First-guess dependence of a physically based set of temperature-humidity retrievals from HIRS2/MSU data

The first-guess dependence of temperature and humidity fields retrieved from HIRS2/MSU data using the GLA (Goddard Laboratory for Atmospheres) physically based retrieval scheme is examined. Retrievals were performed over the ALPEX region for two successive synoptic periods, 1200 UTC March 4 and 00000 UTC March 5, 1982, using three different initial guesses for each period. Results show rather low first-guess dependence for the thickness fields and larger first-guess dependence for the precipitable water fields, especially close to the surface. The humidity retrieval algorithm used is described. The processing system has the property of maintaining the accuracy of a good guess and improving a poor one for both thickness and precipitable water at all levels.