Multi-sensor high spatial resolution leaf area index estimation by combining surface reflectance with vegetation indices for highly heterogeneous regions: A case study of the Chishui River Basin in southwest China
Duo Han, Hong Cai, Lei Zhang, Yiting Wen
Abstract
The leaf area index (LAI) is an indispensable parameter in vegetation canopy research. In highly heterogeneous regions, a low spatial resolution LAI often fails to reveal dynamic changes in vegetation accurately. Most studies estimating high-resolution LAI from global products have focused on flat regions and have only established the relationship between training samples and band reflectance. The estimation result is easily affected by differences in the sensor bandwidth and has low universality among different sensors. This study selected high-quality LAI samples from moderate resolution imaging spectroradiometer (MODIS) products, combined band reflectance and vegetation indices (VIs), and used Random Forest to estimate the 30-m resolution LAI in the upper reaches of the Chishui River Basin in China. After adding VIs, the consistency between the LAI estimation results and the synchronous LAI estimated by the Sentinel-2 Land bio-physical processor (SL2P) improved compared with using only band reflectance. The coefficient of determination (R2) values of Landsat-8 and GF-1 increased by 0.04 and 0.18, and the root mean square deviation (RMSE) values decreased by 0.07 and 0.13; the Sentinel-2 A R2 values increased by 0.09 and 0.24 in two phases, while the RMSE values decreased by 0.07 and 0.22. After terrain correction, the R2 value between the Landsat-8 estimation and SL2P LAI increased by 0.06, whereas the RMSE value decreased by 0.07. The results provide high-resolution LAI data and estimation methods for vegetation research in highly heterogeneous regions and improve the versatility of using MODIS products to estimate high-resolution LAI using different sensors.