Global and Seasonal Variations of Stratospheric Gravity Wave Activity Deduced from

Transcription

1 1610 JOURNAL OF THE ATMOSPHERIC SCIENCES Global and Seasonal Variations of Stratospheric Gravity Wave Activity Deduced from the CHAMP/GPS Satellite M. VENKAT RATNAM, * G. TETZLAFF, AND CHRISTOPH JACOBI Institute for Meteorology, University of Leipzig, Leipzig, Germany (Manuscript received 24 April 2003, in final form 5 January 2004) ABSTRACT Global analyses of gravity wave (GW) activity in the stratosphere are presented using radio occultation data from the Challenging Minisatellite Payload (CHAMP) satellite. Temperature profiles obtained from CHAMP/ GPS radio occultations are first compared with ground-based instruments. In general, good agreement is found between these different techniques. Monthly mean values of potential energy E p, being a measure of GW activity, which is estimated with radiosonde observations, are compared with CHAMP/GPS data and it is found that radiosonde-observed E p values are slightly higher than those estimated with radio occultations. Strong diurnal variation of GW activity has been found. From the global morphology of GW activity, large E p values are noticed, besides at tropical latitudes, even at midlatitudes during winter, but not during equinoxes. This suggests that wave activity at stratospheric heights is not only modulated due to orography (mountain/lee waves) but mainly depends on seasonal variations at the respective latitudes. Significant correlations are found between GW activity and the outgoing longwave radiation (OLR) observations, OLR being a proxy for tropical deep convection. Gravity wave activity is found to be high in the zones of deep convection confirming that convection is the major source of GW generation in the Tropics. Latitudinal and vertical variations of GW activity reveal the existence of large E p values below 25 km and low values between 25 and 30 km in all the seasons near the equator. During the Southern Hemisphere winter, large values are noticed. Large values are also found during equinoxes, and these values are nearly the same in Northern and Southern Hemispheres (NH and SH, respectively) at midlatitudes. During solstices, the E p distribution involves a larger hemispheric asymmetry at middle and higher latitudes. The latitudinal range is wide ( 30 latitude in both hemispheres) with large E p values in all seasons. Large values of E p are noticed during the major stratospheric sudden warming that occurred over Antarctica during September Introduction It is known that gravity waves (GWs) exert a major influence on the large-scale circulation and structure of the atmosphere. Gravity waves are considered to be responsible for much of the spatial (a few to few thousand kilometers) and temporal variability ( 5 min to several hours) in many of the atmospheric thermodynamic variables above the tropopause. The notable sources for the generation of GW include processes like tropospheric weather systems (convective weather fronts, storms, etc.), winds gusting over mountains, jet streams of the troposphere and stratosphere, breaking of tidal and planetary waves, explosions, and instabilities at all heights. Most of the sources for the generation of these GW lie * Current affiliation: Radio Science Center for Space and Atmosphere, Kyoto University, Uji, Japan. Corresponding author address: Prof. Christoph Jacobi, Institute for Meteorology, University of Leipzig, Stephanstraße 3, D Leipzig, Germany. jacobi@rz.uni-leipzig.de in the troposphere (Fritts and Nastrom 1992). In the Tropics, it is generally thought that GW are mostly generated by cumulus convection (Alexander and Holton 1997; Piani et al. 2000; Alexander et al. 2000, and references therein). Recently, Beres et al. (2002) stressed the importance of tropospheric wind shears and that it is essential to include them in GW parameterizations. The general properties of such waves in an isothermal atmosphere were first derived and applied to upper atmospheric phenomena by Hines (1960). In general, the spectral range and energy density of the wave systems are greater at lower heights where the energy available for their generation is greater. However, because of the filtering processes of reflection and dissipation, the spectral range at higher heights becomes narrower. Significant vertical variations of temperature and wind can cause the waves to be reflected and, sometimes, to be ducted between two reflection levels in the atmosphere or between one such level and the ground. Gravity waves play a crucial role in driving the general circulation of the middle atmosphere (Manzini and McFarlane 1998; Charron and Manzini 2002, and reference therein). In any general circulation modeling ef American Meteorological Society

