Comparing accuracy for Leaf Area Index estimation inverting a simple empirical model and a Radiative Transfer model by using multiangular and hyperspectral data

摘要

The Leaf Area Index is a key parameter that is indispensable for many biophysical and climatic models. LAI is required for modeling crop water requirements for precision farming and agricultural resource management. The objective of this study was to investigate different approaches for estimating LAI from EO data. To this aim multiangular CHRIS/PROBA data, from SPARC 2003 and 2004, were used in the inversion of PROSPECT-SAILH models using a numerical optimization technique based on Marquardt-Levenberg algorithm. The optimal spectral sampling to estimate LAI was investigated using a sensitivity analysis. From the same data set, the reflectance in the red and near-infrared bands, from the closer to nadir image, was considered in order to estimate the LAI using an empirical approach based on the CLAIR model. The LAI obtained from the empirical approach was finally employed as prior information in the physical based model. LAI values retrieved with the combined approaches were realistically estimated with a good accuracy (RMSE is 0.51 m2m-2).