Testing the EPIC Richards submodel for simulating soil water dynamics under different bottom boundary conditions

Abstract

Abstract Most biogeochemical models simulate water dynamics using the tipping bucket approach, which has been often found to be too simplistic to represent vadose zone dynamics adequately under shallow groundwater conditions. Recently, a solution to the Richards equation using the Mualem–van Genuchten model (Rich‐vGM) has been added into the EPIC (Environmental Policy Integrated Climate) model to address this shortfall. Its performance was tested using lysimeters operating under free drainage (FD) and at a shallow water table (60‐ [WT60] and 120‐cm depth [WT120]). Model accuracy was also compared with the upgraded tipping bucket‐based method implemented into EPIC (the variable saturation hydraulic conductivity method [VSHC]). Soil water content (SWC) data were split into calibration and validation subsets. Model evaluation also included annual evapotranspiration (ET), percolation (PRK), and upward water movements to assess underlying soil water balance factors. The submodels provided accurate and similar results upon comparison with SWC measures under FD (Nash–Sutcliffe coefficient [NSE] = 0.26 and 0.61 using VSHC and Rich‐vGM, respectively). The Rich‐vGM model accurately reproduced observed SWC and ET (e.g., NSE = 0.70 and percentage bias [PBIAS] = −3.7% for WT120, respectively) although it slightly overestimated PRK (PBIAS = 47.8%, on average). Instead, VSHC proved unable to correctly simulate shallow groundwater conditions (e.g., NSE = −1.85 for WT60 SWC). Under shallow groundwater conditions, the Rich‐vGM method is recommended, despite the additional data required and the need to define the bottom boundary conditions according to water table fluctuations. In conclusion, the Richards solver introduced and tested in EPIC improved the model's ability to represent complex biophysical and biogeochemical processes in terrestrial ecosystems associated with the hydrological balance.