Experimental and Numerical Study of Soil Moisture Dynamics Above a Moving Water Table
Friday, 19 December 2014
We have performed an experimental and numerical study of water flow in the vadose zone with infiltration at the soil surface and a moving water table. Laboratory experiments were patterned after the experiments of Childs and Poulovassilis (1962. The moisture profile above a moving water table, Journal of Soil Science, 13(2), 271-285) and conducted in a Wedron sand packed column (14.5 cm in diameter and 141 cm in length) using different upper boundary fluxes and varying water table velocities. A constant head tank reservoir suspended from a stepper motor was used to control water table movement. Time-domain reflectrometry (TDR) probes were used to measure water content at seven different depths, while tensiometers measured capillary pressures at three depths. Measured water contents and pressures indicated hysteresis effect. The retention curves were affected by the wetting/drying processes, soil surface rainfall conditions, and varied at different locations. A single set of van Genuchten parameters were identified that resulted in adequate description of soil water retention relations for both rising and falling water table cases. The water flow dynamics were simulated using the one-dimensional Talbot-Ogden (T-O) infiltration and redistribution method (Talbot and Ogden, 2008. A method for computing infiltration and redistribution in a discretized moisture content domain, Water Resources Research, 44(8), W08453, DOI: 10.1029/2008WR006815) modified to consider the influence of the groundwater table on the vadose zone. Simulation results agreed satisfactorily with measurements for the evolution of water content profile, though hysteresis effect was not considered in the T-O method. Predicted ponding/deponding times using T-O method were generally acceptable. The performance of T-O method was compared with the numerical solution of Richards' equation (RE). Results showed similar performance for both methods, however the RE solver performed better in cases with a falling water, while the T-O method performed better in cases with a rising water table. The performance of both T-O and RE were best when the water table velocity was less than 20 percent of the saturated hydraulic conductivity.