Correlating pore space morphology with numerically computed soil gas diffusion for structured loam and sand, including stochastic 3D microstructure modeling

Biogeochemical soil processes are closely linked to the structure of soil. In particular, nutrient transport depends on diffusivity and permeability within the soil's pore network. A deeper understanding of the relationship between microscopic soil structure and such effective macroscopic properties can be obtained by tomographic imaging combined with a quantitative analysis of soil morphology and numerical simulations of effective macroscopic properties. In a previous work it has been shown that different parametric regression formulas can be used to predict these relations for finely sieved soils of loam and sand. In the present paper, we validate these formulas and further extend their applicability to structured soils. In particular, 3D CT data of a total of six samples, consisting of three loam and three sand samples, are used as the basis for an extensive structural analysis. As expected, the performance of most regression formulas can be improved by specifically adjusting their parameters for the considered soil structures. However, it turns out that some regression formulas based on, e.g., tortuosity which were fitted for finely sieved soils still reliably predict diffusion for structured soils without adjusting their parameters. For additional validation and improvement of the considered regression formulas, artificially generated soil structures can be utilized. Therefore, a method for the generation of such structures via samples drawn from a parametric stochastic 3D microstructure model is outlined which may serve as a basis for further work.