Research On Retrieval And Spatial Scaling Of Leaf Area Index From Remote Sensing Images

Leaf area index (LAI) is defined as the sum of one-side leaf area on unit area. It is one of the basic parameters characterizing vegetation canopy structure, and its inversion result has important significance on earth’s ecological system research. However, the inversion values of LAI product exists a certain deviation, which is used to estimate vegetation LAI in a large area. Therefore, this paper was applied to establish a scaling model with simpleness, practicality and high accuracy based on the physical structure of the ground component in order to realize the fast, efficient remote sensing estimation of vegetation leaf area index in a large scale.LAI was extracted from Landsat8OLI images and MODIS images based on the method of pixel information decomposition and multiple scatterin model. According to the results of LAI extraction accuracy, the scaling model of LAI was established with MODIS and Landsat8OLI images. Meanwhile, through analyzing the difference of LAI estimation accuracy with multiscale MODIS images and scaling models before and after topographic correction, the effect of topographic correction on them was discussed. The results were showed as follows:(1) The method of pixel information decomposition for the extraction of vegetation coverage could better reflect the component percentage of basic components in the ground, so it had better effect of decomposition than the method of linear mixture model;(2) The correlation between LAI values of Landsat8OLI images based on pixel information decomposition and multiple scatterin model and observed LAI values was higher, and it was0.825, the effect of which was better. But the effect of LAI estimation with MODIS images was worse, and RMSE was22.47%, the estimation accuracy of which was lower. So the estimation results of MODIS LAI needed to be corrected;(3) The lower spatial resolution of multiscale MODIS images, the higher estimation accuracy of LAI, especially that RMSE drop after topographic correction was4.52%. It indicated that MODIS images were suitable for the inversion of leaf area index in a large area, and it was necessary to establish the scaling model between MODIS LAI and OLI LAI with the purpose of improving the LAI estimation accuracy of MODIS data. Meanwhile, topographic correction played an active role in LAI estimation;(4) The linear scaling model between MODIS LAI and Landsat8OLI LAI was built, and the fitting degrees of models after topographic correction increased from0.776to0.902with the loss of the spatial resolution. It illustrated that the scaling models could significantly improve the estimation accuracy of MODIS LAI, and the lower spatial resolution image, the better results of LAI estimation. Moreover, the fitting degrees of models after topographic correction were higher than those values before topographic correction, and the LAI differences between MODIS and OLI images were smaller than those values before topographic correction, which showed that topographic correction could effectively increase the scaling accuracy between OLI images and MODIS images, and it was one of the important factors in the scaling model.