Modeling Water Flux at the Base of the Rooting Zone for Soils with Varying Glacial Parent Materials

Soils of varying glacial parent materials in the Great Lakes Region (USA) are characterized by thin unsaturated zones and widespread use of agricultural pesticides and nutrients that affect shallow groundwater. To better our understanding of the fate and transport of contaminants, improved models of water fluxes through the vadose zones of various hydrogeologic settings are warranted. Furthermore, calibrated unsaturated zone models can be coupled with watershed models, providing a means for predicting the impact of varying climate scenarios on agriculture in the region. To address these issues, a network of monitoring sites was developed in Indiana that provides continuous measurements of precipitation, potential evapotranspiration (PET), soil volumetric water content (VWC), and soil matric potential to parameterize and calibrate models. Flux at the base of the root zone is simulated using two models of varying complexity: 1) the HYDRUS model, which numerically solves the Richards equation, and 2) the soil-water-balance (SWB) model, which assumes vertical flow under a unit gradient with infiltration and evapotranspiration treated as separate, sequential processes. Soil hydraulic parameters are determined based on laboratory data, a pedo-transfer function (ROSETTA), field measurements (Guelph permeameter), and parameter optimization. Groundwater elevation data are available at three of six sites to establish the base of the unsaturated zone model domain. Initial modeling focused on the groundwater recharge season (Nov-Feb) when PET is limited and much of the annual vertical flux occurs. HYDRUS results indicate that base of root zone fluxes at a site underlain by glacial ice-contact parent materials are 48% of recharge season precipitation (VWC RMSE=8.2%), while SWB results indicate that fluxes are 43% (VWC RMSE=3.7%). Due in part to variations in surface boundary conditions, more variable fluxes were obtained for a site underlain by alluvium with the SWB model (68% of recharge season precipitation, VWC RMSE=7.0%) predicting much greater drainage than HYDRUS (38% of recharge season precipitation, VWC RMSE=6.6%). Results also show that when calculating drainage flux over the recharge period, HYDRUS is highly sensitive to model initialization using observed water content from in-situ instrumentation. Simulated recharge season drainage flux is as much as 3.5 times higher when a one-month spin-up period was performed in the HYDRUS model for the same site. SWB results are less sensitive to water content initialization, but drainage flux is 1.6 times higher at one site using the same spin-up analysis. The long-term goals of this effort are to leverage the robust calibration data set to establish optimal approaches for determining hydraulic parameters such that water fluxes in the lower vadose zone can be modeled for a wider range of geomorphic settings where calibration data are unavailable.