Laboratory-Scale Model Simulations of Heterogeneous Flow Using Hydraulic Property Clusters.

Flow heterogeneity can be attributed to differences in hydraulic properties within the soil. The objective of this work was to investigate the effects of the location and number of representative hydraulic properties on the spatial distribution of flow in soil. A multi-step outflow experiment, performed with simulated rainfall on a large undisturbed soil column over a 41-day period, was monitored with time-lapse Electrical Resistance Tomography (ERT). Outflow over the lower boundary of the soil column (755 cm2) was collected in 151 spatially-defined outflow cells. Thirty soil samples of equal volume were taken from the column after the outflow experiment. Water retention properties of the samples were calculated through inversion in the HYDRUS-1D program using tension data obtained from evaporation experiments supplemented with measured retention points. Flow through the column was modeled with HYDRUS-3D with a domain corresponding to the dimensions of the column. HYDRUS-3D was run first using homogeneous materials, uniquely defined by the hydraulic properties of the 30 samples. Simulated outflow values were grouped as high or low flow rates to assign hydraulic properties to binary clusters defined by ERT data. K-means was used to define three clusters based on simulated outflow and the hydraulic properties of the samples. HYDRUS-3D was run with the two types of spatially defined clusters and 19 subregions located at the lower boundary of the domain using water balance calculations for outflow.  Simulated and measured outflow at each subregion were compared to validate the model and assess the effect of the number of hydraulic property clusters and positioning on flow.