In this paper we present the first results of F region field-aligned irregularities (FAI) made during the summer of low solar condition using the Gadanki mesosphere-stratosphere-troposphere radar. FAI echoes were observed on all 20 nights of radar observations and were mostly confined to the postmidnight hours. Echo morphology is found to be very different from the equinoctial postsunset features reported earlier from Gadanki. Echo SNRs are lower by 25 dB than their equinoctial postsunset counterparts but are quite comparable to those of the equinoctial decaying FAI during the postmidnight hours. The Doppler velocities, which lie in the range of ±100 m s−1, are predominantly upward-northward during 0000–0300 LT and downward-southward afterward, in contrast to those observed as predominantly downward-southward associated with the decaying equinoctial postmidnight F region FAI. Spectral widths of the summer echoes, which are well below 50 m s−1 and are very similar to those of the decaying equinoctial irregularities, represent the presence of weak plasma turbulence. Simultaneous observations made using a collocated ionosonde show no ionogram trace during 2200–0530 LT except for a few cases of weak spread F events. Concurrent ionosonde observations made from magnetic equatorial location Trivandrum also show very similar results. The observations are discussed in the light of current understanding on the postmidnight occurrence of F region irregularities in the summer of low solar condition.
 A unique feature of the ionosonde observations of F region irregularities, characterized as spread F echoes (commonly known as spread F), from the Indian longitudes is the solar activity dependent seasonal and local time variations [Chandra and Rastogi, 1970, 1972]. During solar maximum condition, the occurrence of spread F peaks in the postsunset period with occurrence rate higher in equinoxes than in solstices. During solar minimum condition, however, the postsunset peak is seen only in the equinoxes and December solstice, while the occurrence maximum in June solstice shifts to postmidnight hours. Interestingly, in the solar minimum, the occurrence rate of spread F in June solstice is higher than those in other seasons [Chandra and Rastogi, 1972; Subbarao and Krishnamurthy, 1994].
 In the postmidnight hours, they were observed as frequency spread F [Chandra and Rastogi, 1972]. Sastri et al.  and Subbarao and Krishnamurthy  showed that on many occasions range spread F observed in the postsunset hours gradually turned into frequency spread F as time progressed. Subbarao and Krishnamurthy  observed that while most of the time postmidnight frequency spread F was preceded by range spread F, during the summer of low solar activity, the postmidnight frequency spread F occurred without prior occurrence of range spread F.
Subbarao and Krishnamurthy  found that the postmidnight occurrence of spread F in the summer of solar minimum was related to the prior height rise of the F layer and interpreted their observations in terms of generalized collisional Rayleigh-Taylor (RT) instability. Sastri  also observed F layer height rise prior to the occurrence of postmidnight spread F during the June solstice of solar minimum. He, however, found that height rise of the F layer during the postmidnight hours did not always lead to spread F. He surmised that the vertical neutral wind associated with the equatorial midnight temperature maximum (MTM), which occurs in summer and has a high degree of day-to-day variability, might be a potential candidate for altering the growth of the RT instability [Sekar and Raghavarao, 1987] resulting in nonoccurrence of spread F on some nights despite the F layer height was high. Niranjan et al. , using satellite based temperature observations, found that the seasonal variation and occurrence rate of postmidnight spread F were broadly related to the MTM phenomenon. A definitive answer to what is responsible for the post midnight spread F during the low solar condition, however, was not obtained.
 The above investigations on the occurrence of F region irregularities in the summer of low solar condition, however, were based on ionosonde observations. Thus, radar observations of F region field-aligned irregularities (FAI), displaying varying morphology in the radar maps (bottomside, bottom-type, and plume), associated with spread F are expected to help understand the processes involved [Woodman and La Hoz, 1976]. Although some radar studies of postmidnight F region FAI during the summer of low solar condition have been made from Jicamarca [e.g., Fejer et al., 1999; Hysell and Burcham, 2002], such efforts have not been made yet from the Indian sector.
