Analysis of upper tropospheric humidity measurements by microwave sounders and radiosondes

Abstract

This thesis describes results of several analyses of humidity measurements by microwave humidity sounders and radiosondes. The goal of this work is to pave the way for fully utilizing these measurements for climatological applications. High resolution radiosonde data are used to examine the variability of the clear-sky outgoing longwave radiation (OLR). The global variability of OLR is found to be 33 Wm-2, of which a large part can be attributed to temperature variations. The variability after filtering the temperature part is associated with the humidity variability in the horizontal and the vertical. The impact of the vertical structures on the OLR calculations is also investigated in detail. It is observed that smoothed profiles in relative humidity are sufficient to obtain the mean value of OLR, even though the variability cannot be exactly reproduced. AMSU-B Channel 18 brightness temperatures are sensitive to upper tropospheric humidity (UTH). A simple method is developed to transform the brightness temperatures to UTH. This method is validated with high quality radiosonde data. An initial attempt to make a UTH climatology and the usefulness of a robust estimator such as the median in climatological studies are discussed. Finally, a robust method was developed to compare the humidity measurements from satellite humidity sounders and radiosondes. The method is developed and tested using the high quality radiosonde data from the Lindenberg radiosonde station. A case study using different versions of the data shows that the method is sensitive to humidity differences in the different versions. The main result from the case study is that the corrected radiosonde data still have a slight dry bias in the upper troposphere. The method is then applied to assess the performance of different radiosonde sensors and stations. It isfound to be useful for monitoring the global radiosonde network, using the microwave satellite data as a benchmark.