zonal structure of radar backscatter plumes associated with ESF, probably modulated by atmospheric gravity waves, has also been investigated by Fukao et al. . However, LSWS is not yet widely accepted by the research community as the most reliable precursor to ESF, due to a dearth of observations. It should also be mentioned here that LSWS is not easily detectable with overhead measurements using a sensor at a fixed location, especially during its early growth phase, mainly because initially it grows in amplitude without significant zonal drift [Tsunoda and White, 1981; Tsunoda, 2005].
 In this context, since 2008, the beacon transmitter namely CERTO (Coherent Electromagnetic Radio Tomography) on-board C/NOFS (Communications/Navigation Outage Forecasting System, orbital inclination of 13°, see de La Beaujardière and C/NOFS Science Definition Team  and Bernhardt and Siefring  for details) satellite provides a unique way to measure the longitudinal structure of TEC which can be used to detect the LSWS. We report the first observations of LSWS and subsequent ESF development using CERTO beacon on board C/NOFS satellite.
 Recently, we have installed a GNU (GNU is not UNIX) Radio Beacon Receiver (GRBR, see Yamamoto  for details) at Bac Lieu. The C/NOFS as well as the other low-earth orbiting satellites like the Cosmos and OSCAR have been continuously tracked and the line of sight relative TECs are obtained from the differential phase information, using the 150 and 400 MHz transmissions. Typical data duration is 12–15 minutes for each satellite pass. The line of sight TECs are then projected into the relative vertical TECs. As the C/NOFS orbits close to the equatorial plane, we obtain the longitudinal variation of relative TEC (whereas the satellites in the high-inclination orbit give the latitudinal variation), and the perturbation component is derived by subtracting a 2.5 minute running average, which corresponds to a zonal distance of ∼1000 km. This is done keeping in mind that the zonal wavelength of LSWS is usually in the range 100–700 km [Weber et al., 1980; Röttger, 1973].
 Two ESF nights and one non-ESF night observations during the period January 22–25, 2009 are presented. The frequency modulated-continuous wave (FMCW) sounder data from Bac Lieu, and Chumphon (10.7°N, 99.4°E, 3.3° dip lat) and the 30.8 MHz VHF radar observations from Kototabang, Indonesia (0.20°S, 100.32°E, 10.36°S dip lat) are also used. The ionosondes belong to the SEALION (Southeast Asia Low-latitude Ionosonde Network) operated by NICT [Maruyama et al., 2007]. The technical details of the VHF radar are given by Otsuka et al. . The locations of the observation sites and the track of a C/NOFS pass are shown in Figure 1 (top).
3. Results and Discussion
Figures 1a–1d show the TEC observations from four passes of C/NOFS on 22 January 2009. The blue curve in each plot represents the variation in the perturbation component of TEC with the ionospheric penetration point (IPP) longitude. Previous observations of the typical altitude structure of the LSWS show that the density modulations exist in the bottomside of the F-region and the region above this modulation is more or less horizontally stratified [Tsunoda and White, 1981; Tsunoda, 2007]. Hence, the observed modulations (upwellings and downwellings) in the perturbation TEC are directly interpreted as a signature of similarly varying plasma density (LSWS) in the bottom side of the F-region [Tsunoda and Towle, 1979]. The beginning of the first pass (Figure 1a) was at 12:19:33 UT (LT = UT + 7 hrs at Bac Lieu), and at 95°E IPP longitude, the solar zenith angle was 97.5°, which means the solar shadow height was 55.9 km, and the E region was still sunlit. The presence of LSWS is evident even at this time. The zonal wavelength was found to be ∼500–700 km. This is the first direct evidence that the LSWS can form before the E region sunset itself. Four crests can be easily identified at ∼96°E, 101°E, 109°E and 116°E. The next pass (Figure 1b) shows that LSWS still persisted and small-scale irregularities started to appear. An interesting feature that can be seen in Figure 1b is that the west wall of the eastern-most crest region first shows small perturbations. This is similar to the observations by Tsunoda [1981, 1983] and Tsunoda and White  that the ESF backscatter did not initially develop at the top of the eastern crest, where the growth rate of the R-T instability would be largest. Instead, the backscatter was initiated on the west wall of the east crest. However, we cannot add further observational evidence to this point, since we do not have simultaneous radar observations at these locations.