 To study the summertime F region irregularities in low solar condition, we conducted observations on twenty nights during July–August 2008 using the mesosphere-stratosphere-troposphere (MST) radar and an ionosonde located at Gadanki (13.5°N, 79.2°E, 6.4°N magnetic latitude (MLAT)). All observations correspond to geomagnetically quiet condition (Kp < 3). To our surprise, radar echoes from the F region FAI were detected during postmidnight hours on every night. In this paper we present these observations and study them in conjunction with concurrent ionosonde observations made from Gadanki and from magnetic equatorial location Trivandrum (8.5°N, 77°E, 0.5° MLAT). The radar observations of F region FAI reported here are first of its kind from Gadanki and are expected to shed light in understanding the origin of the postmidnight spread F in the summer of solar minimum.
2. Brief Observational Overview
 In May 2008, an IPS-42 ionosonde started functioning from Gadanki. This ionosonde was earlier located at Sriharikota and has been sifted to Gadanki. Subsequently, we conducted simultaneous observations using the ionosonde and the MST radar during two campaign periods: 1–10 July 2008 (first campaign period) and 30 July–8 August 2008 (second campaign period). Thus, these are the first such simultaneous observations from Gadanki. Also the radar observations are the first of its kind from Gadanki as far as summertime F region FAI are concerned. Ionosonde observations could not be made on 2, 5, 6, 7 and 10 July due to power supply problem. A similar ionosonde was also operated from Trivandrum. Ionosonde observations both at Gadanki and Trivandrum were made at 15 min interval.
 The Gadanki MST radar beam was pointed at 15° off zenith due magnetic north to detect the echoes from the F region FAI. Important radar parameters used for the observations are given in Table 1. Radar returns were sampled pulse by pulse and power spectrum was computed online using 256 pulse returns. Also, 8 spectra were averaged online before storing the spectral data for postprocessing. For the specifications given in Table 1, power spectral data were obtained with range and time resolutions of 2.4 km and 30 s, respectively. Signal-to-noise ratio (SNR), mean Doppler velocity, and spectral width (i.e., 2σ, where σ is the square root of velocity variance) were obtained by estimating the three lower order moments. SNR computation was made using noise power reckoned over the entire spectral window of ±283 m s − 1.
Table 1. Radar Parameters Used for Studying the F Region FAI
Peak power-aperture product
3 × 1010 W m2
Beam (3 dB)
Interpulse period (IPP)
Number of fast Fourier transform points
Number of spectral averaging
Nyquist velocity limit
±283 m s−1
2.28 m s−1
3.1. Radar Observations
3.1.1. Echo SNR and Morphology
 Height-time variations of SNR of the F region FAI echoes observed in the first and second campaigns are illustrated in Figures 1 and 2, respectively. FAI echoes observed on 4, 7, and 8 August were weaker and for shorter duration than those of other nights and have been shown as expanded in the insets. Notably, echoes were detected on all twenty nights and were observed mostly in the postmidnight hours (only on 7 July and 2 August, echoes were also observed between 2200 and 0000 LT). It may be mentioned that the sunset and sunrise at 300 km altitude during July–August occur at ∼2000 LT and ∼0430 LT, respectively. Patra et al. [2004, 2005] have shown that in equinox and magnetically quiet conditions, F region echoes commence around 1930 LT, which is close to the sunset time (∼1930 LT in March equinox and ∼1900 LT in September equinox). Thus the onset time of the echoes in summer reported here is much later than the sunset time at F layer altitude and also is in contrast with the equinoctial observations [Patra et al., 2004, 2005]. Furthermore, on ten out of twenty nights, radar echoes were observed till sunrise and beyond (sunrise time at the F region (300 km) is 0430 LT). Also on several occasions, the echoing region is found to extend to altitude as high as 500 km. It may be mentioned that the F region altitude of 500 km over Gadanki maps to 600 km over the magnetic equator.
 Echo SNR is found to be in the range of −15 to 5 dB. Echoes having SNR less than −15 dB are regarded as noise. This implies that the strongest signal observed during the two campaigns is 20 dB above noise. These SNR values are about 25 dB lower than those observed as 45 dB during postsunset hours in equinoctial period of 1998 (low solar condition) [Patra et al., 2005], but are comparable to those of the postmidnight observations [Patra et al., 2005; Sekar et al., 2007] over Gadanki.
