Wind Field Retrieval From Co-Polarized SAR Imagery Without External Wind Direction Assistance

The geophysical model function for sea surface wind speed retrieval from co-polarized synthetic aperture radar (SAR) images typically requires wind direction as prior information. Obtaining wind direction relies directly or indirectly on external sources, making it impossible to achieve wind field retrieval solely based on co-polarized SAR images. This study takes advantage of the fact wind direction is similar to wind waves propagation direction, relying solely on co-polarized SAR images to obtain proper wind direction, thus enabling wind field retrieval without external wind direction assistance. This method underwent testing with Sentinel-1 Stripmap co-polarized SAR images in wind field retrieval experiments. Four cases were initially conducted to demonstrate the effectiveness of the proposed method in both downwind and upwind scenarios. Furthermore, we compared wind field retrieved from 54 co-polarized SAR images with European Center for Medium-Range Weather Forecasts (ECMWF), ASCAT, and buoy data. The root mean square error (RMSE) and the absolute value of Bias of wind direction comparison with ECMWF and ASCAT were both below 22$^{\circ }$ and 1.8$^{\circ }$, respectively. Similarly, the RMSE and the absolute value of Bias of wind speed comparison with ECMWF and ASCAT were both within 2.5 and 0.65 m/s, respectively. The retrieved wind direction was relatively consistent with the buoy wind direction, but there was a significant deviation in wind speed comparison. Statistical comparisons demonstrate the effectiveness of the proposed method for wind field retrieval from co-polarized SAR imagery without external wind direction assistance, and the method showed good performance in wind speed retrieval near the coast.