The main objective of the SoDA (Soil Degradation Assessment) project is to realize a simulator of soil surface degradation by rainfall at the meter scale and including visualization. Soil surface structure and morphology deeply influence a lot of processes of high agronomic and environmental relevance, such as mass and heat transfer through the soil-atmosphere interface, runoff and erosion, seed germination and seedling emergence. The soil surface structure of agricultural field is in continuous evolution: it is strongly affected by tillage, and in between tillage operations, erosion by rainfall and runoff causes a progressive degradation of the structure whose intensity and speed partly depend on the initial state associated to tillage modalities. A soil surface degradation model could allow one to predict this evolution of the soil surface structure, and even to help choosing adequate tillage practices and sowing dates. Erosion modeling has been addressed by soil scientists but also by computer graphic scientists in order to add realism to virtual landscapes. Mixing both of these points of view would be interesting to simulate and visualize the evolution of the soil surface of a cultivated soil. Based on a 3D cellular automata approach using the knowledge accumulated by soil scientists about the physical processes involved in erosion, the principles of our simulator and its first implementation are presented in this paper.