论文部分内容阅读
对我国西北黑河地区的人工林,进行了基于ENVISAT/ASAR数据构造神经网络的反演杨树林叶面积指数研究。首先,分析了白杨树林、沙枣树林的叶面积指数(LAI)与ENVISAT/ASAR不同极化后向散射系数的相关关系,研究表明人工林的空间分布均一性是影响雷达后向散射和LAI关系的首要因素,其次,不同的入射角对后向散射也具有明显的差异。基于上述分析,通过神经网络算法,利用不同时相、不同入射角的ENVISAT/ASAR雷达影像对白杨树林LAI进行了反演研究,对验证样本、训练样本、所有样本实测值与预测值进行了比较验证,其决定系数R2分别为0.61、0.91和0.82,表明基于ENVISAT/ASAR雷达数据利用神经网络算法反演人工林叶面积指数的可行性。
Based on ENVISAT / ASAR data, artificial neural network of Heihe in Northwest China, the leaf area index of poplar forest was studied. First, the relationship between the leaf area index (LAI) and the backward scattering coefficient of ENVISAT / ASAR under different polarizations was analyzed. The results show that the spatial distribution uniformity of plantations is the relationship between LAI and LAI Secondly, different angles of incidence also have significant differences in backscattering. Based on the above analysis, the LAI of poplar woodland was retrieved by using the ENVISAT / ASAR radar images of different phases and angles of incidence using the neural network algorithm. The validated samples, the training samples, and the measured and predicted values of all the samples were compared The coefficients of determination R2 were 0.61, 0.91 and 0.82, respectively, indicating the feasibility of retrieving the leaf area index of plantations by neural network based on ENVISAT / ASAR radar data.