Inverse analysis of upward water flow in a groundwater table lysimeter

The accuracy of numerical water flow models for the vadose zone depends on the estimation of the soil hydraulic properties. In this study, the hydraulic parameters for a silty clay soil in a large lysimeter were determined through inverse modeling of a fallow period with upward water flow from a shallow groundwater table. Parameter uniqueness was studied by simulating a hypothetical soil with known hydraulic properties under comparable conditions. Sensitivity analysis showed that the pressure head h(z,t), the volumetric water content theta(z,t), and the cumulative bottom flux Q(t) were least sensitive to the residual volumetric water content theta(r) and the pore-connectivity parameter lambda in the van Genuchten-Mualem (VGM) model. Parameter response surfaces showed that least squares fitting with theta(z,t) data is more likely to result in a unique hydraulic parameter set than least squares fitting with h(z,t) or Q(t) data. With only theta(z,t) in the objective function, the least squares minimization algorithm was capable of finding the correct soil hydraulic parameters, provided that theta(r) and lambda were fixed and that multiple initial parameter estimates were used. The protocol that was developed for the hypothetical soil was subsequently applied to the actual groundwater table lysimeter. The soil hydraulic parameters for the lysimeter for two (x,y) locations were determined using theta(z,t) data as measured by capacitance sensors. The variability in the optimized inverse of the air-entry value alpha and the saturated hydraulic conductivity K(s) in the VGM model was relatively high because of the high parameter correlation between these parameters. The optimized soil hydraulic properties can be used to study capillary rise from the groundwater table.