In this paper a combined Ground Penetrating Radar (GPR) and Synthetic Aperture Radar (SAR) technique is introduced, which considers the soil surface refraction and the wave propagation in the ground. By using Fermat's principle and the Sober operator, the SAR image of the GPR data is optimized, whereas the soil's permittivity is estimated. The theoretical approach is discussed thoroughly and measurements that were carried out on a test sand box verify the proposed technique.