Estimation of Sahelian-grassland parameters using a coherent scattering model and a genetic algorithm

In this paper, the applicability of a procedure for retrieval of vegetation parameters using a coherent scattering model that considers the botanical properties of Sahelian grassland and a stochastic optimization algorithm is studied. This African vegetation is mainly composed of shrubs and grass. Since the coherent scattering model is computationally time-consuming, a simplified empirical model is constructed by fitting of simulation results obtained by the scattering model. Inputs to the empirical model are the sensitive parameters that, for the studied class of vegetation, are the soil moisture content, grass density, and grass moisture content. The model outputs are the polarimetric backscattering coefficients as a function of the incidence angle. Employing the empirical model and a genetic algorithm, a search routine is implemented to estimate the biophysical parameters of the African vegetation from a data set of backscattering coefficients. The estimation of Sahelian-grassland parameters using the set of C-band HH-polarized measured data shows that this procedure achieves good agreement with the ground-truth data.