In the present communication, allSKy interferometric METeor (SKiYMET) radar observations of gravity wave activity in the mesosphere lower thermosphere (MLT) region over Thumba (8.5°N, 77°E) are presented. The present meteor radar system provides hourly zonal and meridional winds in the MLT region, which can be readily used for studying the tides, planetary waves, gravity waves of periods 2–6 hours, and other long period oscillations in this region. However, these hourly winds are not sufficient for studying short period gravity waves having periods less than an hour, which demand high temporal resolution measurements. Even though the winds are estimated on an hourly basis, information such as zenith angle, azimuth angle, and radial velocity of each detected meteor are archived. Using these details of the meteor, an algorithm is developed to obtain the 15-min temporal resolution wind data. The output of the algorithm is compared with hourly wind data, and it showed a good agreement during the high meteor shower periods. Most of the times high meteor counts are observed during late night and early morning hours (local) over this latitude. Continuous wind measurements during the high meteor shower periods are used for studying the gravity wave activity in the MLT region. As the wave activity is intermittent and nonstationary, wavelet analysis has been used for delineating the wave features. The results showed the upward propagating intermittent gravity waves with periods 1–2 and 4–5 hours. The new aspect of the present communication is the usage of meteor radar for gravity wave studies for the first time over this latitude and studying their seasonal variability.
 Owing to the importance of vertical coupling of the different regions of the Earth's atmosphere, various experimental investigations have been carried out across the globe. Most of these studies identified atmospheric waves as one of the most important mechanisms through which the vertical coupling takes place. These propagating atmospheric waves will be best registered at middle atmospheric heights due to the large wave amplitudes there. In the case of equatorial middle atmosphere, Kelvin, Rossby, planetary waves and gravity waves having periodicities ranging from few days to few minutes are known to play a crucial role in this region. It will not be an exaggeration to state that the dynamics of the equatorial middle atmosphere is dictated by these waves. Understanding and knowledge of these waves and their propagation characteristics in the MLT region is of particular importance to the middle atmospheric community.
 Among all the above-mentioned waves in the equatorial MLT region, the studies on gravity waves are very limited. The limited gravity wave observations in the equatorial MLT region can be directly attributed to the lack of high temporal measurements of winds and temperature in this region. Gravity waves generated in the troposphere propagate vertically and horizontally, dissipate, interact nonlinearly, and profoundly influence the flows of momentum, energy, and constituents on a global basis. However, we are not going to elaborate on the gravity waves; instead we refer to Fritts and Alexander [2003, and references therein]. This seminal review on gravity waves includes description of short period convectively generated gravity waves to long period inertial gravity waves and their propagation characteristics and properties.
 In the contest of MLT region dynamics, it is proven in the past that the gravity waves have profound effect [Vincent and Reid, 1983; Meek et al., 1985a, 1985b; Vincent and Fritts, 1987; Fritts and Yuan, 1989; Fritts et al., 1992; Nakamura et al., 1993; Gavrilov et al., 1995]. Modeling studies also have revealed that the vertically transferred horizontal momentum by gravity waves significantly influences the circulation of the MLT region, decelerating the atmospheric zonal flow away from radiative-equilibrium values, closing the mesospheric jets and inducing a meridional circulation which ultimately cools the summer mesopause to extremely low temperatures [Houghton, 1978; Lindzen, 1981; Holton, 1983; Matsuno, 1982; Dunkerton, 1982; Miyahara, 1984; Holton and Zhu, 1984; Garcia, 1989; McIntyre, 1989] and also warms the winter higher latitude mesopause. The turbulence arising from gravity-wave breaking processes is also important in the transport of heat and constituents. The temporal and geographic variations in gravity-wave fluxes thus contribute to the variability of large-scale properties of the atmosphere. Attempts to understand and model the middle atmosphere therefore require detailed information about properties of the gravity-wave fields. As the quantitative understanding of the MLT region demands the better understanding of gravity waves, numerous experimental and theoretical efforts have been made to divulge these wave characteristics and their interaction with other physical processes in the MLT region. In the earlier efforts, the height profiles of gravity wave amplitudes and phase were treated extensively and in the later phase the possible tropospheric forcing mechanisms for these waves. From the last decade, the interest is shifted towards the gravity wave variability at different scales. Variability of the gravity wave activity on shorter than seasonal time scales has been observed for some time. Several recent studies have been devoted to the variability of the mean winds and wave motions in the lower and middle atmosphere [Nakamura et al., 1993, 1996; Burrage et al., 1996; Garcia et al., 1997; Manson et al., 1999; Vincent and Alexander, 2000]. A number of theoretical studies [Alexander and Holton, 1997; Sassi and Garcia, 1997; Ray et al., 1998; Garcia and Sassi, 1999; Medvedev and Klaasen, 2001] showed that an important contribution to MLT dynamics may accompany various wave motions generating convective and dynamical processes in tropics.
