The spatial distribution of confining layers within a system of two aquifers strongly affects the hydraulics and sensitivity to pollution. The test site is located close to a well field. Wells are switched with short intervals and hydraulic heads are recorded in several observation wells. Because the absolute levels of simulated hydraulic heads do not always coincide with the measurements, the model is calibrated with short term head variations. The characteristic shape of the hydraulic heads at each observation wells contains sensitive information about the structure of the aquifer. A numerical technique is developed which enables to simulate the spatial distribution of the confining layer. The method comprises the use of pilot points and regularisation technique. Cross validation is carried out in order to show the results are physically based. The method is shown to provide significant results even under non optimal conditions.

Several kinds of managed aquifer recharge techniques provide very good purification of surface water since more than 100 years. In order to maintain a reliable supply of clean water, they are becoming increasingly popular all over the world. These methods require low technical effort. At Aquifer Storage and Recovery and ponded infiltration the recharged amounts are technically controlled. The infiltration water has to be pumped and often pretreated. At bank filtration this is dispensable, the approach, of using existing surface water bodies is even more consequent. Exemplarily, at a test site at Lake Tegel, Berlin, Germany, the hydraulic processes are modelled. By means of 3D long term regional and transient hydraulic modelling it was detected that the existing approaches for determining the leakance induce large errors in the water balance and describe the infiltration zone insufficiently. The leakance could be identified to be triggered by the groundwater table, causing air exchange and intrusion of atmospheric oxygen, which reduces clogging by altered redox conditions by at least one order of magnitude. This causes that changes of the groundwater table are mitigated much more than previously assumed. Taking these findings into account, a transient water balance is determined and bank filtration ratios are quantified. A new inverse modelling concept has been developed and applied to a 3D short term local and transient hydraulic model. It comprises spatially distributed pilot points and overparameterisation constrained by regularisation and calibration to head differences. Significance of the results is demonstrated by cross validation. With this approach the spatial distribution of an aquitard have been identified with high precision. The highly transient and heterogeneous flow conditions are specified and a new viewpoint on the geologic formation of Lake Tegel is obtained. The good fit of modelled and observed breakthrough curves of 18O, chloride and temperature by just using transferred parameters obtained with the previous hydraulic methods, show the very good model performance and predictive capabilities. The intrusion of atmospheric oxygen into the unsaturated zone is identified to be the principal redox determining factor during infiltration. Previously inconsistent and also local geochemical conditions are identified to be determined by interaction of infiltration processes with the spatial extent of the aquitard. A theory for chemical clogging of abstraction wells is developed, identifying the strong vertical redox zoning as principal factor of influence.

Successful predictions of the fate and transport of solutes during bank filtration and artificial groundwater recharge depends on the availability of accurate transport parameters. We expand the CXTFIT code (Toride et al., 1995) in order to improve the handling by pre- and post processing modules under Microsoft EXCEL. Inverse modelling results of column experiments with tracers, pharmaceutical residuals and algae toxins demonstrate the applicability of the advanced simulation tool.

A hydraulic and physically based transport model for the catchment of a well field was set up. With the study area situated in a region strongly influenced by surrounding well galleries the boundary conditions had to be worked out during modelling and partially had to be transient. Two important processes were clarified: Bank filtrate extracted at the investigated transect is composed of 3 water qualities from horizontal layers, each with a different age and infiltration area. Sampled wells containing the different water types were identified, providing information for correct chemical interpretation. Secondly, the lake sediments show a pronounced seasonal fluctuation in their leakage coefficient, with its winter values doubling in summer, and lagging 2–4 months behind water temperature.

In the course of the interdisciplinary research project NASRI (natural and artificial systems for recharge and infiltration) many investigations are currently being carried out to assess the risk of break through of persistent organic substances into raw water used for drinking water supply. One part of these studies is the determination of the transport behavior of pharmaceutical residues in test sand filters, so called enclosures, equipped with sampling points at various depths. Breakthrough curves were determined for carbamazepine, primidone (both antiepileptic drugs), clofibric acid (a metabolite of blood lipid lowering agents), diclofenac, ibuprofen (both analgesic drugs) and for chloride, used as a conservative tracer. Retardation coefficients and degradation rates were obtained by using the software Visual CXTFIT. Degradation rates between 0.7 h–1 and 1 h–1 were observed for ibuprofen whereas clofibric acid, primidone, carbamazepine and diclofenac showed no or very little degradation (lambda < 0.06 h–1).

The UBA’s experimental field on the outskirts of Berlin offers a unique possibility of simulating bank filtration, artificial recharge and slow sand filtration on a technical scale. The site consists of a storage reservoir (pond) with an adjacent artificial aquifer consisting of sand and gravel. Additionally the surface water can be conducted into 4 infiltration basins (two slow sand filters and two aquifer infiltration ponds). Three enclosures as well as large scale columns can be used for shorter and longer term simulation of groundwater transport. The whole site is separated from the surrounding aquifer by a layer of clay. A variety of physico-chemical parameters can be measured continuously and observed online. The travel times for the bank filtration passage determined by tracer experiments range from a few days to a maximum of 3 weeks. In the enclosures, infiltration ponds and large scale columns contact time can be varied between a few hours up to 3 months.

River bank or slow sand filtration is a major procedure for processing surface water to drinking water in central europe. In order to model the performance of river bank and slow sand filtration plants, we are studying the different mechanisms by which the elimination of pathogens is realized. An important question concerning the mode of action of slow sand filters and river bank filtration units is the role of the colmation layer or “schmutzdecke” on the elimination of human pathogens. The schmutzdecke is an organic layer which develops at the surface of the sand filter short after the onset of operation. We have inoculated a pilot plant for slow sand filtration with coliphages and determined their rate of breakthrough and their final elimination. In the first experiment, with a colmation layer still missing, the breakthrough of the coliphages in the 80 cm mighty sandy bed amounted to ca. 40 %. In contrast, less than 1 % of coliphages escaped from the filter as the same experiment was repeated two months later, when a substantial colmation layer had developed. Our preliminary conclusions are that the colmation layer is extremely efficient in eliminating of viruses.

The effect of oscillating pumping regimes at the bank filtration site in Berlin Tegel is examined via a scenario based modelling study. There are several scenarios for the pumping regimes, some adopted from the operation of the plant by the Berlin Water Works (BWB), some hypothetical with a regular oscillating regime. A horizontal 2D model of the lower aquifer is set-up, in which the third type boundary condition is used to mimic the influence of an irregularly shaped till layer, overlying the main aquifer. Model results in form of flowpaths are presented for several pumping scenarios. They reveal that there is a substantial influence of the pumping regime on the flowpaths in the vicinity of the well gallery, while in the farfield, including the bank of the surface water body (here: Lake Tegel) the oscillating effect is rather small. It depends very much on the infiltration position on the bank, whether traveltime through the aquifer changes as effect of irregular pumping.