A coherent scattering formulation is developed for radar remote sensing of Sahelian grassland. This African vegetation is composed of shrubs and annual grass. The proposed model includes a vegetation generator tool in order to create vegetation structure with realistic architectures and botanical information. This is important in the development of the coherent scattering model, since the relative position of plant elements needs be preserved as accurately as possible. To correctly account for the coherent attenuation through the crown layer, the crown shape of shrubs must be considered. The crown shape is highly irregular, but for the most part can be encompassed in an ellipsoidal or cylindrical volume depending on the ground truth data. Thus, the extinction of the coherent wave is then calculated only when traveling within the crown volume. On the other hand, the grass generator models the grass as a set of cylindrical stalks and blade leaves arranged in a semi- deterministic fashion. Since grass blades are thin, multiple scattering among adjacent elements can be neglected at microwave frequencies. Backscatter statistics are acquired via a Monte Carlo simulation over a large number of realizations. Depending on the season, it is shown that contribution from soil and grass are the dominant components of the overall backscatter.