Simulation study of a robust algorithm for soil moisture and surface roughness estimation using L-band radar backscatter

Various inversion algorithms have been developed to obtain estimates of soil moisture and surface roughness parameters from multifrequency, multiangle, and multipolarization radar reflectances. Since the penetration depth for radar signals increases with wavelength, an inversion algorithm using widely separated frequencies does not yield comparable probing depths. Furthermore, existing algorithms assume a linear relationship between the radar backscatter coefficient (in dB) and soil parameters, such as the volumetric soil moisture, soil surface roughness and surface slope. This assumption is valid only over a narrow range of soil parameters, thereby restricting its operational use under realistic conditions. Our research specifically explored the use of inversion algorithms based on L-Band radar reflectances at 1 GHz and 2 GHz frequencies in order to retain relatively consistent probing depths. In order to extend the range of applicability, a non-linear exponential-type relationship was developed between radar reflectance at a specified frequency, polarization and incidence angle combination, and soil parameters of interest, viz., soil moisture, surface roughness, and surface slope. An over-constrained inversion algorithm using a six-parameter combination was found to yield relatively accurate estimates of soil parameters over a wide range of soil conditions even in the presence of system error.