Within the scope of the interdisciplinary NASRI research project (Natural and Artificial Systems for Recharge and Infiltration) dealing with river bank filtration processes at Berlin water works, a semi-technical column experiment is ongoing since January 2003 to simulate river bank filtration. Here a 30 m long ensemble of 6 soil columns is operated by surface water sampled from Lake Tegel (Berlin, Germany). In April 2004, oxic conditions in the column ensemble were forced to change in anoxic by gassing the flushing solution supply with nitrogen gas. The objective of reactive transport modelling was to (i) identify the main biogeochemical processes and the governing redox conditions within the soil column system during flushing as a conceptual model for river bank filtration and to (ii) verify the observed hydrochemistry of the pore water.

The present report characterizes the field sites Lake Tegel and Lake Wannsee as well as the artificial recharge site GWA Tegel in terms of their clogging layer, sedimentary, hydraulic and hydrochemical properties. As a result, a solid basis for the interpretation of specific compounds evaluated within NASRI and for subsequent modeling and quantification of the data is given. Major problems or difficulties where identified, in order to focus investigations on aspects not fully understood to date in the next project phase. The combination of different tracers enables the interpretation of the flow regime. With the help of T/He analysis, ages of different water bodies can be estimated. The analysis of tracer showing distinct seasonal variations is used to estimate travel times while water constituents which are either mainly present in the bank filtrate or the background water are used for mixing calculations. The proportions of treated wastewater in the surface water were estimated in front of the transects. The surface water composition varies largely both in time and space, which is a problem at Wannsee, where the surface water sampling point is not representative for the bank filtration input. Estimates for travel times of the bank filtrate to individual observation and production wells are given and vary between days and several months. The production wells are a mixture of bank filtrate and water from inland of the wells and deeper aquifers, proportions of bank filtrate are given where possible to differentiate between contaminant removal and dilution. They vary between < 20 and > 80 %. The new observation wells enable a vertical differentiation of the infiltrate. It becomes clear that at Tegel and Wannsee, there is a strong vertical succession towards larger proportions of considerably older bank filtrate with depth. At the Wannsee transect, the observation wells deeper than the lake do not reflect the surface water signal at all. It will be important to combine the new information with hydraulic information of existing flow models (mainly of the IGB “model” group). The evaluation of the redox conditions shows that redox successions proceed with depth rather than (only) in flow direction. In addition, the redox zoning (as characterised by the appearance or disappearance of redox sensitive species) is very transient. The zones are much wider in winter than in summer, in particular at the artificial recharge site GWA Tegel, probably due to temperature effects. This poses a challenge for the desired modelling and the interpretation of data from redoxsensitive substances.