ABSTRACT Synthetic Aperture Radar (SAR) is a potent instrument for estimating soil moisture across vast farmland areas. However, when soil salinity rises, existing SAR retrieval models for soil moisture become less accurate. Soluble salts in soil water alter the inherent correlation between the SAR backscattering coefficients and soil water, thus compromising the performance of retrieval models. The need to get surface roughness parameters from the field also limits the widespread application of the model. To address these issues, we developed a retrieval model for salty soil moisture assessment under small-scale roughness surfaces. First, we found a method to distinguish between slightly rough and rough surfaces using scattering entropy obtained from full-polarization decomposition. Then, we harnessed the co-polarization ratio to mitigate the effect of surface roughness in slightly rough conditions. Later, we developed a soil moisture retrieval model based on the co-polarization ratio, considering incidence angle, residual roughness, and soluble salt content. The validation of the model was verified using field surveying data, yielding an RMSE of 0.026 cm3 ·cm− 3 for the estimated soil moisture. This accuracy is comparable to or surpasses the level achieved by SAR for non-salinized soil moisture retrieval. Our findings showed that Scattering entropy is an effective parameter for distinguishing scattering mechanisms on farmland surfaces effectively. Specifically, when the scattering entropy of farmland is less than 0.5, the co-polarization ratio can mitigate some surface roughness effects. Additionally, when the soil-soluble salt content is incorporated, our model can accurately estimate the soil moisture of salty soil.
Read full abstract