Simulating Diffusive and Preferential Water Flow in Soils with a Coupled Source-Responsive/Richards-Equation Model

Abstract:

Water movement through soils is often dominated by preferential flow processes such as fingering and macropore flow. Traditional models of flow in the unsaturated zone are based on the diffusion or Richards equation and assume that diffusive (surface-tension viscous) flow is the only flow process. These models are incapable of accurately simulating preferential flow. Several alternative approaches, including kinematic wave, transfer function, and water-content wave models, have been suggested for simulating water movement through preferential flow paths. The source-responsive model proposed by Nimmo (2010) and Nimmo and Mitchell (2013) is unique among such models in that water transfer from land surface to depth is controlled by the water-application rate at land surface. The source-responsive model has been coupled with a one-dimensional version of the Richards-equation based model of variably saturated flow, VS2DT. The new model, can simulate flow within the preferential (S) domain alone, within the diffuse (D) domain alone, or within both the S and D domains simultaneously. Water exchange between the two domains is treated as a first-order diffusive process, with the exchange coefficient being a function of soil-water content. The new model was used to simulate field and laboratory infiltration experiments described in the literature. Simulations were calibrated against measured soil water contents with the PEST parameter estimation package; values for hydraulic conductivity and 3 van Genuchten and 3 source-responsive parameters were optimized. Although exact matches between measured and simulated water contents were not obtained, the simulation results captured the salient characteristics of the published data sets, including features typical of preferential as well as diffusive flow. Results obtained from simulating flow simultaneously in both the S and D domain provided better matches to measured data than results obtained from simulating flow independently in either the S or the D domain, indicating the importance played by flow in each of the domains. The coupled model can be used to investigate the contributions of S- and D-domain flows to infiltration and subsequent redistribution of subsurface water and to identify the conditions that promote preferential or diffuse flow.