Ocean Retrievals for WindSat

Abstract

We have developed an ocean retrieval algorithm for WindSat retrieving sea-surface wind speed and direction, sea-surface temperature (SST), columnar atmospheric water vapor, columnar liquid cloud water, and rain rate. The physical basis for the algorithm is the radiative transfer model (RTM). This model expresses the microwave brightness temperature (TB) in terms of SST, wind vector, and atmospheric profiles of temperature and moisture. The WindSat observations in conjunction with observations from other satellites or numerical weather prediction models are used to determine or refine the wind induced sea-surface emissivity component of the RTM. For WindSat, the wind direction signal for vertical (v) and horizontal polarization (h) can be determined by taking the difference between forward and backward look, which allows a more accurate determination than using only the forward look, as atmospheric uncertainties cancel out. A new feature of the WindSat ocean algorithm compared with algorithms for earlier instruments (SSM/I, TMI, AMSR-E) is the use of the 3rd and 4th Stokes brightness temperatures. To retrieve wind direction, a maximum-likelihood approach finds a set of possible wind vector solutions (ambiguities) that minimizes the difference between the observations and the radiative transfer model. A median filter selects the most likely ambiguity. We present retrievals for a 9-month period and compare to a variety of validation datasets (buoys, ship cruises, numerical weather prediction models, satellites). The performance of WindSat retrievals for SST TS, wind speed W, water vapor V and cloud water L matches closely the ones from other microwave imagers. For wind speeds above 7 m/s, the WindSat wind direction error is below 20 deg. Accurate wind direction retrievals for wind speeds below 5 m/s are difficult due to the lack of sufficient signal size