Forward and inverse dual-permeability model simulation of water flow in a soil column containing a preferential flow path

It is not clear if hydraulic parameters of dual-permeability models (DPM) can be properly identified by inverse analysis of preferential water flow data. In this study, we applied a DPM based on two coupled Richards' equations to compare inverse and forward simulations of preferential water flow observations obtained in well-controlled laboratory column experiments. Infiltration and drainage experiments were conducted using a repacked loam soil column (80 cm long, 24 cm diameter) containing a cylindrical sand region (2.4 cm diameter) as preferential flow path (PFP) along its central axis. The forward DPM simulations relied on fixed hydraulic parameters for both the soil matrix and the PFP, as determined by means of separate infiltration and drainage experiments performed on loam columns and sand columns, respectively. The parameters of the first-order water exchange term were derived based on the column geometry. Two different types of data were utilized for the inverse parameter identification. The first inverse approach relied on observations of accumulated infiltration and outflow, along with water contents and pressure heads in the loam matrix. The second inverse approach was identical except for including region-specific outflow out of matrix and PFP. The results showed that individual outflow out of matrix and PFP could not be properly described when fitting the DPM to bulk-soil related flow data, as opposed to the inverse DPM approach based on region-specific outflow data which reproduced those data well. On the other hand, bulk soil related infiltration and outflow was matched by both inverse approaches. The forward approach in most cases produced satisfactory results for both bulk soil related and region-specific outflow. For natural soils where region-specific flow data are not available, the soil hydraulic DPM parameters can not be inversely identified to correctly describe region-specific flow. This prohibits the subsequent use of the DPM for solute transport analysis. Future work may address the question if DPM hydraulic parameters can be identified when simultaneously using hydraulic and solute transport observations.