2 1JULY 2004 RATNAM ET AL fort, GW effects must be considered properly because effects of the wave-driven pumping can be significant at much lower altitudes, both on tropical upwelling and on polar downwelling. The upward transport and the subsequent deposition of momentum and energy from lower heights form a major part of the dynamical coupling between different parts of the atmosphere. This determines the general circulation and structure of the middle atmosphere (Houghton 1978; Lindzen 1981; Holton 1983; Warner and McIntyre 1996). During the past two decades, with the advent of very high frequency (VHF) radars and lidars, considerable effort has been devoted in characterizing GW. Unfortunately, radars are blank in the upper stratosphere so that lidars have to fill this gap. Although these techniques can provide observations with excellent temporal and spatial resolution, the network of these groundbased instruments is coarse and hence the global morphology of GW activity as acquired with these techniques is poorly known. Satellite observations are able to provide global coverage (Wu and Waters 1996; McLandress et al. 2000) but are of poor spatial resolution and are not suitable to retain the spectral properties of GW. However, they can give a quantitative picture of wave activity. Recently, using global positioning system (GPS)/Meteorological (MET) satellite observations Tsuda et al. (2000) provided a global analysis of stratospheric GW activity with special emphasis on winter months. Using the same dataset, Nastrom et al. (2000) had compared the GW energy observed by VHF radar with GPS/MET data. However, radio occultation data from the GPS/MET experiment are limited to short time intervals only. Therefore, in the present study, GW activity on a global scale, observed by the Challenging Minisatellite Payload (CHAMP)/GPS satellite is presented using almost 10 times the number of occultations seen by GPS/MET. 2. Brief system description and database The CHAMP/GPS satellite was launched on 15 July 2000 into an almost circular and near-polar orbit (inclination 87 ) with an initial altitude of 454 km. The satellite is equipped with instruments for gravitational, magnetic, and ionospheric measurements. In addition, a GPS receiver on board, in combination with precise clocks and position determination, allows the measurements of atmospheric refractivity during radio occultation (RO) events. Details of the system can be taken from Reigber et al. (2000). For the present study we use level-2 version 002 data from May 2001 to February 2002 and level-3 version 004 data from March 2002 to January 2003, which are produced by Geo Forschungs Zentrum (GfZ) Potsdam using their standard methods for RO processing. The atmospheric excess path is derived using a double diffraction technique. Atmospheric profiles are calculated assuming geometric optics and applying an Abel inversion technique (Hocke 1997) in the version 002 dataset. The significant difference between levels 2 and 3 is that in the later version a wider window for interpolation of the phase data, a statistical optimization approach for the bending angles after S. V. Sokolovskiy and D. Hunt (1996, unpublished manuscript) (instead of Hocke 1997) and, for the initialization of the hydrostatic equation, the European Centre for Medium-Range Weather Forecasts (ECMWF) data are used, and the top of the profiles is set to 35 km. The first occultation measurement from CHAMP/GPS was performed on 11 February 2001; since then about 150 to 200 occultations per day were recorded. Besides this data, temperatures observed with ground-based instruments (radiosonde and lidar) are also collected during the above-mentioned time interval to compare with CHAMP/GPS data. Throughout the observational period, antispoofing (A/S) was activated, but the data quality is regained by using a sophisticated occultation antenna geometry, which allowed the soundings with high accuracy and vertical resolution. More details of the data analysis, its processing, initial results of the CHAMP/GPS and validation with the corresponding global weather analysis can be taken from Wickert et al. (2001). 