Determination of the vertical distribution of hydraulic parameters using gadolinium as an anthropogenic tracer and inverse modelling

Bank filtration is widely used for drinking water production around the world. Due to the general composition of river water and the possibility of direct anthropogenic inputs, it is of great importance to understand the interaction between river and groundwater as well as the subsurface flow conditions. These can alter with increased runoff during flood events. Higher gradients between the surface water and groundwater, enlarged infiltration zones or the removal of colmation layers can lead to elevated infiltration rates and thus to changes in the river aquifer system. Due to often high hydraulic permeabilities and the potential spatial proximity between the river and the extraction system, short dwell times may therefore occur. Additionally, associated with stormwater runoff of wastewater treatment plants, higher contamination risks can therefore be expected in the extraction system during flood events.

In our investigations, we use a regional three-dimensional numerical groundwater model to help prevent changes in the quality of the extracted water through optimized operational management. However, the need to predict the flow paths between the river and the water work with maximum precision makes it necessary to complement the regional model with high resolution local models. To better capture vertical heterogeneities constraining local flow paths, a two-dimensional vertical model following the direction of maximum contribution of bank filtration to the water work was additionally created using FEFLOW 7.5 (DHI). Along this transect, the X-ray contrast agent gadolinium was sampled for use as a conservative anthropogenic tracer at different depths.

The sampling of gadolinium every 12 hours during a minor flood event showed a weekly, wastewater-influenced signal in the surface water, which could also be followed in the transect. This signal, together with ²²²Rn tracer ages, complements the time-resolved observations of groundwater levels to calibrate the vertical distribution of hydraulic parameters of the two-dimensional mass transport model. These are supposed to improve the conceptual regional model to ultimately optimize the operation of the waterworks and allow the extraction of the groundwater with the best possible quality.