Soil Moisture Retrieval over Agricultural Fields Using Dual-Polarimetric SAR Data

The model-based polarimetric decomposition technique can utilize information of different channels to describe the complex interaction processes between soil and canopy, elegantly achieving the decoupling of scattering signals from the surface and vegetation. However, the dependence on fully polarimetric SAR data limits the application of this method. In this study, an advanced soil moisture retrieval method coupling model-based decomposition and surface scattering model is proposed for dual-pol SAR data. The generalized volume scattering model based on fully polarimetric decomposition theory is reconstructed as a projection on Stokes vector to facilitate the simple removal of volume scattering contribution. Soil moisture is subsequently estimated iteratively based on a cost function using an Oh semi-empirical model considering surface roughness. The measurements obtained from the ground campaign in Manitoba, Canada and L-band UAVSAR images collected during the campaign are used for validation. The proposed method achieves accurate soil moisture estimation based on reasonable separation of surface and vegetation scattering signals. The root mean square error (RMSE) of soil moisture inversion for the VV-VH mode reached 0.052 m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>/m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> with a correlation coefficient of 0.82, while the RMSE for the HH-HV mode was 0.073 m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>/m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> with a correlation coefficient of 0.68.