 The ESF onset, determined from ionograms taken at Chumphon, was at 2020 LT, and the h'F was 294 km at this time. The ESF continued even after midnight. The ESF onset at Bac Lieu was at 2125 LT, which also continued into the post-midnight hours. The maximum PSSR rate was 46 m s−1 at Chumphon (at 2010 LT) and 20 m s−1 at Bac Lieu (2115 LT). It must be noted that the F-layer at Bac Lieu appeared to descend from 2030 to 2040 LT but later it started to ascend again. This indicates that the ESF could have drifted toward the vicinity of the Bac Lieu sounder, and did not originate there. In contrast, the ESF observed at Chumphon could be triggered there itself by all probability. If we carefully see the LSWS pattern at 1920 LT, and consider the fact that the LSWS does not show much zonal drift, we can associate the ESF at Chumphon to the crest of the upwelling (near 101°E). Although we cannot discriminate the irregularities that are generated over the upwelling near 101°E from those drifted from the other upwellings/crests of the LSWS, the higher PSSR rate, earlier onset and the continuous ascending of the ionogram traces until the onset help us to understand that the ESF was triggered over Chumphon, where the crest of the LSWS formed. Unlike this, the late onset and the descending/ascending of the ionogram traces, enable us to conjecture that the ESF was not triggered over Bac Lieu.
 However, since there is no capability of echo direction finding for the FMCW sounders, we cannot exactly pinpoint the longitude of the generation of irregularities and relate it to the observed crests of the LSWS, either at the growth phase of the ESF or at later stages. For making such conclusions about the growth of irregularities, especially when the ESF is simultaneously observed by ionosondes separated by few degrees in longitude, the observations using a steerable radar like ALTAIR (which provides ∼1000 km scan in the zonal direction) would be more appropriate [Tsunoda, 1981]. The advantages of C/NOFS observations are that they can be used to detect the LSWS in a large-longitudinal area and the zonal structure of the LSWS could be determined. A new finding by this means is that LSWS can form even before the E region sunset. This finding suggests that the PSSR probably does not produce the LSWS. As we move to earlier times before E region sunset, the conductivity control by E region dynamo will be stronger, exceeding that by the F region dynamo. So, in general, the likelihood that dynamo and polarization effects in the E region are controlling LSWS generation increases at earlier and earlier times and this provides a favorable evidence for the generation mechanism of LSWS discussed in detail by Tsunoda . The very fact that the LSWS is found before PSSR, excludes PSSR as a candidate for the generation of LSWS. These observations provide a simple method for detecting the LSWS on a continuous basis, which can provide a valuable database for the formation of LSWS and its relation to the occurrence of ESF.
 The satellite pass at 0031 LT (Figure 1c) shows well-defined plasma depletions in TEC, with the most severe depletions between 100°E and 105°E. The last pass (Figure 1d) shows that the irregularities persisted over this region even after midnight. The VHF radar at Kototabang showed the field aligned irregularity (FAI) echo from 0030 LT (next day) which lasted up to 0300 LT (Figure 1e). It should be mentioned here that the ESF needs to reach 520 km over the geomagnetic equator, in order that the irregularity echoes can be observed at 300 km over Kototabang. The radar sees only a narrow region above Kototabang, which is indicated by plus symbols in the TEC plots. It should also be mentioned here that the elevation of the satellite pass over Kototabang was only 25–27° for the pass shown in Figure1c, and hence we do not intend to do direct comparison of each small-scale depletion with the radar echo, because such low-elevation TEC measurements cannot be simply related to longitudinally-confined and vertically-extended radar echoes [Tsunoda and Towle, 1979]. The comparison is only to show that small scale irregularities existed over this longitude region at close to midnight and post-mid night hours, and had an extent up to ∼10°S magnetic latitude.
