The relation between X-ray-derived measures of soil structure and dual-permeability model parameter values

One dimensional dual-permeability models are used to simulate preferential flow and transport in soils. In these models, soil structure, including large biopores and cracks, is accounted for in a simple way by splitting the total porosity into a micropore and a macropore domain with first-order equations governing the transfer of water and solutes between the domains. The transfer equations are central in dual-permeability models because the values of the included parameters have a strong influence on the degree of simulated preferential flow and transport. These transfer terms include a parameter describing the characteristic length of the soil matrix structure. This characteristic length is well defined for idealised pore geometries. However, such idealised pore geometries are poor representations of macropore networks in intact soil. Our objective was to test if values of parameters governing the degree of preferential transport in dual-permeability models could be estimated from measures of soil structure derived from X-ray tomography images. To achieve this, we calibrated the dual-permeability model MACRO against non-reactive solute breakthrough curves obtained at two flow rates from 33 intact soil columns sampled from a field with large variation in soil properties. Relations between measures of soil structure derived from images of the same columns and values for parameters governing preferential transport were then evaluated. The MACRO model could reproduce all BTCs well except those for three sandy soils. When the saturated water content of the soil matrix, here used to account for possible water repellency, was included in the calibration also the BTCs for the sandy soils were well reproduced. Preliminary results indicate that the fractal dimension of the total imaged pore network is the strongest predictor for the characteristic length of the soil matrix. The other included model parameters were not strongly correlated with any measures of the total imaged pore network. We will also present results for the imaged percolating macropore networks (i.e. the parts of the pore networks connected to both the top and bottom of the imaged region of interest), which is a better representation of the pore network that was active during the transport experiments.