一种组合反演叶面积指数的方法

A method of combining the inversion of leaf area index

针对传统PRO4SAIL+查找表方法反演叶面积指数存在查找表过于庞大,反演速度较慢等问题,该文提出一种基于PRO4SAIL与局部加权多元回归组合模型反演叶面积指数的方法。通过利用卫星传感器光谱响应函数实现了实测端元高光谱向像元多光谱的转化,解决了测量尺度不同导致的反射率差异问题;选取两种叶面积指数植被指数MTVI1和MCARI1作为反演因子,同时只选用40组PRO4SAIL模型模拟数据建立训练组,解决查找表数据量过大的问题;将局部加权多元回归的权重因子距离公式按照反演因子个数从一维空间扩展至多维空间,更符合实际应用。该组合模型的预测决定系数为0.727 1,平均相对误差为11.09%,传统查找表的预测决定系数为0.693 2,平均相对误差为13.63%。实验结果表明:组合模型具有较好的预测能力,反演得到的叶面积指数含量精度较高,可为更好地监测路域植被生态环境提供技术支撑。

Due to the disadvantages of the traditional PRO4 SAIL+look-up table method inversion of leaf area index,such as the large size of the look-up table and slow retrieval speed,a method based on combined model of PRO4 SAIL and local weighted multi-variable regression to invert the leaf area index was proposed.By using the spectral response function of the satellite sensor,the conversion of measured endmember hyper-spectral to pixel multi-spectral is realized,and the problem of the difference in reflectivity caused by different measurement scales is solved;Two vegetation index of leaf area index of MTVI1 and MCARI1 were selected as inversion factors,and only 40 sets of PRO4 SAIL model simulation data were used to establish a training group,which solve the problem of excessive data in the look-up table;The weighted distance formula of local weighted multi-variable regression expands from one-dimensional space to multidimensional space according to the number of inversion factors,which is more in line with practical application.The predictive coefficient of the combination model is 0.730 3,and the average relative error is 11.95%,the prediction coefficient of the traditional look-up table is 0.683 9,and the average relative error is 14.93%.The experimental results show that:the combined model has good predictive ability,and the accuracy of the leaf area index obtained by inversion is higher,which provides technical support for better monitoring of the road vegetation ecological environment.