3. Methodology for calculation of GW energy density It is well known that atmospheric parameters fluctuate on a wide range of scales. In the mesoscale, wind and temperature fluctuations are sometimes described using frequency ( ) and wavenumber (m, k) power spectra since they are observed as a superposition of many waves with various frequencies and wavenumbers. The modeling of atmospheric GW spectra has followed the pioneering efforts of Garret and Munk (1972, 1975), who developed a dynamical theory for the spectrum of internal waves in the ocean. The key element of the Garret Munk model is that the waves obey the polarization and dispersion relations. Since then several GW spectral models have been developed for comparison with mesosphere stratosphere troposphere (MST) radar experiments (VanZandt 1982, 1985; Scheffler and Liu ; Fritts and VanZandt 1987). Under linear GW theory, neglecting the effects of the background wind, Scheffler and Liu (1985) derived an equation relating the observed one-dimensional frequency spectrum to the Garret Munk model GW spectrum. The energy density E 0 is chosen as a measure of GW activity and is defined as (see, e.g., Allen and Vincent 1995; Vincent et al. 1997; Tsuda et al. 2000) [ ] g T E0 u w 2 N T Ek E p, (1) where E k and E p are the kinetic and potential energy per unit mass, respectively. They can be written as

3 1612 JOURNAL OF THE ATMOSPHERIC SCIENCES FIG. 1. Two typical examples showing the comparison of CHAMP/GPS satellite-observed vertical profiles of temperature with radiosonde and lidar located at Gadanki (13.5 N, 79.2 E) on (a) 4 Mar 2002 (warm tropopause) and (b) 1 May 2002 (cold tropopause). where Ek (u w ), (2) g T Ep, (3) 2 N T g dt 2 N. (4) T dz Here u,, and w are the perturbation components of the zonal, meridional, and vertical wind, respectively; g is the acceleration due to gravity; N is the Brunt Väisälä frequency; T and T are the mean and perturbation components of temperature; and is the dry adiabatic lapse rate. According to the linear theory of the GW, the ratio of kinetic to potential energy becomes constant; therefore, it is possible to estimate E 0 from temperature observations only (Tsuda et al. 2000). In Eq. (3), the calculation of E p mainly depends on the estimation of the temperature fluctuation. For this, the procedure adopted by Tsuda et al. (2000), that is, calculating temperature fluctuation by high-pass filtering with a cutoff at 10 km, is closely followed here. 4. Results and discussion a. Comparisons with ground-based instruments Before going into details of the GW activity seen by the CHAMP/GPS satellite, it is desirable to compare the observed temperature profiles with some reference techniques. Wickert et al. (2001) have compared CHAMP observations with corresponding ECMWF profiles in the height range 5 25 km and found excellent comparison within 1 K in both hemispheres, but with some negative bias at tropical latitudes. In the present study, temperatures observed with radiosondes (10 30 km) over two stations at a tropical (13.5 N, 79.2 E) and other at a subtropical latitude (25 N, 121 E) and also with lidar (30 35 km) located at a tropical station has been selected to compare with CHAMP/GPS data. In order to reduce the error due to the temporal and spatial difference of the CHAMP data, only differences of 2 in latitude, 20 in longitude, and 2 h has been accepted as tolerable for coincidences of ground-based and satellite-derived profiles. Figure 1 shows two typical examples of the comparison of CHAMP temperature profiles with radiosonde and lidar measurements taken at Gadanki (13.5 N, 79.2 E) on (Fig. 1a) 4 March 2002 (example with warm tropopause) and (Fig. 1b) 1 May

4 1JULY 2004 RATNAM ET AL FIG. 2. Statistical comparison between temperature profiles derived from CHAMP measurements and from radiosonde (Taiwan) and lidar measurements (India). The differences (a) CHAMP radiosonde are plotted for 59 profiles and (b) CHAMP lidar are plotted for 20 profiles of nearest coincidence (example with cold tropopause). In general, CHAMP and ground-based observed profiles are matching well. There exists some difference below 5 km (sometimes up to 10 km), which is due to water vapor and that occurs from incomplete temperature retrieval at these heights. The correction due to the ionospheric residuals will also create a problem above 45 km, and sometimes even from 35 km upward (Rocken et al. 1997; Syndergaard 2000). Hence we will restrict our studies to the height range between 10 and 35 km only. In addition, one can see some differences near the tropopause, which are of different sign. In the 1 May 2002 profile, the sharp tropopause is not completely resolved by the radiosonde. Hence these GPS measurements will be highly useful in studying especially the characteristics of the tropical tropopause where most of the time a sharp tropopause can be seen (Randel et al. 2003). Comparison between temperature profiles derived from CHAMP measurements and from radiosonde (Taiwan) and lidar measurements (India) for all seasons is shown in Fig. 2. The differences observed with (Fig. 2a) CHAMP radiosonde measurements are plotted in between 8 and 25 km and (Fig. 2b) CHAMP lidar measurements are plotted in between 27 and 35 km during May 2001 to June With the aforementioned selection criteria, we obtained 59 profiles for CHAMP radiosonde and 20 profiles for CHAMP lidar comparisons. The comparison of the CHAMP profiles with radiosonde (Fig. 2a) shows excellent agreement. The mean deviation is smaller than 1 K between 8 and 25 km with a standard deviation of less than 2 K. Near the tropical tropopause (around 17 km), the mean deviation is about 1 K, with colder CHAMP data, as also has been observed by Wickert et al. (2001) using ECMWF analysis. This is possible due to a better vertical resolution of RO measurements in comparison to the analyses that were available on standard pressure levels (radiosonde). The mean deviations between CHAMP and lidar measurements are found to be less than 2 K with equal amount of standard deviations and always showing warmer CHAMP measurements. Since the number of observations with nearest coincidence is small, a more careful validation based on a larger dataset is required. b. GW potential energy observed with CHAMP GPS and radiosonde In this section, estimations of E p from the temperature perturbations T using ground-based instruments and CHAMP data are presented. Figure 3a shows the vertical profiles of temperature observed with CHAMP/GPS ( km), radiosonde (0 27 km), and lidar (27 35

5 1614 JOURNAL OF THE ATMOSPHERIC SCIENCES FIG. 3. Vertical profiles of (a) temperature, (b) temperature perturbation estimated using monthly mean, (c) temperature perturbation estimated using a high-pass filter, (d) the phase shift between CHAMP and radiosonde profiles (practical difficulty in getting the GW properties with satellite observations), (e) Brunt Väisälä frequency squared, and (f) potential energy observed on 4 Mar 2002.

6 1JULY 2004 RATNAM ET AL FIG. 4. Scatterplot of potential energy observed with (a) radiosonde observations at 0000 vs 1200 UTC, and (b) CHAMP/GPS vs radiosonde (average of 0000 and 1200 UTC) observations during Jun 2001 to May 2002 in the height region between 20 and 25 km. km) observed on 4 March The location of the CHAMP profile is 131 N, 13.6 E. In order to see the temperature fluctuation defined as deviations from the mean, monthly averaged profiles of temperature observed with radiosonde and lidar are also plotted in the same figure. The overall comparison between these various techniques show that the variations generally are matching well. Figure 3b shows T calculated as deviations from the monthly mean. It is clear that large fluctuations exist on a long-period time scale and the estimation of potential energy using these perturbations will lead to unrealistic values. Instead of this, T is estimated from the single CHAMP profile by applying a high-pass filter with a cutoff at 10 km following Tsuda et al. (2000). This is demonstrated in Fig. 3c. Another practical difficulty using this high-pass filter arises near the tropopause especially in the Tropics, where frequently a sharp tropopause exists. Therefore, the estimation of E p using high-pass filtering is also not possible near the tropopause, especially at low latitudes. Even though this method serves best to obtain the global morphology of GW energy, some spectral information on GW (amplitude and phase) will be lost due to the low horizontal resolution (200 to 400 km) and spherical symmetry assumptions in the temperature profile retrieval (Belloul and Hauchecorne 1997; Lange and Jacobi 2003). Due to these assumptions, very smallscale and large-scale GW are filtered out, and the E p values estimated through these T will be too small. Nevertheless, due to its global coverage one may obtain an overall picture of GW activity at least qualitatively. Fig. 3d shows typical T profiles observed between 20 and 25 km. The amplitude observed by CHAMP is smaller than that obtained by radiosonde, and additionally there is a phase shift between the two profiles. Since a 5000-km horizontal distance exists between the two profiles, some planetary wave effects may arise, hence it is not clear whether the same phenomena would be observed if both observations were available at the same place and time. Figure 3e shows the Brunt Väisälä frequency squared, N 2, observed with CHAMP. As expected, N 2 is low in the troposphere and shows a rather abrupt transition near the tropopause, while high values of N 2 are observed in the stratosphere. Figure 3f shows the vertical profiles of E p observed with radiosonde, lidar and CHAMP/GPS occultations on the same day. At most of the heights, E p values estimated from groundbased