Surface wind vector over the oceans is vital information for scientists and forecasters in the understanding of the Earth's global weather and climate. As the demand of global wind vector information increasing, there were numbers of satellite missions in orbit that carried instruments for surface wind measurement during the past three decade. However, there was still not sufficient numbers of instruments flown simultaneously in operation to fulfill the operational meteorological and non-real-time scientific wind vector requirements. In year 2003, the Naval Research Laboratory launches the WindSat instrument on board Coriolis satellite. WindSat was the first polarimetric passive microwave radiometer that flown in space capable of measuring ocean surface wind speed and direction. WindSat operates at five discrete frequencies at 6.8, 10.7, 18.7, 23.8 and 37.0 GHz. The 10.7, 18.7 and 37.0 GHz channel are fully polarimetric establishing all four Stokes parameters and are primary channels used for wind vector retrieval. The 6.8 and 23.8 GHz channels are dual-polarization used in conjunction with other channels to retrieve relevant surface and atmospheric parameters that complement the wind vector retrieval.

WindSat's wind vector retrieval was achievable via a model function that establishes relationship between the third and fourth Stokes parameters and the surface wind truth. Although the principal polarization (vertical and horizontal) contain unusable weakly wind direction dependence, some certain combination between the vertical and horizontal brightness temperature are almost cancel the atmospheric effect and are predominantly dependent on the sea surface temperature, wind speed and direction. The brightness temperature combination can be expressed as ATBv-TBh where A is a constant. For WindSat's earth incidence angle, the surface reflectivity of the horizontal polarization is about twice the vertical's, therefore the A is approximately 2. For low to moderate wind speed, the wind vector dependence on ATBv-TBh is weak and lies within the instrument noise level. However, the wind vector signal in ATBv-TBh reveals the opportunity to use it in combination with the third and forth Stokes model function for wind retrieval improvement. In this paper, we investigate the ATBv-TBh characteristic for the WindSat's 10.7, 18.7 and 37.0 GHz channels and present preliminary results of the wind vector model function.