A Monte Carlo backscatter model for a layer of a mixture of discrete scatterers

Abstract

This paper presents a Monte Carlo model for radar backscatter from a layer consisting of a mixture of randomly orientated discrete scatterers. The random layer is bounded on top by a Kirchoff rough surface using the scalar approximation, and bounded at the bottom by a plane lossy half-space. The discrete scatterers are modelled as spheroids and ellipsoids. The multiple-scattering processes are considered to be chains of collisions between the photons in the incident beam and the discrete scatterers. An algorithm is developed to track the trajectory of each photon and to collect the intensity of the backscattered photon flux, from which the radar cross-sections are calculated using the ensemble averages of the backscattered power. The technique so developed is used to study the backscatter from wet snow modelled as a three-phase mixture of elliptical scatterers (representing the water inclusions), spherical scatterers (representing the ice particles) and a background of air. A comparison is made with the results from the two-phase radiative transfer theory. It is found that our three-phase Monte Carlo model predicts higher levelsfor both the co- and cross-polarized returns and appears to give better agreement with experimental data. This is apparently due to modelling the water inclusions in the snow as a separate scatterer type, rather than combining the water and air as an average background medium. Comparisons are also made with measured data obtained by Stiles and Ulaby. They show good agreement, both in trends and levels, for co-polarized backscatter. Computations for cross-polarized backscatter are also presented, but no experimental data arc available for comparison.