|GUBER, ADREY - Michigan State University|
|YAKIREVICH, ALEXANDER - Ben Gurion University Of Negev|
|CADY, RALPH - Us Nuclear Regulatory Commission|
|NICHOLSON, THOMAS - Us Nuclear Regulatory Commission|
Submitted to: Vadose Zone Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/1/2014
Publication Date: 12/2/2014
Citation: Pachepsky, Y.A., Guber, A., Yakirevich, A., Mckee, L.G., Cady, R., Nicholson, T. 2014. Scaling and pedotransfer in numerical simulations of flow and transport in soils. Vadose Zone Journal. 14(12).
Interpretive Summary: The saturated hydraulic conductivity characterizes the ability of soils to store and conduct water and solutes. This is the most important parameter used in all hydrological and contaminant transport models. This parameter is difficult to measure, and estimating it from soil composition is a common practice. The estimates are obtained using empirical relationships between the saturated hydraulic conductivity and soil composition. These empirical relationships, called pedotransfer functions, are obtained from measurements of hydraulic conductivity on small samples. The saturated hydraulic conductivity is known to increase as the volume of soil increases. Our soil hypothesis was that estimated hydraulic conductivities have to be upscaled, i.e. increased, to be used in simulations if computational cells are substantially larger than soil volumes in hydraulic conductivity measurements. We used the USDA-ARS hydraulic conductivity pedotransfer functions and the power law upscaling. The long-term tracer experiment was carried out at the USDA-ARS OPE3 experimental site in Beltsville, MD. Monitoring and simulations of the tracer in soil were done in the line of groundwater wells. Using only pedotransfer functions caused a substantial difference between measured and simulated concentrations. After pedotransfer estimates were upscaled, the accuracy of simulations was the same as that of the calibrated model. Results of this work indicate the opportunity of replacing the highly resource consuming model calibration procedure with much less demanding search for upscaling coefficients, and that they can be of interest and use in a wide variety of hydrological and contaminant transport modeling projects.
Technical Abstract: Flow and transport parameters of soils in numerical simulations need to be defined at the support scale of computational grid cells. Such support scale can substantially differ from the support scale in laboratory or field measurements of flow and transport parameters. The scale-dependence of flow and transport parameters essentially precludes the direct use of measured or pedotransfer-estimated parameter values in numerical simulations. The hypothesis of this work was that a support-based scaling law can be introduced that can convert pedotransfer-estimated saturated hydraulic conductivity values into values to be used over grid cells for finite element-based simulations of water flow and tracer transport in variable saturated soils. A four month-long experiment was conducted at the USDA-ARS experimental site where chloride as a tracer was applied with a pulse of irrigation water and its transport in groundwater and variably saturated shallow coarse-textured soils was monitored in two rows of wells on daily basis. The HYDRUS-3D software was used to set and calibrate the Richards model for flow simulations and the convective-dispersive equation for transport simulations. Saturated hydraulic conductivity values were estimated with class pedotransfer functions derived from the USDA database containing results of about 1000 measurements in soil of different textures and bulk density. A power law scaling for the saturated hydraulic conductivity was suggested based on literature data. When only two parameters of the scaling law rather than nine values of hydraulic conductivity from nine soil materials were calibrated, using the scaled saturated hydraulic conductivity values resulted in the accuracy of simulations that was similar to the accuracy of the calibrated model results. Upscaling of pedotransfer-estimated saturated hydraulic conductivities can provide reasonable estimates for the numerical flow and transport modeling in variably saturated soils.