Assessment of variational method schemes for synthetic aperture radar wind retrieval

Synthetic aperture radar (SAR) high spatial resolution sea surface wind retrieval is of great significance for studying atmospheric and ocean processes in the marine coastal area. The wind speed retrieved directly by CMOD geophysical model function is greatly affected by the prior wind direction error. A wind direction error of 30 introduces SAR wind speed uncertainty up to 40%. The variational method (VAR) considers the uncertainty of radar observation and background wind, which can reduce the influence of prior wind direction on the retrieved wind speed. However, different observation errors or geophysical model function choices will form different VAR schemes, which lead to different wind results. Based on Sentinel 1 SAR data and NDBC buoy data, different variational schemes are systematically assessed in this paper. The tests show that when the observation error is 0.1 dB and the geophysical model function is CMOD7, the RMSE of retrieved wind speed is 1.11 m.s(-1), which is the smallest overall. However, other VAR schemes can obtain relatively better retrieval accuracy in a specific wind speed range. The accuracy of retrieved wind speed by each VAR scheme is significantly higher than that of background wind speed and CMOD direct retrieved wind speed, but the retrieved wind direction is not significantly improved compared with the background wind direction. The results can provide a reference for the selection of VAR methods under different application requirements.