基于ASAR GM数据时序特征的农田表层土壤水分的反演

了解农田表层土壤水分的时空变化对反演涝渍害有着十分重要的意义， 针对ASAR GM数据特点提出了在没有地面观测数据地区农田大尺度土壤水分的反演方法，即运用MODIS NDVI数据（空间分辨率为250 m）的农田作物时序特征提取土地利用现状，并以此来校正ASAR GM数据（空间分辨率为1 km）中的混合像素；利用大面积降水后土壤表层湿度水平差异不大的特点，运用非线性回归模型，获取水云模型中的经验系数，运用水云模型将植被和土壤对后向散射的贡献区分开；在获得一个地方长时间土壤后向散射系数空间分布集后，假设其土壤粗糙度不变的情况下，其后向散射系数与土壤表层湿度成正比，运用其时序特征计算出其土壤表层相对体积含水量。运用上述方法，利用79景ASAR GM数据和同时期的MODIS数据，反演了湖北省四湖地区棉田的土壤表层相对体积含水量的时空分布，将其计算值与观测值比较分析后发现，其变化趋势相同、相关系数（R2=0779，n=25）也很高，圴方根误差为963%，表明上述反演方法还是可行了，适用于大尺度农田土壤墒情的反演

RETRIEVING SURFACE SOIL MOISTURE OVER CROP FIELDS BASED ON TIME SERIES CHARACTERISTICS OF ASAR DATA

The knowledge of the surface soil moisture content can be very helpful for retrieving the spatial temporal distribution of water logging in agriculture fields. According to the characteristics of ASAR GM data, this paper retrieved surface soil moisture over the cotton fields using 79 ASAR GM images data from 2007-2011 during cotton growing seasons, the auxiliary parameters were calculated using MODIS data, the main methods include the follows: we obtained spatial distribution of land use classification using crop time series characteristics and MODIS data, corrected the matrix pixels data of backscattering coefficient using the land use classification data, and separated backscattering coefficient influenced by soil and vegetation using the semi empirical water cloud model, the parameters of which were calculated by ASAR GM data under water saturation state using non linear statistic mode and the vegetation water content in these methods came from NDVI data calculated from MODIS data, the finally corrected soil backscattering coefficient was used to calculated soil surface moisture by ASAR GM time series data. The advantage of our approach is that no auxiliary data is needed.By comparing the method value and measured data in the cotton field in Sihu area, the results indicated that this method is correct (R2=0779 n=25), the estimation precision reaches 963% in root mean square error