Assessment and Validation of MODIS and GEOV1 LAI With Ground-Measured Data and an Analysis of the Effect of Residential Area in Mixed Pixel

Abstract

Leaf area index (LAI) is a critical variable for simulating the carbon or nitrogen cycles and water and heat energy balance of ecosystem. MODIS and Geoland2 version 1 (GEOV1) LAI products were validated based on the ground-measured maize, winter wheat, and grass LAI data in several years. This study also investigated the residential area effect in mixed pixels on global LAI product accuracies in North Plain and Northeastern Plain in China. The MODIS and GEOV1 LAI products showed marked difference in variations of maize and winter wheat LAI at different key growth stages, and the GEOV1 LAI can present much clear differences and variations during crop growth periods. The MODIS and GEOV1 LAI products often underestimate the maize and winter wheat LAI, with the exception that GEOV1 LAI overestimate when maize LAI is large. For grass, the MODIS and GEOV1 LAI both overestimate a little. Overall, the GEOV1 LAI is often larger than the MODIS LAI. The GEOV1 LAI showed better regressions (with ${\rm{R}}^{2} $ of 0.868, 0.496, and 0.216) with the ground-measured LAI than MODIS LAI (with ${\rm{R}}^{2}$ of 0.258, 0.350, and 0.129) for maize, winter wheat, and grass, respectively. The residential area in mixed pixel make marked impact on MODIS and GEOV1 LAI data at different maize and winter wheat growth stages, and it maybe a main error source of the MODIS and GEOV1 LAI underestimations. The quadratic polynomial fitting relationships (most of the regressions ${\rm{R}}^{2} $ exceeded 0.90) can describe well the effect of residential area percent in mixed pixel on global LAI product.