instruments are showing higher values. c. Diurnal variation of potential energy Monthly comparison of E p values between 20 and 25 km observed with radiosonde at 0000 UTC and 1200 UTC and also with CHAMP/GPS radio occultations observed during June 2001 to May 2002 over Taiwan (25 N, 121 E) are shown in Fig. 4. The line entered in Fig. 4 is the best fit between the two observations. The E p values observed at 1200 UTC are larger than those observed at 0000 UTC, which might be due to increased daytime GW activity due to convection that, besides wind shear (Beres et al. 2002), is believed to be the major source of GW in the Tropics and subtropics (Alexander and Holton 1997). For about 66% of the time, E p values observed at both times have the same values, and rest of the time the 1200 UTC value is observed to be larger. Comparison of E p values estimated with radiosonde and CHAMP/GPS shown in Fig. 4b reveals that the radiosonde observed E p values are slightly larger than those observed with CHAMP/GPS data. This can be attributed to the smaller magnitudes of T using the high-pass filter of the radio occultation processes. From theoretical estimations, Tsuda et al. (2000) found

7 1616 JOURNAL OF THE ATMOSPHERIC SCIENCES FIG. 5. Distribution of 5093 CHAMP/GPS occultations observed during the month of Apr 2002 across the globe. that there will be a decrease in amplitude of up to 21% for GPS/MET RO data. Using a 2D model, Lange and Jacobi (2003) studied the influence of geometric wave parameters and the measurement geometry on plane GW in the range of km horizontal and 1 10-km vertical wavelengths and found that the radio occultations can resolve more than 90% of the simulated GW with 60% amplitude level and more than the 50% of the derived amplitudes are above 90%. To show the diurnal variations of the GW activity at different longitude bands at particular latitude, data have been selected in such a way that they should cover most of the local times. The distribution of the 5093 occultation events taken in April 2002 across the globe is shown in Fig. 5. Occultations are distributed throughout the globe, even when only 1 month of data is considered. Another interesting fact is the good coverage of polar latitudes, which will provide reliable statistical results even there. The diurnal variation of E p at km observed with CHAMP/GPS during April 2002 is shown in Fig. 6. This figure shows the departure of E p from the diurnal mean over bands of longitudes (120 in steps of 60 ) near the equator ( 10 ). April 2002 was selected because of the good coverage of all local times (more than 80%) in this month. Missing values in Fig. 6 are interpolated. The number of observations over the respective longitudes is also shown in parentheses and the uncertainty of E p is also given; clearly the uncertainty is smaller than the observed phenomena. The variations indicate a westward propagating semidiurnal wave with zonal wavenumber 2 peaking at noon and in the first half of the night. It may be concluded that in further investigations the diurnal variation of E p should be taken into account. However, the database used for the present study does not yet consist of data in all months at all local times so that it is difficult to obtain global coverage at a particular time of the day. Hence, all available occultations are averaged here while presenting the global morphology of GW activity, leaving the task of analyzing the diurnal variation to a further study when more observations will be available using the upcoming Gravity Recovery and Climate Experiment (GRACE) satellite observations. d. Global distribution of GW activity in the lower stratosphere The global distribution of E p values observed during March, July, September, and December 2002 thus covering all the seasons, averaged between 20 and 25 km, are shown in Fig. 7. Potential energy contour intervals are 2 J kg 1. The number of occultations used for this figure amounts to 3656, 4397, 4288, and 3936 in the respective months. Since the number of occultations per day has increased tremendously after February 2002, we present the global distribution of E p from March 2002 onward. From Fig. 7, it is evident that at tropical and subtropical latitudes ( 30 latitude) E p values are large in all the months, which could be due to larger convection than expected and partly due to equatorial waves. An interesting feature in July and December 2002 is that large values are visible even at midlatitudes over the continents and smaller values are found over the oceans during winter months of both hemispheres. Similar observations were also reported by Tsuda et al. (2000) for boreal winter months.

