Long-term mean vertical velocities observed by the Mesosphere-Stratosphere-Troposphere (MST) radar over a tropical station, Gadanki, India (13.5°N, 79.2°E), are presented for the first time in this paper. Profiles of mean vertical velocities show wave like structure with a vertical wavelength of ∼6 km, and the observed values range from 3 to 20 cm s−1 on average. During monsoon and postmonsoon seasons, larger magnitudes are observed up to 20 cm s−1. From the present study, an interesting feature, reversal in vertical velocities from downward to upward between 5 and 10 km during the monsoon season, is observed. This upward motion in long term averaged vertical velocity is thought to be due to reversal in wind direction around this height range and also due to horizontal velocity convergence, which is frequently observed over this region. One more reversal is observed near the Tropical Easterly Jet (TEJ) (around 16 km), which is from downward to upward and always showing minimum vertical velocities close to zero. This reversal can be attributed to instabilities associated with the jet streams. Large negative values are observed in the lower troposphere below 6 km, which are attributed to large vertical wind variances observed in this region indicating gravity wave activity.
 The vertical wind measurement with a high resolution contains a great deal of information on gravity-wave motions. The long-term mean value of the vertical wind velocity is very useful in studying the large-scale circulation pattern [Fritts, 1984] and wave transport phenomenon [Fritts, 1989, and references therein]. Doppler radars operating at UHF and VHF wavelengths have the unique capability to measure vertical velocity of clear air. Nastrom and Gage  studied climatology of vertical wind variability in the troposphere and stratosphere using the Poker Flat MST radar, Alaska. The vertical velocities in the troposphere are expected to be small in large-scale systems and in long period time averages [Nastrom and Van Zandt, 1994]. Several studies were made over a number of radar sites on vertical velocities. Using MU radar (34.85°N, 136.10°E), Fukao et al.  observed the transition from downward motion in the troposphere to upward motion in the lower stratosphere near the peak in the zonal wind profile during winter months. Yoe and Ruster  also found reversal in vertical velocities near the height of maximum horizontal wind using SOUSY VHF radar. Similar patterns of downward motion in the troposphere and upward motion in the above were seen by Nastrom and Van Zandt  using Flatland radar (40.5°N, 88.5°W) and are thought to be due to vertically propagating gravity waves. In the tropical latitudes, vertical velocities show always upward motions throughout the troposphere during precipitation and downward motion at all the heights during periods of nonprecipitating systems [Balsley et al., 1988; Gage et al., 1991]. Large vertical velocities of the order of 10 ms−1 were observed during convective precipitation [Rao et al., 1999].
 The aim of this paper is to study mean vertical motions over Gadanki and to examine some of the possible theories to explain the observed results. This paper is organized as follows. In section 2, system description and data used for the present study are given. Results and discussions are presented in section 3, which covers mean vertical velocities (section 3.1), monthly variations (section 3.2), seasonal variations (section 3.3) and annual variations (section 3.4). Mean vertical velocities during the jet stream conditions are given in section 4. Finally, summary and conclusions drawn form the present study are given in section 5.
2. System Description and Database
 Indian MST radar is a high power coherent pulsed Doppler radar operating at 53 MHz and is located at Gadanki, India (13.5°N, 79.2°E), a tropical station. The antenna consists of 1024 crossed three element Yagi antennas occupying an area of 130 m × 130 m. A total transmitted power of 2.5 MW (peak) is provided by 32 transmitters. It generates a radiation pattern with a main beam of 3°, gain of 36 dB and a side lobe level of −20 dB. Its average power-aperture product is 7 × 108 Wm2. The main beam can be operated in total 82 beam directions, two in vertical (in east-west and north-south polarizations) and ±20 off zenith (in east, west, north and south directions). It makes use of the Doppler Beam Swinging (DBS) technique for measuring wind field. The DBS method for measuring the three components of the wind vector requires spectral measurements at a minimum of three noncoplanar beam positions. Usually, radar is operated using six beams (two in vertical and four in off-vertical) to determine the wind components in a least square sense as described by Sato . Complete details of system description can be had from Rao et al.  and Kishore .
 The Indian MST radar is operated in two modes. In one mode, the radar is operated daily for about half-an-hour around 1700 LT (12 GMT). This time is chosen as it coincides with the radiosonde launch time at Chennai, located at a radial distance of 120 km from the radar site. Most of the times data exceeds for about 20 days in a month. In the second mode, the radar is operated for about 15 minutes in each hour in a 24-hour cycle (diurnal cycle) and every month two diurnal cycles are used. Three years of Indian MST radar data, starting from November 1995 to June 1998, are used to study the seasonal and monthly variation of mean vertical velocities observed over a tropical station, Gadanki, India (13.5°N, 79.2°E). Data are not available during May 1996, December 1996, January 1997, February 1997 and March 1997 due to some technical problems and maintenance of the radar. First the observed vertical velocities on each day are averaged to remove small-scale fluctuations and treated that averaged profile as representative of that day. Similarly, daily averaged profiles are then averaged for a month and finally for a season in order to study monthly and seasonal mean vertical velocities. Besides these observations, radar data collected in campaign (alert) mode during special events like convection and precipitation are also used while comparing the clear air velocities in the present study.
