A radar backscatter model for forest stands

Abstract

A Monte Carlo method for calculating the radar backscatter cross-sections from a forest stand is presented in this study. The model for the forest stand consists of two layers: the top layer contains a mixture of randomly orientated dielectric circular disks or small needles representing the leaves of the deciduous trees or coniferous trees respectively, as well as cylindrical-shaped scatterers, representing the branches of the foliage, while the bottom layer consists of randomly distributed vertical cylinders modelling the tree trunks. The phase matrices of the disks and needles are formulated using the generalized Rayleigh–Gans approximation whereas those of the branches and tree trunks are obtained from the infinite-cylinder approximation. The ground is modelled as a Kirchhoff rough surface based on the stationary-phase approximation. The relative contributions from the tree trunks, branches and trunks, leaves and trunks, and the entire forest stand are studied separately. It is found that leaves seem to dominate at higher frequencies while branches at lower frequencies. Effects of various physical and geometrical parameters such as dielectric constants, volume fractions, sizes and orientation distributions of the scatterers in a forested canopy are also studied. Finally, the results from the model are compared with the measured angular distributions of the radar backscattering cross-sections from some field measurements reported in the literature.