Soil moisture retrieval is one of the most important aspects of Earth science observation where it can be used to predict various climate events. Several empirical and theoretical approaches have been proposed in the past for soil moisture retrieval. We propose a novel soil moisture retrieval algorithm based on electromagnetic scattering model. Being a physics-based model, our sliced regression inversion algorithm works for a wide range of geophysical parameters without any dependence on previously acquired datasets. Our SRI technique builds on and improves the look-up table approach by approximating the relationship between radar backscattering coefficients and the underlying physical variables by piece-wise linear fits.