Integrated hydrologic models characterize catchment responses by coupling the subsurface flow with land surface processes. One of the major areas of uncertainty in such models is the specification of the initial condition and its influence on subsequent simulations. A key challenge in model initialization is that it requires spatially distributed information on model states, groundwater levels and soil moisture, even when such data are not routinely available. Here, the impact of uncertainty in initial condition was explored across a 208 km2
catchment in Denmark using the ParFlow.CLM model. The initialization impact was assessed under two meteorological conditions (wet vs
dry) using five depth to water table and soil moisture distributions obtained from various equilibrium states (thermal, root zone, discharge, saturated and unsaturated zone equilibrium) during the model spin-up. Each of these equilibrium states correspond to varying computation times to achieve stability in a particular aspect of the system state.
Results identified particular sensitivity in modelled recharge and stream flow to the different initializations, but reduced sensitivity in modelled energy fluxes. Analysis also suggests that to simulate a year that is wetter than the spin-up period, an initialization based on discharge equilibrium is adequate to capture the direction and magnitude of surface water–groundwater exchanges. For a drier or hydrologically similar year to the spin-up period, an initialization based on groundwater equilibrium is required. Variability of monthly subsurface storage changes and discharge bias at the scale of a hydrological event show that the initialization impacts do not diminish as the simulations progress, highlighting the importance of robust and accurate initialization in capturing surface water–groundwater dynamics.