Comparison of methods to estimate LAI from remote sensing in maize canopies by inversion of 1-D and 3-D models

Abstract

The inversion of canopy reflectance models is a procedure widely employed for the retrieval of LAI from remote sensing. However the accuracy of the estimates depends on a range of factors, most notably the realism of the canopy representation of the models employed and the a priori knowledge of some vegetation characteristics. The objective of the present work was to compare the performances and the operational limitations of the use of two contrasting type of radiative transfer models: a classical one-dimensional canopy reflectance model, PROSPECT+SAIL (PROSAIL) and a three-dimensional dynamic maize model (Lopez-Lozano et al., 2007) having greater realism in the description of the canopy structure. The assessment was based on data gathered in dedicated field experiments. The inversion of the two models was carried out by using the same technique, the simplex numerical optimisation algorithm. The results revealed that although the 3-D model inversion was slower as compared to that of PROSAIL, it provided slightly better estimates of LAI, because of the direct use of gap fraction data and possibly the implicit injection into the inversion process of a priori information embedded into the 3-D maize model.