8 1JULY 2004 RATNAM ET AL FIG. 6. Departure of potential energy from the diurnal mean over respective bands of longitudes near equator observed during Apr The numbers within parentheses show the number of points used for generating the respective plot. However, this later feature is not clearly observed during March 2002, which reveals that the large E p values observed over the midlatitude continents are either not due to mountain/lee waves (topography) or they could not propagate to the lower stratosphere. From these two examples it is clear that the GW activity observed with radio occultations will not only depend on source distribution at least at stratospheric heights, but also may be influenced by interactions of GW with the background wind as reported by Alexander (1998). It is also well known that GW activity will be stronger at midlatitudes in winter (Allen and Vincent 1995) and at tropical latitudes during equinoxes. Another significant enhancement has been observed over Antarctica during September 2002, which is probably connected with the FIG. 7. Global distribution of GW activity observed in lower stratosphere (20 25 km) during different months covering all seasons; E p contour intervals (shaded regions) are shifted for every 2Jkg 1. Contour lines show areas of deep convection (OLR 220 W m 2 ) that occurred during respective months.

9 1618 JOURNAL OF THE ATMOSPHERIC SCIENCES unusually strong stratospheric warming, that occurred during that month (Kirstin et al. 2003, manuscript submitted to J. Atmos. Sci.) and should be further investigated. In order to study the coherence with tropical convection, the outgoing longwave radiation (OLR) is used as a proxy for tropical convection. Daily OLR data on a grid (with the data gaps filled by interpolation) are obtained from the Climate Diagnostics Center Web site (at The contour lines ( 220) in Fig. 7 show regions of deep convection, defined as OLR 220 W m 2. It can be seen that, irrespective of land/sea near the regions of deep convection, E p values are large. In general, deep convection is seen over the landmass of Africa and South America and over the Indian and Pacific Oceans. To show one-to-one coherence of GW activity with that of deep convection, a band of latitudes where deep convection is observed is selected for further study. Figure 8 shows departures of E p and OLR from their zonal means observed in the months mentioned in Fig. 7. Negative (positive) values of E p and OLR indicate decrease (increase) or increase (decrease) of GW energy and convection, respectively. The top panel of the figure shows the regression analysis performed by considering all the months shown in the bottom panels. From the figure, a very strong correlation is observed at all months presented here. Hence it may be concluded that convection is a primary source for the observed GW activity in the Tropics. e. Latitudinal variation of GW activity In this section we present the latitudinal variation of E p observed during different seasons, that is, NH winter (November February), spring equinox (March April), summer (May August), and autumn equinox (September October) with the database consisting of all data from May 2001 to January The latitudinal and seasonal variations of E p values are shown in Fig. 9. The total number of occultations used for this figure is , , 8811, and 7323 for NH winter, summer, spring, and autumn, respectively. The salient features noticed from Fig. 9 are large values of E p at low latitudes below 25 km in all the seasons. The latitudinal range of high potential energy is wider as expected (up to 30 in both hemispheres) in almost all seasons and in the winter it is even wider. Potential energy values are nearly the same at midlatitudes during all the seasons in both hemispheres. These values are low between 25 and 30 km near the equator in almost all seasons, especially during winter and summer and increasing above that height region. At higher latitudes, in winter (November February in NH and May August in SH), E p values are larger in both hemispheres. During March April, the latitudinal distribution of E p values is nearly symmetric between NH and SH in the entire height range considered. An unusual FIG. 8. Departures of E p (open circles) and OLR (closed circles) from zonal mean over respective latitudes observed in different months covering all seasons. Negative values in the E p and OLR data indicate the decrease and the increase of GW energy and convection, respectively. The top panel shows the regression analysis performed by considering all the months shown in the bottom panels. enhancement is observed at SH high latitudes during September and