基于梯度算子的高分三号SAR风场反演方法

Wind retrieval method from Gaofen-3 SAR image based on a gradient operator

鉴于高分三号(GF-3)合成孔径雷达(SAR)图像具有黑白相间条纹特征,提出一种基于梯度算子的风场反演方法。对SAR图像重新采样后,利用Sobel梯度算子求解海面风向值,接着使用扩展后的CMOD5散射模型推算出风速值。为了测试该方法,对GF-3 SAR图像进行风场反演实验,并使用欧洲中尺度天气预报中心(ECMWF)再分析风场数据和美国国家浮标中心(NDBC)浮标风矢量数据对反演得到的SAR风场质量进行评估。实验结果表明:GF-3 SAR风场相比于ECMWF再分析风场,风速、风向均方根误差分别为1.79 m/s和11.95°;相比于NDBC浮标数据,风速、风向绝对偏差分别为0.24 m/s和9.03°。因此,本文方法能有效地从GF-3 SAR图像中反演出较高精度的风场,研究结果可为GF-3 SAR风场反演技术提供参考。

Given the fact that Gaofeng-3(GF-3)synthetic aperture radar(SAR)can capture the wind streak whose gray values are changing in black and white on image,a wind retrieval method based on Sobel gradient operator is proposed to retrieve wind field from GF-3 SAR images.After resampling the SAR image,the Sobel gradient operator was applied to calculate the wind direction over the sea surface,and then an extended CMOD5 scattering model was explored to calculate the wind speed.To test this method,an experiment was done to process the GF-3 SAR image,whose retrieved wind field is then evaluated using the European Centre for Medium-Range Weather Forecasts(ECMWF)reanalysis wind field data and National Data Buoy Center(NDBC)buoy wind vector.The experiment showed that compared with the ECMWF reanalysis wind field,the root mean square error of the retrieved SAR wind speed and direction was 1.79 m/s and 11.95°,respectively,while compared to NDBC buoy data,the absolute deviation of wind speed and direction was 0.24 m/s and 9.03°,respectively.Therefore,the proposed method acted well to retrieved wind filed with higher precision from GF-3 SAR images.The research results could provide a reference for the wind field retrieval technology of GF-3 SAR.