Field evidence of a dynamic leakage coefficient for modelling river–aquifer interactions

Abstract

Summary In groundwater flow modelling, the interaction between rivers and aquifers is usually modelled with spatially and temporally constant leakage coefficients. We used conventional model calibration techniques to investigate the time-varying river–aquifer interactions in the sandy gravel aquifer of the upper Limmat valley in Zurich (Switzerland). The aim of the study was to determine whether the leakage coefficients have to be treated as time-dependent in order to adequately model the dynamics of the groundwater flow. A transient horizontal two-dimensional groundwater flow model was established together with a one-dimensional hydraulic model for river flow, as well as a scheme calculating groundwater recharge and lateral inflow from meteorological data and a soil water balance model. The groundwater flow model was calibrated using hydraulic head data from May and June 2004 and July and August 2005. The verification period covered 13 years using hydraulic head data from 90 piezometers. The comparison of the model results with the measurements in the verification period revealed three phenomena concerning river–aquifer interaction which all showed up as systematic deviations between model and observations. (1) The major flood event in May 1999 had a significant and persistent influence on the river–aquifer interaction. In an impounded river section upstream of a weir, the infiltration of river water was enhanced by the flooding probably due to erosion processes. (2) Seasonal river water temperature fluctuations influenced the infiltration rate, due to the temperature dependence of hydraulic conductivity of the river bed. (3) Depending on geometry and hydraulic characteristics of the riverbanks the leakage coefficient can be a function of the river stage. With higher water levels, additional areas can contribute to the infiltration of river water. Therefore, in modelling groundwater flow with strong river–aquifer interactions, it can become necessary to consider dynamic leakage coefficients and to recalibrate periodically.