Comparison of two inversion methods for retrieval of soil moisture and surface roughness from polarimetric radar observation of soil surfaces

This paper presents comparison of the direct inversion method (DIM) and the genetic algorithm-based inversion method (GAIM) for retrieval of soil moisture and surface roughness from polarimetric radar observation of soil surfaces. Those inversion methods are based on the polarimetric semiempirical model (PSEM), which was developed empirically to estimate the backscattering coefficients from the volumetric soil moisture content and the surface roughness parameters ks and kl, where k = 2pif/c, s is the rms height and l is the correlation length. The copolarized ratio p and the vh-polarized backscattering coefficient of the semiempirical model have been inverted directly to retrieve the soil parameters in the DIM. The cross-polarized ratio q can additionally be used to improve the inversion results in the DIM. A good agreement between the values of ks estimated by the inversion method and those measured in situ for ks<3.5, with a correlation coefficient of 0.822. The genetic algorithm (GA) is applied for an inversion method to retrieve both the volumetric soil moisture contents and the rms surface heights from multipolarized radar observations of bare soil surfaces. The cost function for each chromosome is evaluated using the semiempirical scattering model. Good agreement was found between the values of the rms height and the soil moisture content estimated by the inversion technique and those measured in situ. It was also found that the GAIM shows higher accuracy than the DIM in the estimation of the surface parameters from the radar observations