Abstract: This research presents a forward modeling approach for target scattering centers on a dielectric rough surface, which is directed to the investigation of the electromagnetic scattering for typical ground backgrounds. Initially, a rough surface is randomly generated by the Monte Carlo method, and a three-dimensional rough ground model is simulated using a Gaussian spectrum with different ground materials characterized by different permittivity. Subsequently, a forward approach is employed, and ray tracing and diversity techniques are used to decompose the scattering into a number of sources, which including the complex ground environment and BMP2 infantry fighting vehicle target from the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset. Then, the solution is analyzed by considering different permittivity of rough surfaces and solving the problem using the high-frequency approximation. Finally, the reconstructed synthetic aperture radar (SAR) images and the MSTAR dataset are compared, the result shows that the proposed method improves image similarity, and confirming the efficiency of the proposed forward modeling method for target scattering centers on a dielectric rough surface.