基于Sentinel-1A的长江口近岸风矢量场反演研究

Sea surface wind vector retrieval off the Yangtze Estuary based on Sentinel-1A

选取长江口外近岸海域,以欧空局发射的Sentinel-1A为SAR图像数据源,利用多级小波变换和傅里叶变换获得高精度风向信息,再利用基于GMF的3种CMOD系列模型反演风速,最后将反演结果与同时相的ECMWF模式风场数据以及五组实测数据进行对比.结果表明,除个别包含复杂纹理特征的SAR子图像以外,SAR风场反演精度整体优于ECMWF模式风场.经过多级小波变换后的风向反演结果更优,均方根误差分别为31.6°,29.7°,23.5°.经过二级小波变换之后的整景SAR图像结合风向信息代入到CMOD系列模型中反演得到的风速精度最优,均方根误差控制在0.8 m/s.比较适合长江口外近海海域的是MOD-IFR2和CMOD4,均方根误差分别为1.08 m/s和1.05 m/s.

This paper selected SAR images in the Yangtze Estuary from Sentinel-1Awhich was launched by ESA. First ly, i t extracts high-precise wind direct ion using multi level wavelet transform and fourier transform. Secondly, wind speed informat ion are obtained through three kinds of CMOD models of GMF, and compare wi th five sets of in-situ data and as well as wind vector from ECMWF. The results indicats that except for some sub-images containing complex texture, the precision of SAR-derived wind field is bet ter than that of ECMWF. The multiscale wavelet transform is conducive to wind field retrieval （RMSE of wind direct ion are 31.6°, 29.7°, 23.5° respectively）, especially the wind speed retrieval after the second level wavelet （RMSE of wind speed is 0.8 m/s） . Among CMOD models, CMOD- IFR2 and CMOD4 are suitable for coastal area of the Yangtze Estuary （RMSE of wind speedis 1.08 m/s and 1.05 m/s respectively）.