 An important aspect that can clearly be noted from the SNR maps is the morphology of the maps, which is different from those observed as bottomside/bottom-type and plume structures observed in the equinox at Gadanki [Patra et al., 1997; Rao et al., 1997; Patra et al., 2005] as well as those extensively reported from the magnetic equatorial location Jicamarca [e.g., Woodman and La Hoz, 1976; Hysell and Burcham, 2002]. Moreover, the present observations appear like decaying irregularities despite the fact that their commencement took place in the postmidnight hours. On 3 August, they were observed as descending periodic striations.
3.1.2. Doppler Velocity and Spectral Width
 In order to present detailed variations of Doppler velocity as a function of height and time, we present in Figure 3 three examples each from July observations (7, 9, and 10 July) and August observations (2, 3, and 5 August). Positive (negative) velocities represent irregularity velocities upward-northward (downward-southward) along the radar beam. Velocities are both upward-northward and downward-southward and are well within ±100 m s−1. Also polarity of the velocity varies with both height and time. However, the velocities observed after 0300 LT, except for 9 July, are found to be predominantly downward-southward.
Figure 4 shows scatterplots of the mean Doppler velocities (corresponding to SNR > −12 dB) observed during 0000–0600 LT on all the nights (1–10 July and 30 July–8 August) in the altitude region of 250–450 km. The height region of 250–450 km maps to 350–550 km over the magnetic equator. The Doppler velocities are predominantly upward-northward prior to 0300 LT and downward-southward afterward. Figure 4 also indicates that the upward-northward velocities decrease in magnitude with time and become downward-southward after 0300 LT. For the postsunset F region irregularities observed from Gadanki, Patra et al.  observed similar decreasing trend in the Doppler velocities, but the Doppler velocities were larger than 100 m s−1 at times and were found to be near zero/downward-southward after ∼2300 LT. It is interesting to note that even though the current observations correspond to postmidnight hours, the velocities are upward-northward prior to 0300 LT, representing the presence of eastward electric field during those hours.
Figure 5 shows scatterplots of spectral widths (corresponding to SNR > −12 dB) as a function of time for different heights. Spectral widths are well within 50 m s−1 and are found to decrease to 20 m s−1 toward early morning hours. For the postsunset F region FAI observed from Gadanki, Patra et al.  reported spectral width values as high as 300 m s−1. They found that the spectral width values gradually decrease with time and become 20–50 m s−1 after 2300 LT. Thus the observed spectral widths reported here, in general, are remarkably lower than those of the postsunset FAI, but are similar to those observed after 2300 LT when the postsunset FAI are in the final stage of decay process. We may recall that SNR of the postmidnight FAI reported here are also similar to the postmidnight counterpart of the postsunset FAI. Thus the SNR and spectral width observations essentially represent a weak turbulent process associated with the postmidnight FAI presented here.
3.2. Ionosonde Observations From Gadanki
 During July–August 2008, F region echoes in the ionogram, in general, were found to disappear after 2200 LT and reappear after 0530 LT, except on a very few occasions of spread F events. In contrast, the occurrence of E layer echoes (hereafter referred to as sporadic E or Es echoes) was found to be much more than their F layer counterpart.
Figure 6 shows sample ionograms observed at Gadanki on the nights of 31 July and 6 August when F region FAI were detected by the Gadanki MST radar. In fact, only a few such ionograms were observed on these two nights. The ionograms are diffuse and weak and the maximum frequency of reflection/scattering is limited to 3 MHz. Contrary to these, on the night of 7 August, when radar echoes were weak, ionosonde spread F echoes were observed for a relatively longer duration. Figure 7 shows the type of ionograms observed on 7 August.
