Inversion of soil moisture with radar backscattering data

The influence of surface roughness on the radar backscatter has been studied using several existing theoretical approaches. The simulation results showed a very strong sensitivity of the radar backscattering coefficients on the variations of the parameters characterizing the surface roughness. In order to assess the use of multifrequency or multipolarization data for soil moisture inversion without full knowledge of the surface roughness parameters, neural networks are applied to simulated data. The copolarised multifrequency data have been simulated using the IEM model and the multipolarized data (co and crosspolarization) have been simulated using a polarimetric model the authors developed under the small perturbation assumption