Influence of Different Bandwidths on LAI Estimation Using Vegetation Indices

Abstract

Leaf area index (LAI) is a valuable indicator used in vegetation growth monitoring. To optimize the index selection according to the type of remote sensing data and to improve the inversion accuracy of LAI, this article analyzes the influence of different bandwidths on the accuracy of the inversion model based on vegetation indices. First, the simulation dataset is generated by the PROSAIL model, and on this basis, 15 vegetation indices with high correlation coefficients with LAI. Then, by analyzing the sensitivity of these 15 indices to the variations in bandwidth, and to the coefficient of determination (R2) of the LAI inversion model with the variations of bandwidth, the influence of different bandwidths on the accuracy of LAI estimation by each index is determined. The results show that bandwidth is one of the most important factors in determining the accuracy of LAI inversion, and the influence on different vegetation indices can be divided into the following three categories. First, narrowband vegetation index, the accuracy of inversion models built by vegetation indices decreases with the increase of bandwidth, including SR[800,680], OSAVI, MTVI2, SR[752,690], RDVI, NDCI, and NVI. Second, middleband vegetation index, the accuracy first increases and then decreases with the increase of bandwidth, including SR[700,670], Carte5, and SR[675,700]. Third, broadband vegetation index, the accuracy increases with the increase of bandwidth, including SPVI, Carte2, OSAVI2, MTVI1, and NDVI705. The study provides a scientific basis for vegetation index optimization in the process of LAI inversion.