Study of Wind Speed Retrievals From Sentinel-1 Images Using Physical Models

Sea surface wind plays an important role for many applications such as meteorological forecasting, oil slick observation and ship detection. From the Sentinel-1 SAR images, we propose in this paper the methods which permit to retrieve rapidly wind fields. Wind directions can be directly extracted from the SAR images by the Local Gradient method with high resolution. Wind speed is estimated by the physical (SPM) and empirical (CMOD.5) models. Compared to the empirical models, the physical ones can give the estimations of wind speed in different frequencies and for both co-polarizations. They are thus more general and more interesting. Surface wave roughness spectrum is the most important parameter in the physical models to retrieve accurately wind speed. We study two approaches in this paper: semi-empirical model proposed by Elfouhaily and al., and empirical model studied by Hwang. To evaluate, the wind speed estimated by the SPM with two models of wave spectrum and the CMOD.5 is compared for the different cases of incident angle.