3. Results and Discussion
3.1. Mean Vertical Velocities
Figure 1 shows typical profiles of mean vertical velocities observed during clear time (around 1630 LT) and precipitation time (around 1830 LT) on 6 June 1996 averaged for about half an hour each. From the figure it is clear that under clear conditions (solid circle), subsiding motion below 11 km in troposphere and ascending motion in the upper troposphere are observed with magnitudes reaching up to 0.5 ms−1. During convective precipitation [Rao et al., 1999] upward motion (open circle) with magnitudes upto 10 ms−1 are noticed in most of the tropospheric heights between 4 and 15 km. These observations are similar to those reported earlier [Balsely et al., 1988; Gage et al., 1991] but those are for long period averages.
 In order to see whether there is any bias in the observed vertical velocities when we consider vertical beam alone with that obtained using all the beams, vertical velocities are first estimated from the radial velocities from vertical beam position alone and are compared with that estimated using all the beams (using off-vertical beams) and are shown in Figure 2. This is an averaged figure drawn using the data from Indian MST radar for about three years. The error bars presented in this figure are the standard deviations observed, when the data collected using vertical beam of the radar are averaged over the period of approximately three years. From the figure it can be seen that more or less there is a close agreement between the two (mostly in lower heights), suggesting that there is no bias in vertical velocity due to tilted refractivity surfaces [Nastrom and Van Zandt, 1994].
3.2. Monthly Variations
Figure 3 shows monthly mean vertical velocities observed over Gadanki for the entire period during November 1995 through June 1998 along with error bars. The error bars in the figure represent the standard deviations observed when the data is averaged over each (similar) month during the observational period. From the figure it can be seen that vertical motions are showing wave like structure with constant upward and downward motions with wavelength ∼5 to 8 km thickness except during winter months (December, January and February). The magnitudes observed in the lower troposphere when compared to that of upper troposphere are very large. Earlier literature [Balsley et al., 1988; Huaman and Balsley, 1996] suggests that the observed vertical velocities will be less in the tropical latitudes when compared to mid and polar latitudes. But here interestingly large magnitudes are observed even in the monthly (even in long term average and will be shown in later sections) averages particularly in the lower troposphere. These observed large vertical velocities could be attributed to strong monsoon circulations that prevail half of the year over the site.
3.3. Seasonal Variations
 Seasonal variation of mean vertical velocities for the entire period is shown in Figure 4. From the figure it can be seen that during winter, the magnitudes of mean vertical velocities are very small with completely subsiding motion below 12 km and ascending motion above it. The mean vertical velocities in this season are around 1 cm s−1 thus showing similar values in clear air conditions with that observed at various other tropical radar sites [Balsley et al., 1988; Gage et al., 1991; Huaman and Balsley, 1996]. In summer season, upward motion with large magnitudes is observed in middle and upper troposphere when compared to other seasons. These large magnitudes may be attributable to convection that generally prevails in summer season over the radar site. During postmonsoon and monsoon seasons, downward motion occurs below 16 km except in a small region between 5–9 km where an upward motion is observed. This feature is consistent throughout the monsoon period. This is attributed to reversal in wind direction around this height range observed as shown in Figure 5 and perhaps may also be due to horizontal velocity convergence [Asnani, 1993]. Figure 5 shows horizontal and vertical winds along with wind direction in two different months of January (Figure 5a) and July (Figure 5b) averaged for three years. From the figure it is clear that in the month of July (and through out monsoon), it is always found that the wind direction suddenly changes around 7 km and it is always found that the observed vertical velocities are upward. During January (and the rest of the seasons) there was no sudden change in wind direction and hence no upward motion. The reversal in wind direction observed over the radar site is due to monsoon circulation, which is in conformity with winds observed over India during monsoon season [Koteswaram, 1958].
3.4. Annual Variations
 Annual variations of mean vertical velocities along with the variance during the year November 1995–October 1996, November 1996–October 1997, November 1997–June 1998 and mean for three years are shown in Figures 6a, 6b, 6c, and 6d, respectively. From the figure it is observed that in the lower troposphere, large downward velocities ∼6 cms−1 are observed during clear air conditions. From the figure it can also be seen that, even in long period averages, a clear wave like structure can be seen with constant upward and downward motions with ∼6 km wave length. From the variance plot it can be seen that in the lower troposphere below 7 km as the variance increases, the magnitudes of downward motion increase showing gravity wave effects on vertical wind bias [Nastrom and Van Zandt, 1994]. This suggests that the gravity wave activity is large at this latitude and hence large magnitude in vertical velocities and its variance. Several investigators observed gravity wave activity in the lower troposphere over the MST radar site [Nagpal et al., 1994; Revathy et al., 1994]. This phenomenon is quite different to earlier studies [Huaman and Balsley, 1996] in the tropical regions in which they observed mean vertical velocities and its variance are both much smaller. Instabilities driven by wind shear might be the possible generation mechanism of gravity waves [Nastrom et al., 1990; Eckermann, 1995].
