Wind speed retrieval from Chinese Gaofen-3 synthetic aperture radar using an analytical approach in the nearshore waters of China’s seas

Abstract

ABSTRACT Empirical wind-retrieval algorithms for synthetic aperture radar (SAR), such as the geophysical model function (GMF) CMOD5N in the C-Band, are currently well developed. The retrieval accuracy of GMFs, however, is lower in the presence of marine phenomena such as oceanic fronts and mesoscale eddies, making SAR wind retrieval challenging under these conditions. In this study, we proposed a scheme for wind retrieval from Chinese Gaofen-3 (GF-3) SAR images using an analytical approach to obtain the wind speed in the nearshore waters of China’s seas. Approximately 300 images were acquired from 2018 to 2021 in the copolarization channels [vertical–vertical (VV) and horizontal–horizontal (HH)]. The images were collocated with the measurements from the Advanced Scatterometer (ASCAT) and the Chinese Haiyang-2B (HY-2B) scatterometer. Our analytical approach was based on the Bragg resonant and non-Bragg (NB) scattering model for the calculation of the normalized radar backscatter cross-section (NRCS), which accounted for the influence of wave-current interactions and breaking waves in nearshore waters. We used the sea surface currents obtained from the HYbrid Coordinate Ocean Model (HYCOM) to simulate the Bragg resonant roughness. We derived the contribution of the NB breaking waves from the SAR-measured NRCS values in both the VV and HH channels. By comparing the retrieval results obtained using the analytical approach with the scatterometer products, we concluded that the root mean square error (RMSE) of the wind speed was 2.01 m s−1 with a 0.91 correlation coefficient (COR), which was less than the 2.93 m s−1 RMSE and 0.72 COR for the GMF CMOD5N. The variation in the bias of the HYCOM currents was approximately 1 m s−1 for current speeds greater than 0.3 m s−1. These findings verified the stability of the analytical approach for SAR wind retrieval in complex sea states.