A Method for the Use of Satellite Retrievals as a Transfer Standard to Determine Systematic Radiosonde Errors

Abstract

This paper presents the results of a feasibility study to evaluate a method for the use of satellite measurements as a transfer standard to determine temperature biases between radiosonde types. The method was evaluated on a sample of satellite observations that were paired with nighttime radiosonde observations. Only nighttime cases were used in this study in order to avoid the additional complication of heating of the radiosonde by solar radiation; however, the method should be equally valid in daylight. Radiances were calculated from radiosonde temperature profiles and compared to radiances for the same location as measured from the satellite. With the use of radiosonde?satellite pairs for two different radiosonde types, the satellite radiances were used to remove radiance differences due to atmospheric temperature differences from the radiosonde radiances. This step allowed radiosondes at different locations to be compared. Biases between the U.S. civilian instrument and the Väisälä instrument were derived and compared with published values; the results of the new method were found to be consistent with results obtained from direct comparisons of radiosonde instruments when measurements were made under similar atmospheric conditions. However, the direct measurements were made in a limited range of atmospheric conditions, whereas the satellite measurements were made under a wide range of atmospheric conditions. Results based on the indirect satellite comparisons showed substantial variation in the biases obtained under different atmospheric conditions. This variation is consistent with differences in temperature errors that are the result of differences in the radiation balance between the instrument and its surroundings. The results demonstrated the ability of the method to provide estimates of radiosonde biases. They also show that radiosondes are subject to substantial errors owing to longwave radiation and other sources. These errors are not only large (0?2 K), but also highly variable.