A Genetic Algorithm for the Retrieval of Soil Multi-scale Roughness and Moisture Parameters

This paper investigates applications of neural network (NN) and genetic algorithms (GAs) to invert geophysical parameters from synthetic radar data. The GA belongs to the family of stochastic approaches for optimization. They are known to explore large domains in short periods of time providing fairly good estimates of the solution. NN and GA have already been applied in various geophysical problems. Extracting soil moisture and roughness parameters of natural surfaces from this data has been problematic for many reasons and many researchers have encountered many problems like the lack of information about the characteristics of natural surface roughness. In addition, the relation-ship between the backscattering coefficients is non-linear and the problem of retrieving parameters is frequently ill-posed and it may be impossible to separate the contributions from different mechanisms making the retrieval of several parameters simultaneously necessary. In this work a multi-layered modified 2D Multi Scale Small Perturbation Model (MLS SPM 2D) including both the surface scattering and the volumetric scattering, within the soil has been used to calculate the total surface reflection coefficients. Based on the GA, a fitting algorithm has been derived from parameterization of the SPM model for a wide range of soil dielectric constant, incidence angle, RMS height, and multi scales fractals parameters for each layer. Samples were generated with the original MLS 2D SPM model followed by a neural network to obtain the statistic soil moisture and multi scale roughness parameters algorithm. The soil moisture and MLS roughness parameters optimized with the GA inversion algorithm agree very well for VV-copolarisation and HH-copolarisation.