Interest in the amplitude and phase of the atmospheric diurnal temperature wave above the surface has been revived recently because of the possible aliasing of diurnal effects in long-term satellite observations of upper-air temperature. The drift in local equatorial crossing time of polar orbiting satellites can introduce spurious trends (or mask real ones), and removing the artificial signal from the data record requires information on the true diurnal cycle of temperature.

For this purpose, estimates of the climatological variation of free air temperature have been deduced from a general circulation model (Mears et al., J. Climate, 2003) and from the satellite data themselves (Christy et al., J. Atmos. Ocean. Tech., 2003), but an independent, observationally-derived estimate has not been made recently. A dataset with good vertical resolution through the troposphere and stratosphere, adequate temporal resolution of the diurnal cycle, global coverage, at least a full year of observations, and no biases related to solar elevation angle, would be best suited for such analysis. Lacking such an ideal observational resource, we turn to radiosonde observations, which, in past years, were made four times daily, at some stations, for some periods of time, on a routine basis. For our analysis, we employ data from about 96 stations (almost exclusively in the 10S-65N latitude range) to estimate the amplitude and phase of the diurnal temperature wave, at 17 levels from the surface to 10 hPa, for the layers sampled by the Microwave Sounding Unit's tropospheric and stratospheric channels (MSU channels 2 and 4), and (for tropical stations) for the tropopause level. We will show seasonal, vertical and geographic variations, discuss possible effects of observational error, compare our results to previous studies, and suggest potential applications.