Presented at the 2012 International Geoscience and Remote Sensing SymposiumAn Observing System Simulation Experiment (OSSE) for the Aquarius/SAC-D mission has been developed for assessing the accuracy of soil moisture retrieval from passive and active L band. So far, this OSSE has been successfully exploited to study the artifacts in the retrieved soil moisture associated to: (1) uncertainties and aggregation of the ancillary parameters needed for the retrieval and (2) instrumental noise effects. However, effects due to forward and retrieval model incompatibilities have not yet been studied. In this paper, OSSE attempts to capture the influence of this effect over estimated soil moisture. The emissivity of real surfaces is very complex and is strongly dependent on land cover type and condition. In particular, surface covered by average to dense vegetation presents complex scattering properties, heavily related to canopy structure. The OSSE implements a forward model using a theoretical approach based on the electromagnetic modeling of vegetation elements and high order radiative transfer theory. In this way, the difficulties related to retrieving soil moisture from passive data with a simple model are studied. The accuracy of the soil moisture estimation is analyzed on a set of selected footprints in order to illustrate the impact of discrepancies between both models. In general, retrieved soil moisture performs worse over dense vegetated areas and under wet conditions. Furthermore, accuracy is highly dependent on land cover.