A fully polarimetric multiple scattering model for agricultural fields

A discrete model has been developed at Tor Vergata University, which describes vegetation as an ensemble of discs and cylinders overlying a homogeneous half-space with a rough interface. The model predicts the backscatter coefficient including multiple scattering effects. The first version, which was based on a scalar approach (2 Stokes components) has now been improved in order to get a fully polarimetric approach (4 Stokes components). Soil scattering is represented, for each pair of incidence and scattering directions, by a 4×4 scatter matrix, computed by means of the small perturbation approximation (for smooth soils) and the geometrical optics approximation (for rough soils). The vegetation layer is subdivided into many thin layers, each filled with discs and/or cylinders. The scattering and transmission properties of a thin layer are represented, for each pair of incidence and scattering directions, by two 4×4 scatter matrices (for upper and lower half-spaces respectively) and, for each incidence direction, by a 4×4 extinction matrix. Matrices are computed by means of Rayleigh-Gans or physical optics approximations, depending on frequency. Finally, the contributions of the various thin layers and that of the soil are combined by means of the matrix doubling algorithm. In this paper, the authors´ recent modeling advances are described. In particular, the differences between the scalar approach and the polarimetric approach are emphasized. The polarimetric model results are compared with experimental radar data obtained over agricultural fields of the Italian Montespertoli site during the MAC-91 Campaign. In particular, the crop biomass effects are investigated at L-band in the case of sunflowers. Some polarimetric features, like backscatter coefficients in linear and circular polarizations, and phase difference between VV and HH channels are considered