Ocean retrievals for WindSat: radiative transfer model, algorithm, validation

We have developed an ocean retrieval algorithm for WindSat. The ocean retrievals are for sea-surface temperature (SST), wind speed (SSWS) and direction (SSWD), atmospheric water vapor, cloud water, and rain rate. The algorithm is based on our AMSR-E algorithm, which has been adapted to the WindSat channel set and incidence angles. One major change to the AMSR-E algorithm is the addition of the wind direction retrieval, which is made possible by WindSat measurements of the 3rd and 4th Stokes brightness temperatures. The 3rd Stokes parameter at 10.7 GHz is corrected for Faraday rotation. To compute wind direction, a maximum-likelihood approach is used similar to the method used for microwave scatterometers. A set of possible wind vectors solutions (i.e., ambiguities) is found that minimize the difference between the observations and the radiative transfer model. A median filter is then used to select the most likely ambiguity. The ocean algorithm uses all 22 WindSat channels. Retrievals for a 9-month period are presented and are compared to a variety of validation datasets. The physical basis for the algorithm is the radiative transfer model (RTM). This model expresses the microwave brightness temperature in terms of SST, wind vector, and atmospheric profiles of temperature and moisture. The WindSat observations in conjunction with Reynolds SST and numerical weather prediction model surface winds are used to refine the sea-surface emissivity component of the RTM. Some differences are found between the new emissivity model and earlier models that were based on aircraft and previous satellite observations.