October 2002 months. During winter and summer seasons, the E p distribution involves a large hemispheric asymmetry at middle and high latitudes, which is more pronounced in NH summer. The E p values at N in November February are significantly larger than those at S. Similarly, E p values at S in May August are significantly larger than those observed in at N in November February, suggesting that the latitudinal variation of the GW activity in the stratosphere depends on hemisphere as well as on season. Most of the features mentioned above are also noticed by Tsuda et al. (2000) using GPS/MET radio occultations. The only fundamental differences are the wider latitudinal range of large equatorial values and the large values of E p at SH polar latitudes during September October Investigations are still going

10 1JULY 2004 RATNAM ET AL FIG. 9. Latitudinal variation of potential energy observed in the stratosphere in different seasons during May 2001 Jan on to reveal the possible reason for this significant enhancement and wider latitudinal range. 5. Summary and conclusions A global analysis of GW activity in the stratosphere is presented using CHAMP/GPS satellite temperature perturbation measurements. Initially, CHAMP/GPS vertical temperature profiles are compared with radiosondes and lidar measurements. In general, good agreement is found between these different techniques. The potential energy is calculated from the temperature perturbations (in a 2 10-km vertical window) and the background Brunt Väisälä frequency. Monthly E p values are calculated from radiosonde observations and are compared with those estimated with CHAMP/GPS in order to estimate the accuracy of the satellite measurements. In general, E p values estimated with radiosonde observations are slightly larger than those estimated with RO, which may be due to horizontal resolution and spherical symmetry assumptions implemented in the retrieval of the RO temperature profiles. Significant diurnal variation of E p has been found, revealing that satellite measurements observed at different local times should be considered separately as soon as a sufficiently large database is available. Making use of the high accuracy and good vertical resolution, along with global coverage of the CHAMP/ GPS satellite observations, global and seasonal variations of E p are studied. From monthly variations it is found that the GW activity is large not only at equatorial latitudes but also at midlatitudes during winter in both hemispheres, while during equinoxes it is large only at tropical latitudes. A good correspondence is found between stratospheric GW activity and outgoing longwave radiation data. Global and seasonal variations reveal that the largest values of E p are found at low latitudes below 25 km in all seasons. During SH winter, large values of E p are noticed, while during equinoxes E p values at midlatitudes are nearly equal in both hemispheres. Potential energy values are found to be very low between 25 and 30 km near the equator, especially during solstice. It is also found that there is a large hemispheric difference in E p values during solstice conditions. An interesting feature noticed is that the large value of E p in SH polar latitudes during September October 2002 could be due to a stratospheric warming. Even though the database used here is already large, it should still be increased in order to cover all the local times across the globe to obtain a higher statistical significance of the monthly mean results. Thus, using CHAMP data from the following years and including RO temperatures from following GPS Low Earth Orbiter missions, will enable us to derive still more reliable monthly climatologies of the stratospheric GW activity, which may be used in numerical circulation models to obtain a more realistic GW parameterization. Since the available data are not yet sufficient to study the diurnal

11 1620 JOURNAL OF THE ATMOSPHERIC SCIENCES variation of GW activity in all the months and seasons, we leave the task to a further study. Acknowledgments. We wish to thank GfZ Potsdam for providing CHAMP/GPS data through the ISDC data center and NOAA for providing OLR data through the Climate Diagnostics Center. Our sincere thanks to W. N. Chen for providing Taiwan radiosonde observations. We are highly grateful to National MST Radar Facility (NMRF) for providing lidar and radiosonde data. This research was funded by the DFG under Grant JA 836/ 4-2. REFERENCES Alexander, M. J., 1998: Interpretations of observed climatological pattern in stratospheric gravity wave variance. J. Geophys. Res., 103, , and J. Holton, 1997: A model study of zonal forcing in the equatorial