Gravity waves are known to have profound impacts on the atmosphere ranging from their influences on large-scale mean flow structures to micro-scale mixing and transport of chemical species. They are a very important coupling mechanism between the lower and upper atmosphere. Traditional methods, such as hodograph analysis, rotary spectral analysis, etc., can be applied to radiosonde observations to retrieve gravity wave parameters, but these methods only captures low – frequency gravity waves (intrinsic frequency ω around f ~ 10f) that are typically generated from large – scale geostrophic adjustment processes and topography.

Here, amplifying on some earlier work by Lane et al. (2003), we show that, by applying similar approaches to ascent rate profiles (we define a new variable – “vertical energy density”, VE hereafter), we can identify middle frequency gravity waves (f << ω << N). By studying the climatology of VE, we find that they are likely to be generated by a major gravity wave source – convective activity.

A nine - year time series (1998 - 2006) of US high-resolution radiosonde data show that VE maximizes in summer at mid-latitudes within the troposphere (2 – 8.9 km), and maximizes at local afternoon – early evening during summer over most areas of the contiguous US. The diurnal pattern of VE corresponds well with that of convective precipitation, indicating that these waves are likely generated by convective activity. Furthermore, VE within both the troposphere and lower stratosphere (18 – 24.9 km) over 5 tropical western Pacific island stations is highly correlated with convective precipitation and inversely correlated with outgoing longwave radiation. The apparent dominant vertical wavelength based on Fourier analysis of the low-pass filtered perturbation field of ascent rate gives similar patterns with the diurnal and seasonal variations of VE. VE calculated within the lower stratosphere, however, does not show similar relationships to convection, but analysis of the vertical wavelength does retain some of the features.

We further estimated the mean intrinsic frequency through the ratio among kinetic energy density (KE), potential energy density (PE) and VE by assuming these waves are hydrostatic. The derived ω from horizontal wind and temperature perturbation fields (2f – 3f) in general agrees with what is derived from hodograph analysis, while ω calculated from the ascent rate and temperature perturbation fields in the troposphere have much higher frequencies (~ 20f), suggesting that we are observing middle frequency gravity waves in the ascent rate profiles. The estimated ω in the lower stratosphere is too high, indicating that our hydrostatic assumption is likely not valid for that region.

In all, our study suggests the potential of using radiosonde data to study the convection source for generating middle frequency gravity waves, and to further investigate the mechanisms of coupling between the lower and upper atmosphere through gravity waves.