 All the above discussion suggests that a better knowledge of the climatology of gravity wave activity in the MLT region is necessary for any realistic modeling of this region. The global atmospheric communities is now interested in vertical coupling of the different regions of Earth's atmosphere and as discussed above gravity waves play a crucial role in this coupling process (e.g., Climate and Weather of Sun Earth System (CAWSES)). However, the climatology of gravity waves needs observations across the globe as the gravity wave activity is expected to be a highly local phenomenon, strongly controlled by the geographical distribution of sources, which show variations from one geographical location to the other. Even though the satellites are best suited for global coverage, the temporal resolution of their measurements renders them not suitable for gravity wave studies. On the other hand, long-term coordinated observations from MF, MST and meteor radars across the globe can provide a global climatology of gravity waves. However, most of the gravity wave observations in the past are by using either MF or MST radars. So far, meteor radar observations of gravity waves are very limited. This can be attributed to hourly wind measurements provided by meteor radars, which are not sufficient for gravity wave studies. Because of this limitation, earlier studies from the present observational site were limited to diurnal tides and their seasonal variability and other long period oscillations in the MLT region [Reddi and Ramkumar, 1997]. Now, it is realized that it will be very useful to have high temporal resolution wind measurements, which can be used for gravity wave studies. Proceeding on similar lines, in the present study an attempt is made to derive the high temporal wind measurements during meteor shower periods from the meteor radar observations. The central objective of the present study is to demonstrate the capability of meteor radar to investigate the gravity wave activity in the height region of 82–98 km for the first time over this latitude. The meteor radar observations during 2004–2005 are used to bring out some salient features of the gravity waves over the equatorial MLT region.
2. Data Analysis
 The details of SKiYMET meteor radar used for the present study are given in Kumar et al. (this issue). Here, we briefly discuss the estimation of some key parameters from the meteor radar observations. A detailed description of parameter estimation from SKiYMET meteor radars can be found in Hocking et al. . In the present radar observations, the complex correlation method is adopted to measure the phase differences between the receiving antennas in order to determine the echo arrival angle, and both the complex auto and cross correlation methods are used to measure the rate of change of the relative phase in order to determine the radial velocity of the meteor trail. The accuracy of the arrival angle determination is essentially depending upon the error in the phase determination. For each meteor, the arrival angle of the reflection point is calculated from the mean phase difference between the receivers, averaged over possible independent estimations using the combinations of receiving antennas. After collecting a large number meteor information, these data is then used to determine zonal and meridional winds. Usually this is done by clustering the radial velocities into one-hour bins, and 3 km height bins. Then a least squares fitting procedure will be applied to determine mean wind components for this height-time bin.
 If θ and ψ are zenith and azimuth angles, respectively, then the radial velocity Vr can be projected on a three-dimenisonal plane as follows:
where U, V and W are the zonal, meridional and vertical components of the wind field, respectively. Three scattering points are sufficient to determine the three unknown components of the motion. However, to improve the accuracy of the measurements, all the meteors observed in that particular hour will be used to estimate the hourly winds. For any set of measurements of zenith angle, azimuth angle and radial velocity the simultaneous equations for U, V and W can be solved to find the best estimates of winds. The accuracy of wind measurements is directly proportional to the number of meteors observed during that hour. As the meteor count has diurnal variation, the accuracy of wind measurements also will have diurnal variation. Figure 1 shows a typical diurnal variation of meteor counts observed over present observational site. It is evident from this figure that the large number of meteors occurs during 00 to 04 hours and 18 to 23 hours UT (LT = UT + 5:30 hours). This figure also confirms that there are more than sufficient meteors during particular hours in a given day. This aspect has been used to estimate the high temporal resolution wind data during the high meteor rates, which can be readily used to divulge the gravity waves in the MLT region. It has been observed that during the high meteor rates, there are sufficient meteors to estimate winds every 15 min. The diurnal variation of meteor counts shows that if we have two consecutive days observations then it is possible to have 10 hours of high meteor counts starting from 18 hours to next day 04 hours (UT). Now, coming to the temporal resolution, it has been observed that one can have meaningful wind estimates with 15 min time resolution during high meteor rates. The time resolution depends on number of meteors available during that interval. On some occasions during meteor shower periods it is observed that the winds can be estimated with 10 min time resolution. However, we estimate zonal and meridional winds every 15 min for the present study.