stratosphere by convectively induced gravity waves. J. Atmos. Sci., 54, , J. H. Beres, and L. Pfister, 2000: Tropical stratospheric gravity wave activity and relationship to clouds. J. Geophys. Res., 105, Allen, S. J., and R. A. Vincent, 1995: Gravity wave activity in lower atmosphere: Seasonal and latitudinal variations. J. Geophys. Res., 100, Belloul, M. B., and A. Hauchecorne, 1997: Effect of periodic horizontal gradients on the retrieval of atmospheric profiles from occultation measurements. Radio Sci., 32, Beres, J. H., M. J. Alexander, and J. R. Holton, 2002: Effects of tropospheric wind shear on the spectrum of convectively generated gravity waves. J. Atmos. Sci., 59, Charron, M., and E. Manzini, 2002: Gravity waves from fronts: Parameterization and middle atmosphere response in a general circulation model. J. Atmos. Sci., 59, Fritts, D. C., and T. E. VanZandt, 1987: Effects of Doppler shifting on the frequency spectra of atmospheric gravity waves. J. Geophys. Res., 92, , and G. D. Nastrom, 1992: Sources of mesoscale variability of gravity waves. Part II: Frontal, convective, and jet stream excitation. J. Atmos. Sci., 49, Garret, C., and W. Munk, 1972: Space time scales of internal waves. Geophys. Fluid Dyn., 3, , and, 1975: Space time scales of internal waves: A progress report. J. Geophys. Res., 80, Hines, C. O., 1960: Internal atmospheric gravity waves at ionospheric heights. Can. J. Phys., 38, Hocke, K., 1997: Inversion of GPS meteorology data. Ann. Geophys., 15, Holton, J. R., 1983: The influence of gravity wave breaking on the general circulation of the middle atmosphere. J. Atmos. Sci., 40, Houghton, J. T., 1978: The stratosphere and the mesosphere. Quart. J. Roy. Meteor. Soc., 104, Lange, M., and Ch. Jacobi, 2003: Analysis of gravity waves from radio occultation measurements. First CHAMP Mission Results for Gravity, Magnetic and Atmospheric Studies, Ch. Reigber, H. Lühr, and P. Schwintzer, Eds., Springer Verlag, Lindzen, R. S., 1981: Turbulence and stress owing to gravity wave and tidal breakdown. J. Geophys. Res., 86, Manzini, E., and N. A. McFarlane, 1998: The effect of varying the source spectrum of a gravity wave parameterization in a middle atmosphere general circulation model. J. Geophys. Res., 103, McLandress, C., M. J. Alexander, and D. L. Wu, 2000: Microwave limb sounder observations of gravity waves in the stratosphere: A climatology and interpretation. J. Geophys. Res., 105, Nastrom, G. D., A. R. Hansen, T. Tsuda, M. Nishida, and R. Ware, 2000: A comparison of gravity wave energy observed by VHF radar and GPS/MET over central North America. J. Geophys. Res., 105, Piani, C., D. Durran, M. J. Alexander, and J. R. Holton, 2000: A numerical study of three-dimensional gravity waves triggered by deep tropical convection and their role in the dynamics of the QBO. J. Atmos. Sci., 57, Randel, W. J., F. Wu, and W. R. Rios, 2003: Thermal variabilty of the tropical tropoause region derived from GPS/MET observations. J. Geophys. Res., 108, 4024, doi: /2002jd Reigber, C., H. Lühr, and P. Schwintzer, 2000: CHAMP mission status and perspectives. Eos, Trans. Amer. Geophys. Union, 81 (Suppl.), F307. Rocken, C., and Coauthors, 1997: Analysis and validation of GPS/ MET data in the neutral atmosphere. J. Geophys. Res., 102, Scheffler, A. O., and C. H. Liu, 1985: On observation of gravity wave spectra in the atmosphere by using MST radars. Radio Sci., 20, , and, 1986: The effects of Doppler shift on gravity wave spectra observed by MST radar. J. Atmos. Terr. Phys., 48, Syndergaard, S., 2000: On the ionosphere calibration in GPS radio occultation measurements. Radio Sci., 35, Tsuda, T., M. Nishida, C. Rocken, and R. H. Ware, 2000: A global morphology of gravity wave activity in the stratosphere revealed by the GPS occultation data (GPS/MET). J. Geophys. Res., 105, VanZandt, T. E., 1982: A universal spectrum of buoyancy waves in the atmosphere. Geophys. Res. Lett., 9, , 1985: A model for gravity wave spectra observed by Doppler sounding systems. Radio Sci., 20, Vincent, R. A., S. J. Allen, and S. D. Eckermann, 1997: Gravitywave parameters in the lower stratosphere. Gravity Wave Processes: Their Parameterization in Climate Models, K. Hamilton, Ed., NATO ASI Series 150, Springer, Warner, C. D., and M. E. McIntyre, 1996: On the propagation and dissipation of gravity wave spectra through a realistic middle atmosphere. J. Atmos. Sci., 53, Wickert, J., and Coauthors, 2001: Atmospheric soundings by GPS radio occultation: First results from CHAMP. Geophys. Res. Lett., 28, Wu, D. L., and J. W. Waters, 1996: Satellite observations if atmospheric variances: A possible indication of gravity waves. Geophys. Res. Lett., 23,