At microscopic scale, soil pores constitute microbial habitats with changing physicochemical settings, where cells and organic residues are heterogeneously distributed. Soil Organic Matter (SOM) decay is thus controlled not only by the activity of the micr obial communities but also by abiotic processes such as the diffusion of extracellular enzymes and dissolved organic carbon (DOC) in hydrologically connected soil pore space. We present a carbon decomposition model (inspired by Schimel and Weintraub, 200 3) where the processes are spatialized in the soil structure. Decomposition is seen as sequential with a first stage for particulate organic matter (POM) hydrolysis and a second one for DOC assimilation by bacteria. DOC and enzymes may enter adsorbed pools and hence be temporarily preserved from degradation. The biological model is coupled to the TRT - Lattice Boltzmann model (Ginzburg, 2005). The latter applies statistical physics principles that simulate collisions and propagation of fluid particles using a discrete method. It reproduces water flows and solutes diffusion at the pore scale. This approach enables to take into account easily the complex 3D geometry of soil pore space and the heterogeneous distribution of the water/air, POM and bacteria within the pores. 3D images of the soil pore space are obtained from X - Ray computing microtomography scans of soil samples. A parameters sensitivity analysis of the coupled model will finally allow us to estimate soil structure impact o n biological, physical an d physicochemical mechanisms controlling SOM decomposition.