基于CHRIS/PROBA的植被叶面积指数估算模型研究

摘要

The ESA-mission CHRIS-PROBA (Compact High Resolution Imaging Spectrometer onboard the Project for On-board Autonomy) was used for providing space borne imaging spectrometer and multiangular data to assess the LAI. Five spectral vegetation indices (VI) were derived from CHRIS-PROBA image, including normalized difference vegetation index (NDVI), perpendicular vegetation index (PVI), modified soil adjusted vegetation index (MSAVI), ratio vegetation index (RVI), atmospheric resistance vegetation index(ARVI). Three hundreds LAI-VI correlation models were established. The VI-LAI correlation coefficients varied greatly across vegetation, vegetation indices, as well as image angular. In all models, from the perspective of angular, the best model is 0° image,R2 =0. 591 ,RMSE =0. 650, the worst model is -55° image,R2 = 0. 551 ,RMSE =0.821, from the perspective vegetation types, the best model is coniferous forest, followed by the broadleaf forests, shrubs, coniferous forests and grasslands, from the types of vegetation model, exponential model is better than one regression model, from the perspective vegetation index, the best model is PVI, followed by MSAVI, NDVI, RVI, ARVI.%选用江西省余干县多角度高光谱遥感数据CHRIS/PROBA,提取了5种植被指数(VI),即归一化植被指数(NDVI)、垂直植被指数(PVI)、调整土壤植被指数(MSAVI)、比值植被指数(RVI)、大气阻抗植被指数(ARVI),与地面实测的植被叶面积指数进行了回归分析,建立300个LAI-VI关系模型.结果表明:在所有的模型中,从5个角度来看,0.提取叶面积指数效果最好,R2=0.591,RMSE =0.650;-55.提取叶面积指数效果最差,R2 =0.551,RMSE=0.821；从植被类型来看,针阔林最好,其次为阔叶林、灌木、针叶林和草地；从植被模型种类来看,指数模型好于一次回归模型；从植被指数来看,PVI最好,其次为MSA VI、NDVI、RVI、ARVI.在LAI-VI关系建模过程中,基于多角度高光谱遥感数据提取植被指数,有利于充分挖掘遥感影像信息,能够提高LAI估算精度.