The monitoring of tropical ecosystems is complex due to the type of terrain and adverse weather conditions present during most of the year. This paper proposes the development and implementation of a methodology based on RADARSAT-2 images and a semi-empirical model to obtain surface parameters in a tropical forest. The model takes into consideration the sensor configuration, and vegetation and soil parameters to represent the behavior of the wave in the scene. To retrieve the surface parameters, the model is implemented in an optimization framework using all polarizations (HH, HV, VH and VV). Finally, as outputs of the optimization process, we obtain the trunk diameter (Dt), the crow height (hc), and the optical penetration depth (τ). By using τ, the vegetation water content (VWC) is estimated. The accuracy of this methodology is about 83% when compared to ground data.