Radiometric profiling of temperature, water vapor and cloud liquid water using various inversion methods

We have developed a tunable temperature profiler based on a highly stable synthesizer that can observe at multiple frequencies on the shoulder of the 60‐GHz atmospheric oxygen feature. We are developing a similar radiometer to obtain the vertical distribution of water vapor by making observations on the pressure‐broadened water vapor line from 22 to 29 GHz. Information on cloud liquid water profiles is also contained in these two wave bands. Various mathematical retrieval methods for temperature, water vapor, and cloud liquid water profiles were tested based on these radiometer designs. These include neural networking, Newtonian iteration of statistically retrieved profiles, and Bayesian “most probable” retrievals. On the basis of realistic radiometer errors and performance, very good retrieval capability is demonstrated. The performance of the various retrieval methods are presented and compared. Examples of profile retrievals are also presented. Data were not binned into seasons to reduce computer time; better retrieval results for all methods would be expected with binning.