 For the present study, we have chosen consecutive days with high meteor rates. The meteor information during the 18 hours to next day 04 hours (UT) is used to estimate winds for every 15 min and 3 km height intervals. Thus, time series of zonal and meridional wind for 10-hour duration and 15 min time resolution is used for studying the gravity waves in the MLT region. Wavelet analysis has been used to find the time periods of gravity waves. It has been shown in the past that the gravity wave activity in the MLT region is nonstationary, i.e., the characteristics of the wave changes with time and also localized in time. To study these types of waves, wavelet analysis is very useful over conventional Fourier analysis. The wavelet analysis can provide the time history of gravity waves. Morlet wavelet has been chosen as mother wavelet for the analysis as it can be used a general-purpose wavelet. The Morlet wavelet consists of a complex exponential modulated by a Gaussian , where t is time, s is the wavelet scale, and ω is a nondimensional frequency. For ω = 6, which is used for the present study, there are approximately three oscillations within the Gaussian envelope. The wavelet scale s is almost identical to the corresponding Fourier period of the complex exponential. The wavelet power spectrum is defined as the absolute value squared of the wavelet transform and gives a measure of the time series variance at each scale (period) and at each time.
3. Results and Discussion
 An algorithm is developed to estimate the 15-min time resolution zonal and meridional winds using above discussed procedure and the output of the algorithm is compared with the hourly estimates. This comparison is shown in Figure 2. This figure corresponds to 90 km altitude on 20 January 2005. From this figure it is evident that during the high meteor rates the 15 min resolution data obtained by the present algorithm agrees with hourly estimates. However, Figure 2 shows that the 15 min wind values are smaller than the hourly values during 20–23 hours. This could be due to unsymmetrical distribution of meteors about zenith during smaller time samples. However, this aspect should be investigated further. The small fluctuations in 15 min resolution winds are believed to be due to gravity wave activity. Thus, this plot confirms the consistency of the algorithm and also demonstrates the potential of this data for studying the gravity wave activity. To further confirm the observed fluctuations are due to wave motion, a hodograph is plotted using the zonal and meridional winds and is shown in Figure 3. The clockwise rotation of winds with height and ellipsoid shape of hodograph confirm beyond doubt that the observed fluctuations are due gravity wave motions. However, it is to be noted that 8-h tide's hodograph is also elliptical (it will be if the components are not 90 degrees and the zonal and meridional amplitudes are not equal). The 8-h oscillations are also quite irregular in occurrence. So, there is a possibility that this is an 8-h tidal oscillation. To obtain the zonal and meridional wind fluctuations, we have used the hourly wind estimates as background wind and subtracted the same from 15 min data. Figure 4 (top) shows the zonal and meridional wind fluctuations observed on 30 June 2005. This plot readily reveals oscillatory nature of the wind fluctuations. The variance estimated from these fluctuations ranges from 0–150 m2 s−2, which are consistent with the earlier studies reported by Manson et al.  and Eckermann et al. .
Manson et al.  reported the gravity wave spectra and direction statistics using MF radar observations in the two height layers (61–73, 76–88 km) and in two time period bands (10–100 min and 1.5–6 h). In the present study, to know the typical time periods present in the zonal and meridional fluctuations shown in Figure 4 (top), wavelet analysis has been carried out and is shown in Figure 4 (bottom). The x-axis of this figure shows the time and y-axis shows the time period of the observed gravity waves and intensity represents the gravity wave amplitude in m s−1. These wavelet spectra clearly show the presence of gravity waves with time period 4.5 hours, both in zonal and meridional winds. As these spectra give the time history of gravity waves one can visualize the evaluation of the wave amplitudes. In this particular case the 4.5-h wave is present throughout the observations. One of the interesting observations from the present wavelet spectra is the presence of 1–2 h gravity waves in the meridional winds. This particular wave is localized in time and also present in meridional winds only. One more general observation from these spectra is the wave amplitudes in zonal winds are twice the magnitude of the meridional winds. Thus, the present results show that the meteor radar observations during shower periods can be used to study the gravity wave activity. To confirm the observed gravity waves are propagating vertically, the phase profile of 4.5-h gravity wave is constructed and is shown in Figure 5. This phase profile shows the downward propagation of phase in the height region 88–98 km, which implies that the wave is propagating upwards. All this exercise is carried out to show the applicability of meteor radar observations to study the gravity wave activity over this latitude, which is so far not carried out due to the lack of high temporal resolution wind information.