 Considering that the Es activities are pronounced during July and August over Gadanki [Venkateswara Rao et al., 2009], we have examined whether the detection of F layer trace was obstructed by the blanketing type Es layer. Figures 8a and 8b show the top frequency of Es echo trace (ftEs) as a proxy of the Es activity corresponding to the two campaigns, respectively. Since F layer trace was absent most of the time, it was not possible to scale the blanketing frequency to have any idea on the background electron density of the Es layers. Ionosonde observations are not available for 2, 5, 6, 7 and 10 July due to power supply problem as mentioned earlier. It may be noted that on several nights, ftEs values exceeded 3 MHz (the maximum frequency of F layer observed in the current data set). On several occasions, however, we have observed Es trace having ftEs values much smaller than 3 MHz or no Es trace at all. During those occasions also F layer trace was not observed in the ionogram. This suggests that the F layer electron density must be low enough for not being able to reflect ionosonde signals.
3.3. Ionosonde Observations From Trivandrum
 At Trivandrum also, ionosonde echoes from the F layer were found to disappear at about 2200 LT and reappear after 0600 LT except for some nights of spread F event. At Trivandrum, unlike at Gadanki, no E layer echo was observed in the ionograms.
Figure 9 shows examples of spread F echoes observed by the Trivandrum ionosonde. They are quite similar to those observed at Gadanki and can be classified as weak frequency spread F. It may also be mentioned that no range spread F was observed prior to the occurrence of frequency spread F, implying that they are not linked to postsunset range spread F. In contrast, on 7 August, when radar echoes were not observed at Gadanki, Trivandrum ionosonde detected range spread F during 2130–2300 LT and strong frequency spread F during 2300–0300 LT (not presented).
Figures 10a and 10b show virtual height of the bottom of the F layer (h'F) and the maximum frequency of HF wave that got reflected/scattered from the F layer (foF) observed during 5–10 July and 30 July–8 August, respectively. Observations made during 1–4 July have large data gaps and thus are not presented. The blank portions in Figures 10a and 10b are due to the invisibility of the F layer trace. Duration of occurrence of spread F is also indicated. At Trivandrum, spread F was observed only on 7 nights, i.e., 5 July, 6 July, 9 July, 10 July, 31 July, 6 August, and 7 August. Out of these, range spread F was observed only on 10 July and 7 August prior to 2300 LT. No clear height rise prior to the spread F occurrence has been observed. It may also be noted that the values of the foF just before the disappearance of F layer trace is found to be ∼2 MHz, implying that the peak electron density at that time is close to 5 × 104 electron/cm3.
 Important aspects of the F region FAI observed in the summer of low solar condition by the Gadanki radar are (1) their occurrence during the postmidnight hours on all the 20 nights, (2) height extent of the FAI as high as 500 km, (3) morphology of the radar maps very different from their equinoctial counterpart, and (4) Doppler velocities predominantly upward-northward prior to 0300 LT. Ionosonde observations made both at Gadanki and Trivandrum that make the above observations intriguing are (1) rare occurrence of ionosonde spread F echoes and (2) F layer peak electron density of ∼5 × 104 electron/cm3 or less as characterized by the absence of ionogram trace during 2200–0530 LT.
4.1. Comparison With Earlier Observations
 Radar observations of F region FAI have clearly revealed that the local time duration of their occurrence during the summer of low solar condition is in excellent agreement with that of ionosonde frequency spread F reported earlier [Chandra and Rastogi, 1970, 1972; Subbarao and Krishnamurthy, 1994]. Simultaneous ionosonde observations, however, have revealed that during the low solar condition of 2008 (10.7 cm solar flux = 61), most part of the night (i.e., 2200–0530 LT), except for some occasions associated with frequency spread F event, the ionograms were devoid of F region echo. It was also observed that foF decreased to ∼2 MHz at 2200 LT. Since the F layer peak electron density was 5 × 104 electron/cm3 (corresponding to 2 MHz) or lower, nondetection of ionosonde echoes can be attributed to the limited sensitivity of the ionosonde to detect reflected echoes at lower frequencies and weak spread F echoes in general. Despite, the ionosonde observations from both the locations revealed the occurrence of frequency spread F echoes during the postmidnight hours, which seem to broadly agree with those reported earlier from the Indian sector [Chandra and Rastogi, 1970, 1972; Subbarao and Krishnamurthy, 1994].
