Research On Retrieval Of Sea Surface Wind Filed From Dual-polarization SAR Images

As the main bond between the ocean and the atmosphere,sea surface wind regu-lates and controls the changes of the ocean and the atmosphere.Therefore,acquisition of accurate sea surface wind filed information has important scientific research value and practical significance for promoting technological development and marine economic.As an active microwave imaging radar,synthetic aperture dadar?SAR?that possesses charac-teristics of high resolution,all-weather,all-time,and strong ability to penetrate clouds is an important means of sea surface wind retreival.How to use SAR to retrieve sea surface wind field has become a research hotspot.In this paper,we analyze two main algorithms of SAR wind direction algorithm,fast Fourier transform?FFT?method in the spectral domain and local gradients?LG?method in the spatial domain.The spectral method is almostly aligned with wind streak,but it works fine only on open oceans and large areas,such as 20km×20km.To improve the accu-racy and resolution of FFT-retrieved wind directions,a combined wind direction retrieval method?CWDR?that is based on a modified cycle median filter?MCMF?is proposed for improving accuracy of FFT-retrieved wind direction utilizing wind direction from LG.In our experiment,the proposed method is validated at different spatial resolutions,and the resolution of FFT-retrieved wind directions can be improved to 3km×3km.In addition,dual-polarization SAR images and in situ buoy winds are collected and used to analyze the relations of cross-polarization normalized radar cross section?NRCS?with sea surface wind speeds and incidence angles.Based on the characteristic of cross-polarized NRCS that it's weakly dependent on wind direction,a wind speed retireval method for dual-polarization SAR?WSRD?is proposed by combining co-and cross-polarization geophysical model functions.For the match-up dataset,the bias and cor-relation coefficient of DWSR is-0.9 ms^-1and 0.846,respectively.Besides,two SAR images are used to retrieve wind filed by WSRD with wind directions from CWDR and the great results demonstrate the effectiveness of proposed models.