 Having done this, our immediate objective was to establish the seasonal variation of gravity wave activity for this latitude. Even though meteor showers occur only during particular seasons, we have observed that high numbers of meteors are available at least 5–6 days in a month. These days are used to generate the time sequence of zonal and meridional wind with 15 min time resolution for an interval of 10 hours. Again, this data is used to generate the zonal and meridional wind fluctuations. Gravity wave activity is determined from the variance of the wind data. Although spectral frequency resolution would add useful information, a benefit of the variance approach is that the results are not contaminated by the Doppler effect. Here it is emphasized that the variance observed in the winds are attributed to the presence of gravity waves. The variance data thus obtained is averaged for each day to study the monthly variation of gravity wave activity. As mentioned earlier only 5–6 days of data is used to obtain the monthly means. Figure 6 (top and bottom) shows the monthly means of zonal and meridional wind variance averaged in 82–90 km height region. The striking feature of this figure is the pronounced semiannual oscillations in the gravity wave activity showing peaks near equinoxes. The semiannual oscillation in the background winds over equatorial MLT region is well established. However, as mentioned in the introduction the gravity wave studies over these latitudes are limited in the MLT region and thus present results are of interest.
 Gravity wave studies in the mid and high latitudes have also shown the semiannual oscillation in the gravity wave activity using MF radar observations [Manson et al., 1997, 1999; Tsuda et al., 1994]. Tsuda et al.  carried out a comparative study of gravity wave activity in the mesosphere over Shigaraki (35°N, 136°E), Adelaide (35°S, 139°E) and Saskatoon (52°N, 107°W). The authors have shown the annual variation in the lower atmosphere and semiannual variation in the mesospheric gravity wave activity with large peak in summer and minor enhancement in winter. This particular observation was consistent at all other latitudes observed by the authors i.e., at 35°N, 44°N, 52°N and 35°S. Again, none of these latitudes include the low latitudes and equatorial region. However, Manson et al.  estimated gravity wave variances using global MLT-MF radar network spread across 0–70°N. The authors show a max at low latitude near months 4 and 11, quite similar to the present study. In a seminal work reported by Hirota , it has been emphatically shown that the gravity wave activity in the low latitude mesosphere shows the semiannual oscillation with peaks near equinoxes using rocket data up to 65 km. The seasonal variation of gravity wave activity can be directly attributed to seasonal variation of their sources. In midlatitudes, the primary source for excitation of gravity waves is jet stream and the secondary source is convection associated with meteorological disturbances [Tsuda et al., 1994]. In equatorial and low latitudes, excitation of gravity waves is very active primarily due to the intense solar radiation and when insolation is high during equinoxes the activity peaks as shown in the present study. One more important mechanism, which gives equinoctial maxima, is filtering by background winds. The zonal winds, which show SAO at these MLT heights, will provide increased gravity wave variances in the equinoxes due to filtering and exclusion in other months. However, the data used for the present study is only for one-year duration and data for some more years are required to verify the consistency in the seasonal variation of gravity wave activity in the MLT region. The seasonal variation discussed here, however, is compared with earlier studies from other geographical locations, which showed the consistency of the observed seasonal variation in the gravity wave activity over this latitude.
 Newly installed meteor radar at Thumba has been used to study the gravity wave activity in the low latitude MLT region. An algorithm has been developed to derive the high temporal resolution wind data from meteor observations during the shower periods. The radar derived zenith and azimuth angles along with radial velocity are used to derive the zonal and meridional winds with 15 min time resolution. This high temporal resolution data is used to bring out some salient future of gravity wave activity over this latitude. The hodograph analysis, wavelet analysis and phase profiles have shown the presence of gravity waves in the output of the algorithm. The magnitude of gravity wave variances is comparable with the earlier studies. Thus, the present analysis has shown the potential of meteor radar to study the gravity wave activity in the MLT region. Making use of high meteor rate observations in every month, the seasonal variation of gravity wave activity is studied. The present observations have shown the semiannual oscillation with maxima near the equinoctial months. Even thought the seasonal variation is obtained using one year of observations, the results are consistent with the results reported by the earlier studies. Subsequent studies at this latitude will be focusing on gravity wave interactions with tides and planetary waves.
 Authors are very much thankful to W. K. Hocking and Brian Fuller for successfully installing the meteor radar at Thumba. One of the authors (Maria Antonita) is thankful to the Indian Space Research Organization for providing the fellowships to carry out the research work reported here.