C-band retrieved soil moisture is known as uncertain, due to volume scattering and complicated effect of roughness or meteorological disturbance. Accordingly, despite several merits of remote sensing, application of satellite observation was limited to bare soils or homogenous terrain, instead of estimating high resolution roughness in heterogeneous areas. The accuracy, temporal and spatial resolution of the remotely sensed soil moisture estimates is often insufficient for the use in hydrological modelling (Brocca et al., 2010). In this context, this study suggests a stochastic approach to consider satellite measurements as probability distribution rather than a deterministic single vale. Such stochastic retrievals were used to �relax� the estimation of physical parameters inverted with a forward model of Advanced Integral Equation Model (AIEM). From the results, we demonstrated that it was possible to mitigate non-linear errors in retrievals arising from rainfall, and vegetation growth. Spatial distribution analysis also showed that a relationship between soil moisture and backscatter coefficients was reasonably good. Hydro-meteorological data showed that SAR soil moisture well deciphers streamflow hysteresis. These results imply that soil moisture derived from the RADARSAT-2 backscatters can be reasonably employed for river discharge forecasting (Hirpa et al., 2013).