ARS | Publication request: ANALYSIS OF UNSATURATED WATER FLOW IN A LARGE SAND TANK

Submitted to: Soil Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: September 4, 2002
Publication Date: N/A

Interpretive Summary: Process-based numerical computer models are increasingly used to predict water flow and solute transport in the unsaturated zone between the soil surface and the water table. The use of such models presumes having reasonable estimates of the water retention (water holding) and hydraulic conductivity (permeability) properties of the soils. Previous studies have revealed the importance of how these properties are estimated for specific applications/soils. This study was initiated to acquire experimental data about the hydraulic properties of sandy soils to serve as a base for the numerical predictions. Specific objectives were to clarify the effects of (i) the invoked procedure for estimating the soil hydraulic parameters and (ii) using increasingly refined spatial definitions of the hydraulic properties on simulated two-dimensional water content and flow velocity distributions. Water flow in and drainage from a large sand tank (approximately 5 m x 3 m at the base, 6 m x 5.6 m at the top) was investigated using soil hydrologic and geophysical methods. The observed flow fields and water contents were analyzed using the HYDRUS-2D computer software model developed at the Salinity Laboratory. The observed variability in the drainage rate with time was reproduced best when an average water retention curve was used and the saturated water content was set equal to the porosity, whereas cumulative outflow was predicted best when all hydraulic parameters were fitted to the retention data. Using more refined estimates of the hydraulic parameters (e.g., assuming a layered profile) did not improve the predictions of the discharge rate as compared to assuming a homogeneous sand tank. This suggests that elaborate efforts to measure the hydraulic properties of all soil types in a layered or otherwise heterogeneous soil profile may not necessarily lead to better predictions of water flow and contaminant transport in the subsurface, and that sometime a reasonable estimate of the average properties may be sufficient.

Technical Abstract: A realistic, physically based simulation of water and solute movement in the unsaturated soil zone requires reasonable estimates of the water retention and unsaturated hydraulic conductivity functions. A variety of studies have revealed the importance of how these unsaturated soil parameters are assessed and subsequently distributed over the numerical mesh on modeling outcome. This study was initiated to acquire experimental data about the water flow characteristics of sandy soils to serve as a base for numerical analyses. Specific objectives were to clarify the effects of (i) the invoked procedure for estimating the soil hydraulic parameters and (ii) using increasingly refined spatial definitions of the hydraulic properties on simulated two dimensional water content and flow velocity distributions. Water flow in and drainage from a large sand tank (approximately 5 m x 3 m at the base, 6 m x 5.6 m at the top) was investigated using soil hydrologic and geophysical methods. Numerical analyses of variably saturated flow along a two-dimensional cross-section were carried out in attempts to describe the heterogeneous flow fields using the Richards equation-based HYDRUS-2D code. The unsaturated soil hydraulic properties were described using van Genuchten-Mualem type expressions. Information from both in situ and laboratory measurements was employed to obtain parameter estimates. The observed variability in discharge rate with time was reproduced best when an average water retention curve was used and the saturated water content was set equal to the porosity, whereas cumulative outflow was predicted best when all van Genuchten hydraulic parameters were fitted to the retention data. Using heterogeneously distributed hydraulic parameters (assuming a layered profile or a random distribution of the saturated hydraulic conductivity) improve neither predictions of the cumulative discharge rate nor the variability in the outflow rate when compared with the homogeneous case. Efforts to construct or numerically simulate heterogeneous flow experiments may, therefore, not always be justified when water flow in sandy substrates is studied.