Forest remote sensing: Inversion using a coherent scattering model and a genetic algorithm

In remote sensing for forest monitoring, Synthetic Aperture Radar (SAR) imaging is a powerful tool to investigate wide forested areas, particularly at low frequencies (less than 1GHz) where the electromagnetic waves penetrate deeply under the forest canopy. Radar measurements allow us to obtain information on the forest biomass, such as density, heights and dielectric properties of trees. In order to extract this information, it is necessary to analyze the wave-tree interactions involved when SAR measurements are performed. At low frequencies, the small branches and leaves of trees do not significantly contribute to the electromagnetic fields scattered by a forest area and can be safely neglected. The main contributions to the scattered field come from the trunks and primary branches. In this context, we have built up an approximate model where the latter are considered as finite circular cylinders. The total scattered field is then obtained by coherently adding the contributions of the different trunks and branches and the contribution of each of them is obtained through the coherent sum of various elementary scattering mechanisms such as single, double and triple bounces.