Impact of topographic resolution on simulated regional groundwater flow and residence time.

International audienceLarge regional numerical models are often used for the prediction of groundwater fluxes to surface water or to assess the impacts of changes in either climate, land use or groundwater exploitation. Such models typically use a large grid size and a low-resolution land surface topography. The objective of our study was to assess if such large models can provide accurate predictions of fluxes to rivers based on representative recharge and hydraulic properties. For this purpose, a simplified 95 km-long 2D cross-section hydrogeological model was developed using the topography of an area from the Appalachian Highlands to the St. Lawrence Lowlands in southwestern Quebec (Canada). Numerical simulations were completed with a fixed numerical grid but with four levels of topographic resolution using data points at intervals of 30, 90, 500 and 1000 m. The regional rock aquifer models all used the same depth-decreasing hydraulic conductivity (K) profile based on field conditions. Steady-state groundwater flow and age simulations were done by imposing heads corresponding to the topography in order to obtain flowpaths and surface fluxes (inflow and outflow). Simulated inflows were compared to recharge values independently estimated with the HELP infiltration model calibrated with total flow and baseflow in gauged watersheds. Results show that topographic resolution has a major impact on surface fluxes, both entering and exiting the aquifer. In order for smooth-topography models to represent fluxes to streams, it is necessary to use “equivalent” K values that are significantly higher than estimates based on thousands of specific capacity tests. For simulations using large regional groundwater flow models with smoothed topography, a choice has to be made between the proper representation of surface fluxes using unrealistic K values and the underestimation of surface fluxes using representative K profiles