Chandra and Rastogi  reported disappearance of ionosonde echoes during 0300–0500 LT. It may be mentioned that their observations were made during 1965 when 10.7 cm solar flux was 69 and the foF values during the midnight were 4–5 MHz. Observations reported by Subbarao and Krishnamurthy  were made in 1976 when 10.7 cm solar flux was 66 and the foF were 2–3 MHz. Thus, it is reasonable to consider that the disappearance of F layer ionogram trace for longer duration in 2008 than in 1965/1976 is related to lower solar flux condition in 2008 than in 1965/1976.
 Further, the ionosonde frequency spread F echoes reported here are weaker than those reported by Chandra and Rastogi . Notably, the radar observations of postmidnight F region FAI reported here also show low echo SNR (−15 to 5 dB). The observed correlation between weak radar echoes and weak ionosonde frequency spread F echoes is consistent with that reported by Rastogi and Woodman  based on Jicamarca radar and Huancayo ionosonde observations.
 Coming to the morphology of the radar maps, they are remarkably different from those of the equinoctial postmidnight observations reported earlier from Gadanki [e.g., Patra et al., 2005; Sekar et al., 2007]. Morphology of some of the events, however, is somewhat similar to some events reported earlier by Rao et al.  using Gadanki radar observations, and Fukao et al.  and Saito et al.  using the Equatorial Atmosphere Radar (EAR) located at Kototabang (0.2°S, 100.32°E, dip latitude 10.36°S), Indonesia. Hysell and Burcham  also reported Jicamarca radar observations of postmidnight FAI with distinctly different morphology. Hysell and Burcham  found that such structures occur following periods of geomagnetic activity and are most common in solar minimum. The aforementioned observations, however, are either the extension of the postsunset F region FAI into the postmidnight or related to magnetic activity. In contrast, the present observations relate to mainly postmidnight hours and magnetically quiet period. It may be stressed that not even on one out of twenty nights plume was observed in the premidnight hours.
 In this context, midlatitude F region FAI structures reported by Fukao et al.  may be worth comparing since their observations correspond to postmidnight hours and the summer of low solar condition. They observed ascending and then descending structures in which the velocities were found to be northward-upward and downward-southward, respectively. The present observations seem to be more of the type reported by Fukao et al. . It is, however, intriguing that such features are being observed from Gadanki, which is close to the magnetic equator. In the following, we discuss these aspects in details.
4.2. Generation Mechanism
 Before we discuss on the generation mechanism of the F region irregularities in the postmidnight hours, it is important to address whether the summertime postmidnight irregularities manifesting weak radar echoes were generated during those hours or the overhead passage of the postsunset irregularities in their decay phase. The radar observations of FAI suggest that postsunset irregularities of the type observed in equinox have not been observed even on one out of twenty nights. Also, the ionosonde frequency spread F echoes from both Trivandrum and Gadanki, excluding the two nights of Trivandrum observations (10 July and 7 August), appeared late in the night. It may also be mentioned that F region FAI observed during the summer of 2007 from Kototabang, using a 30.8 MHz radar, were found to occur during postmidnight (Y. Otsuka, personal communication, 2009). Thus it is reasonable to say that these irregularities are not generated in the postsunset hours. However, there may be a possibility that the postmidnight occurrence probability is due to the equator ward movement of the irregularities generated at higher latitude in the premidnight hours.
 Coming to the generation mechanism, Subbarao and Krishnamurthy  invoked the generalized collisional RT instability to account for the postmidnight occurrence of ionosonde spread F echoes during the summer of low solar condition. Their proposal was basically based on the observed height rise of the F layer prior to the occurrence of spread F echoes. The radar observations presented here, which showed FAI velocities predominantly upward-northward during 0000–0300 LT, representing the presence of eastward electric field, and height extent of the postmidnight FAI to 500 km at times, are indicative of the role of convective process viz., the RT instability. Thus, while the RT instability appears to be a plausible candidate to account for some of the features, there remain a few aspects that need to be accounted for. First, the origin of the eastward electric field during the postmidnight and magnetically quiet condition, required for the uplifting of the F layer and thus the growth of the RT instability, needs to be addressed. Also, the morphology of the radar observations, not of the type of radar plumes, is not quite compatible in terms of RT instability [Woodman and La Hoz, 1976].