Figure 7 shows mean vertical velocities as a function of hourly vertical velocity variances (σw2) observed from the diurnal cycles during the entire period from November 1995 through June 1998. The mean and variances for the vertical velocities are computed each hour at each height and finally the means are sorted into groups depending on variance, and the mean of vertical velocities from each group of variance is plotted. The observed variances range between 0.01 and 0.13 m2 s−2 and above these values it is found that there were no consistent variances. From the figure it is clear that greater the hourly averaged mean variance, more negative the observed mean vertical winds in the lower troposphere. Above 10 km, there is no apparent correlation between the observed variances and mean vertical velocities.
4. Mean Vertical Velocities During Tropical Easterly Jet (TEJ) Conditions
 To study the mean vertical velocities during jet stream conditions and quiet conditions, the days are sorted out according to horizontal wind speeds. These sorted out jet days and rests as quiet days are then averaged to measure mean vertical velocities during the jet and quiet conditions. Here, the jet conditions are taken as the days in which magnitudes of zonal velocities exceed 30 ms−1 [Rao et al., 2000] and quiet conditions are considered when magnitudes lie below 30 ms−1. Figure 8 shows profiles of mean vertical velocities along with zonal velocities during TEJ conditions during 1996 and 1997. Over Indian subcontinent, jet streams normally occur during monsoon months (June, July and August). During jet stream conditions, there appears a transition in vertical wind from downward to upward motion with a depth of 3 to 5 km near the jet wind maximum. In 1996, the reversal in vertical velocities occurs near the height of the zonal wind maximum. But during 1997, the reversal is observed below the height of zonal wind maximum. The height of reversal in vertical wind is somewhat different as observed by Fukao et al.  using MU radar in which they always observe the reversal at wind maximum (>60 ms−1). Here it is found that the height of reversal is varying and always not at wind maximum. The observed disparity may be attributable to fundamental differences in the nature of jet stream circulation at the two locations or on the extent of time averaging of vertical velocities. Over the Indian MST radar site, the horizontal winds rarely exceed 60 ms−1, and it may be difficult to observe sharp reversals in vertical wind with downward motion below and upward motion above jet stream wind maximum. The smaller depth of ascent near jet stream maximum can be attributed to lower jet speeds. Figure 9 shows the characteristics of vertical velocities with respect to observed zonal wind during quiet conditions. From the figure it can be seen that during quiet conditions no such reversal is observed near the zonal wind maximum.
 In order to assess the dependence of reversal in vertical wind on time averaging, a case study is presented by taking diurnal cycle data on 23 July 1996 and is shown in Figure 10. The 12-hour average vertical velocities (Figure 10a) show a sharp reversal with positive values above 16 km altitude, which is at a slightly higher height above the zonal wind maximum. The other profiles in Figures 10b, 10c, 10d, and 10e, with different time averages also showed a reversal just above the zonal wind maximum with fluctuation in vertical wind from negative to positive values. In this example it is found that the larger the averaged time, larger the height with positive velocities above the reversal. From this it is clear that the reversal in vertical wind near zonal wind maximum depends on time averaging of vertical wind data and the shape of the jet as also reported by Yoe and Ruster . These updrafts near jet stream wind maximum can be due to instabilities produced by the large wind shear at these heights [Muschinski, 1996].
5. Summary and Conclusion
 The mean vertical motions observed using Indian MST radar located at Gadanki, India, show wave like structure with vertical wave length of ∼6 km. The observed values range from 3 to 20 cm s−1 on average. Positive values are observed between 5 and 10 km with thickness varying from 2 to 3 km in monsoon months. Reversal in horizontal wind direction due to monsoon circulation and due to horizontal velocity convergence causes this upward motion. Another transition in vertical wind from downward to upward is observed near the height of zonal wind maximum when the magnitude of zonal wind exceeds 30 m s−1. This second feature is generally observed in monsoon season. Earlier literature suggests that the observed vertical velocities will be less in the tropical latitudes when compared to mid and polar latitudes. But here interestingly large magnitudes are observed even in the monthly (even in long term averages) averages particularly in the lower troposphere. Observations at Gadanki show the existence of correlation between variance in vertical velocity (σw2) and mean vertical velocity () in the lower troposphere. This suggests that the gravity wave activity is large at this latitude and hence large magnitude in vertical velocities and its variance. In the long term averaged vertical velocities also, a clear wave like structure is seen with constant upward and downward motions with ∼6 km wavelength.
 The authors wish to thank UGC-SVU Centre for MST Radar Applications and National MST Radar Facility (NMRF), Gadanki, for providing necessary facilities to carry out this work. One of the authors (V.V.M.J.R.) is thankful to the Commissioner, Department of Technical Education, Government of Andhra Pradesh, India, for permitting to carry out these studies. M.V.R. is thankful to CSIR for providing SRF to carry out this work.