Figures 2a–2d show the TEC observations from four consecutive passes of C/NOFS on 25 January 2009. The beginning of the first pass (Figure 2a) was at 11:30:39 UT (LT = UT + 7 hrs at Bac Lieu), and at 95°E IPP longitude, the solar zenith angle was 86.56°. This means that the solar shadow height was zero, and the E region was still sunlit. Four crests can be easily identified at ∼103°E, 109°E, 112°E and 117°E. In this case, the E region sunset over Bac Lieu was at 1844 LT, which means that the E region was sunlit even while the C/NOFS passed over Bac-Lieu, thus giving further confirmation that LSWS can form well before the E region sunset. It should be mentioned here that these are the first observational evidence for the formation of LSWS well before the E region sunset. The ESF onset at Bac Lieu was at 1930 LT (h'F = 268 km), which also continued up to post-midnight hours, whereas there was no post-sunset ESF at Chumphon. However, it should be mentioned here that, the PSSR was very weak both over Bac Lieu (maximum ∼10 m s−1 at 1925 LT) and Chumphon (also ∼10 m s−1 at 1925 LT) on this day indicating that the PSSR cannot be the single reliable factor determining the occurrence/non occurrence on a given night. ESF was first generated over Bac Lieu, which is situated near the west wall of the first upwelling at 103°E (Figure 2a). These are the first direct observational evidence for the theory proposed by Tsunoda [1981, 2005] that the irregularities first appear on the west wall of an upwelling.
 The next satellite pass (Figure 2b), at 2014 LT shows that the amplitude of the LSWS has increased, especially between 104°–117°E. This also corroborates with the earlier observations [Tsunoda and White, 1981]. After ∼2 hrs, the entire region between 104°–117°E shows irregular structures, which continues further. In these two passes, there is some zonal drift of the LSWS, and there are some weak irregularities in the 95°E–104°E region also, at 2341 LT (Figures 2c and 2d), even though there is no clear ESF observed in the ionograms from Chumphon. Further continuous observations are needed to form more robust conclusions about such events. Figures 2e–2h show the TEC observations from four passes of C/NOFS on 23 January 2009, which is a non-ESF day, both over Bac Lieu and Chumphon. It can be seen that there is no significant LSWS formation, at 1938 LT, and there is no subsequent irregularity generation. The TEC remain rather smooth, even at post midnight hours.
 As mentioned earlier, the C/NOFS observations provide the zonal picture of the irregularities. To see the field-aligned nature of the irregularities observed by C/NOFS, we also looked into the data obtained by polar orbiting satellites. Figure 3 shows the latitudinal variation of TEC and the perturbations obtained using a Cosmos 2407 pass at 2135 LT, on January 25, 2009. It can be seen that the amplitude of the TEC variations is smaller along the meridional direction than along the zonal direction than along the meridional direction, due to the field-aligned nature of the perturbations. This is an indication that the irregularities detected by C/NOFS correspond to ESF irregularities. This result also signifies that the TEC variation has a wave elongated along the meridional direction. Consequently, modulation of the ionosphere in the zonal direction becomes an important parameter for the plasma bubble generation.
 The first observations of LSWS and subsequent ESF development by ground based TEC measurements using radio beacons on board C/NOFS satellite are presented. The observations indicate that the LSWS can form much before the E region sunset, which indicates that the source of the LSWS is not PSSR, but is perhaps located in the E region. The LSWS does not have significant zonal drift in the initial stages, and the zonal wavelength appears to be ∼500–700 km, which corroborates with the earlier results. The irregularities form around the crests of the LSWS, mostly initiated as small-scale structures near the west walls of the upwellings. These observations further confirm that the presence of LSWS is a necessary pre-requisite for the subsequent ESF development. It must be mentioned here that the excitation of LSWS has a theoretical basis also [Tsunoda, 2007]. The continuous long-term C/NOFS observations from several ground stations in the equatorial region would provide a simple method to detect the LSWS, which would provide a better understanding to the enigmatic day-to-day variability of ESF.
 The work of ST is supported by the Japan Society for the Promotion of Science (JSPS) foundation; that of RTT was supported by the National Science Foundation under grant ATM-0720396. We thank Susumu Saito for providing the software to scale the ionograms. ST thanks Huixin Liu for the careful reading of the manuscript.