 In regard to the existence of eastward electric field during the midnight hours, Nair et al. , using HF radar observations from the magnetic equatorial station Trivandrum, showed that the downward plasma drift during nighttime (due to westward electric field) is the lowest (5 m s−1 or less) during the summer and low solar flux. Similar features have also been reported from Jicamarca [e.g., Scherliess and Fejer, 1999]. Since the westward electric field is small during the summer of low solar condition, at times it could possibly turn eastward providing condition for the RT instability to grow.
 As far as morphology of the radar maps are concerned, Fukao et al. , using the EAR, observed low-latitude F region FAI structures, not of the type of classical plume, and found that they are similar to those of middle and upper atmosphere (MU) radar observations of midlatitude F region FAI reported earlier by Fukao et al. . Notably, observations made by Fukao et al.  were made during the summer of low solar activity and were interpreted by Kelley and Fukao  in terms of Perkins instability as the primary instability generating large-scale structures and the E × B instability/wind driven instability as secondary instability generating short-scale structures. Kelley and Fukao , however, found that the propagation direction was not consistent with that required for the Perkins instability. They speculated the possible role of Es layers.
Haldoupis et al.  and Kelley et al. , based on midlatitude observations of ionosonde spread F, airglow observations of traveling ionospheric disturbance (TID), and Es radar backscatter observations, proposed that enhanced polarization electric field from the E region when map along the magnetic field lines to the F region can create rising and falling regions in the bottomside F region, which eventually help forming spread F echoes in ionogram. More recently, Otsuka et al.  reported midlatitude radar observations of F region irregularities having close link to TID and strong Es backscatters. Based on the propagation direction of the structures, which were not consistent with the Perkins instability, they interpreted their observations in terms of the gradient drift instability, wherein the driving electric field is of E region origin. Yokoyama et al. , through numerical simulation study of coupled Perkins and Es layer instabilities, however, showed that the propagation of the F region irregularities can be controlled by E region neutral wind.
 In this context, it may be mentioned that summertime F region FAI observed using the multibeam capability of the 30.8 MHz radar from Kototabang revealed westward motion of the irregularities (Y. Otsuka, personal communication, 2009), which is opposite to the commonly observed eastward motion during nighttime. Although it is not known what causes the westward motion of the irregularities, the role of Es layer is a good possibility. As far as the low-latitude Es activity is concerned, Venkateswara Rao et al.  showed that at low latitude both Es activity and the occurrence of quasiperiodic radar echoes peak in summer and just prior to midnight. Thus it is quite possible that the low-latitude Es activity provides necessary electric fields for the F region to be unstable and the eastward electric field required for the growth of the RT instability is of low-latitude Es origin. It appears that both gradient drift instability [Otsuka et al., 2009] and RT instability [Subbarao and Krishnamurthy, 1994] may be required to account for the postmidnight F region irregularities. Zonal motion of the E and F region irregularities and airglow measurements similar to that done from midlatitudes may provide new insight in understanding the observations reported here.
5. Concluding Remarks
 Gadanki radar observations of F region FAI during the summer of low solar condition have been shown to be very different from their equinoctial counterpart. It appears that they could be due to the manifestation of gradient drift and RT instabilities with Es layer providing important free energy. Although such coupling processes have been shown to be operational at midlatitudes and are being intensely studied for a deeper understanding, the low-latitude observations provide added features in understanding the E-F region coupling processes in detail. Zonal motion of the irregularities and airglow measurements would provide additional new insight into the low-latitude processes, which we would like to carry out in future investigation. Further, observations from Kototabang in Indonesia and Piura in Peru will be of immense value in furthering this research.
 The radar observations reported here were made under the scientific program “Study of Atmospheric Forcing and Responses” (SAFAR). The authors wholeheartedly appreciate the efforts made by the MST radar technical team for making the observations reported here. D.V.P. was supported by the “Development of Korean Space Weather Centre” of KASI and KASI basic research funds. The authors appreciate both of the reviewers for their very useful comments.
 Amitava Bhattacharjee thanks Narayan Chapagain and Tatsuhiro Yokoyama for their assistance in evaluating this paper.