Calibration and Reliability in Groundwater Modelling: Credibility of Modelling

(Proceedings of ModelCARE 2007 Conference, held in Denmark, September 2007). IAHS Publ. 320, 2008, 163-168.

How successful are standard models of groundwater–river interactions

Fritz Stauffer, Tobias Doppler &
Harrie-Jan Hendricks Franssen

Institute of Environmental Engineering, ETH Zurich, CH-8093 Zurich, Switzerland

stauffer@ifu.baug.ethz.ch

Abstract In horizontal two-dimensional aquifer models the interaction between rivers and aquifers is usually taken into account by a leakage concept. Such approaches are based on various assumptions and simplifications like Dupuit’s assumption for the representation of the flow conditions between river and aquifer. The leakage concept is primarily based on the observation of an additional hydraulic resistance caused by a clogging of the riverbed, which may be strongly spatially variable. Moreover, the resistance can be time-dependent due to various processes. Furthermore, it might be of importance whether infiltration or exfiltration conditions prevail. Also highly important are the hydrogeological conditions below or close to the river, which may be strongly non-homogeneous. The prevailing infiltration or exfiltration rate furthermore depends on the geometry of the riverbed, which may vary over time due to variable flow regimes. In cases where the groundwater table is below the river bottom, the infiltration rate depends on the conditions in the capillary zone. Near-shore vegetation can exhibit a further influence. Various concepts exist to model the interaction between river and groundwater. Often a linear relationship between the specific leakage rate and the difference between river head and groundwater head is assumed. The resulting leakage parameter of extended river reaches cannot be measured directly. The usual way is therefore to estimate it by model calibration using field data. However, how successful are such parameters when modelling uncalibrated periods? Are the leakage parameters constant over time? Is the relationship linear? The problem of the interaction between river and groundwater has been investigated based on field data for two examples in Switzerland. It was shown that the leakage coefficient can be time dependent due to sedimentation and flood events. Moreover, large fluctuations of the water temperature lead to temperature-dependent infiltration conditions. Furthermore, the leakage coefficient is in general nonlinear and depends on the river head by exhibiting increased bank infiltration due to an increase of the wetted perimeter of the river bed.

Key words groundwater; river–aquifer interaction; leakage coefficient; parameter estimation