Preliminary analysis of wind and wave retrieval from Chinese 1mC-SAR image

Abstract

Chinese satellite carrying synthetic aperture radar (SAR) with spatial resolution up to 1 m, denoted as 1mC-SAR, is the successor of Gaofen-3 (GF-3). The main purpose of this study is to conduct the preliminary analysis of wind and wave retrieval from more than 400 1mC-SAR images acquired in quad-polarization stripmap (QPS), which are located at China coastal waters on April 2023. The co-polarized (vertical−vertical [VV] and horizontal−horizontal [HH]) geophy sical model function (GMF), denoted as CSARMOD-GF, is employed for wind speed retrieval from those images taking prior information on wind directions from European Centre for Medium-Range Weather Forecasts (ECMWF). Validation against the wind products from Haiyang-2 (HY-2) (2B/2C/2D) scatterometers yields a 1.78/1.91 m/s root mean squared error (RMSE) with a 0.22/0.23 scatter index (SI) for SAR retrievals at VV/HH polarization channels. Moreover, the accuracy of SAR-derived winds at spatial resolution of 2 km for QPS-I and 6 km for QPS-II is relatively higher than that achieved from the retrievals at spatial resolution of 4 km for QPS-I and 12 km. The wave slop spectrum is inverted from co-polarized image by polarimetric technique, in which the term of wind speed is included in the model transfer function (MTF) of tilt modulation. Significant wave height (SWH) retrievals are compared with the simulations by the third-generation numeric wave model, denoted as WAVEWATCH-III (WW3), showing a 0.53 m RMSE with a 0.36 SI. This behavior is also confirmed as comparing with SAR-derived wave spectra and WW3-simulated wave spectra, i.e. a 0.79 correlation coefficient (Cor) and a 0.92 squared error (Err). The variation of bias in wind speed and SWH indicates an increasing tendency with the growth of sea state, meaning that calibration is a heedful issue for 1mC-SAR.