The Effect of Variable Soil Moisture Profiles on P-Band Backscatter

Radar measurements at P-band are sensitive to profile soil moisture. Associated backscatter measurements depend on the distribution and variation of the soil moisture profile. Existing scattering models account for this variation by approximating the soil moisture profile as consisting of a number of homogeneous layers. Since the inversion of the scattering models during the retrieval process can be based on only a few polarimetric backscatter measurements, the number of obtainable independent layers in the profile representation is limited. The purpose of this paper is to gain insights into the effects of the layering representation on the resulting modeled forward scattering. These insights form the rational basis for the design of retrieval algorithms. The effects of reflections between layers and other sources of error on simulated backscattering coefficients are first illustrated using several case studies. To determine the combined effect of different error sources for realistic soil moisture profiles, ten years of conditions at a grassland in California are studied. Depending on the layering strategy and the polarization, the root-mean-square error (RMSE) of backscattering coefficients due to misrepresenting the profile alone can be up to 2 dB, although errors can be up to 10 dB in particular cases. The error generally decreases as additional layers are added. The HH-polarization is more sensitive to the subsurface than the VV-polarization and has greater errors. Using a profile-dependent layer placement strategy decreases the RMSE of the backscatter simulation by less than 1 dB relative to a strategy with fixed layering.