Managed aquifer recharge (MAR) provides efficient removal for many organic compounds and sum parameters. However, observed in situ removal efficiencies tend to scatter and cannot be predicted easily. In this paper, a method is introduced which allows to identify and eliminate biased samples and to quantify simultaneously the impact of (i) redox conditions (ii) kinetics (iii) residual threshold values below which no removal occurs and (iv) field site specifics. It enables to rule out spurious correlations between these factors and therefore improves the predictive power. The method is applied to an extensive database from three MAR field sites which was compiled in the NASRI project (2002e2005, Berlin, Germany). Removal characteristics for 38 organic parameters are obtained, of which 9 are analysed independently in 2 different laboratories. Out of these parameters, mainly pharmaceutically active compounds (PhAC) but also sum parameters and industrial chemicals, four compounds are shown to be readily removable whereas six are persistent. All partly removable compounds show a redox dependency and most of them reveal either kinetic dependencies or residual threshold values, which are determined. Differing removal efficiencies at different field sites can usually be explained by characteristics (i) to (iii).

Berlin’s drinking water is produced from groundwater replenished by 60 % from surface water from the city’s abundant rivers or lakes using bank fi ltration or artifi cial groundwater recharge. Compared to other bank fi ltration sites world wide, the situation in Berlin is characterized by low hydraulic conductivities but nevertheless high capacities. Interdisciplinary research projects have shown that travel times and redox conditions during subsurface passage are highly transient due to seasonal effects and discontinuous pump operation. Trace organics like pharmaceuticals and x-ray contrast media are attenuated during subsurface passage to a varying degree. Substances that were found to be poorly removed under oxic conditions or even persistent include carbamazepine, primidone, sulfamethoxazole, 1,5 NDSA, MTBE and EDTA. Under anoxic to anaerobic conditions others like phenazone and diclofenac show little removal. However, none of these substances occur at relevant concentrations in the fi nished drinking water due to low initial concentrations in the surface water or additional removal during post-treatment (aeration and fi ltration for iron and manganese removal).

Bank filtration (BF) and aquifer recharge (AR): aquifer storage recharge (ASR), aquifer storage transport recharge (ASTR); are natural and semi-natural methods for drinking water treatment and constitute a major barrier within water supply system. Recent investigations have shown that about 60 % of Berlin’s drinking water is produced via BF or AR (Zippel & Hannappel 2008). Most drinking water therefore originates from surface waters within the cities limits and is pumped from wells adjacent to it’s many lakes and rivers. Since more than 100 years this system has been supplying safe drinking water so that post-treatment is limited to aeration and subsequent sand filtration. Disinfection is usually not applied (SenStadtUm 2008). The research project NASRI (“Natural and Artificial Systems for Recharge and Infiltration”, KWB 2002 – 2006), funded by the Berliner Wasserbetriebe (BWB) and Veolia (VE) had the aim to characterize the specific hydraulic and hydrochemical conditions at selected BF and AR sites in Berlin and to assess the behaviour of major water constituents, trace organic substances, algal toxins and pathogens during subsurface passage. For this, field investigations at three transsects (Lake Tegel BFsite, Lake Tegel AR-site and Lake Wannsee), laboratory and technical scale experiments were carried out by 7 different working groups. The results of the investigations were documented in 6 extensive research reports and were the basis for nearly 50 scientific publications. In 2007 the IC-NASRI project (Integration & Consolidation of the NASRI outcomes) was initiated by VE and BWB in order to support the practical implementation and optimization of bank filtration and aquifer recharge for drinking water production with the experience gained during the NASRI project. The aim was to derive practical guidelines for design and operation of BF & AR systems by i) further interpretation of the NASRI data and ii) integrating experience from other BF / AR sites world wide. Although subsurface passage is characteristic to many systems of managed aquifer recharge (MAR) the investigations within IC-NASRI concentrated on systems where drinking water is produced by infiltration of surface water either from the banks of a lake / river or from infiltration ponds (or similar systems like ditches or irrigation fields). A transfer of the presented results to other MAR systems, which use different recharge methods (e.g. ASR) or different sources (e.g. treated wastewater) therefore needs to be considered carefully, even though many statements may be true for them as well. This reports aims at providing engineers and scientists involved in drinking water production by BF & AR with up-to-date information on settings of similar systems world wide and on the systems’ performance with regard to drinking water treatment. The aim was to give the reader a condensed overview of the topic whereas further details can be taken from the large number of references given in the bibliography.

Managed aquifer recharge provides efficient removal for many organic water constituents but it is a difficult task to quantify removal under field conditions: Observed concentrations often scatter and may be biased by subsurface mixing of different waters. Removal efficiency is affected by different environmental parameters, such as redox potential, travel times, threshold values, and also field site specifics. In addition, it is crucial to know the corresponding surface water concentration for all samples. We developed a method, which overcomes these difficulties, quantifies the efficiency and removal kinetics and is applicable to extensive databases. It combines both, statistical and graphical evaluation which allows the determination of precise values and also interpretation based on expert knowledge. The database of this study was collected within the NASRI project between 2002 and 2005 at two bank filtration sites (Tegel BF, Wannsee BF) and one basin aquifer recharge site (Tegel AR) in Berlin. In total, 38 organic constituents were analysed (Table 1).

Managed Aquifer Recharge (MAR) comprises a wide variety of systems in which water is intentionally introduced into an aquifer and subsequently recovered, e.g. for drinking water or irrigation purposes. The objective is i) to store excess water for times of less water availability and / or ii) to introduce an additional barrier for purification of water from different sources (e.g. surface water, treated waste water) for a specific use. Common MAR techniques in Europe are (Figure 1): river bank filtration (RBF) and artificial groundwater recharge – usually via ponded infiltration (AR). Riverbank filtration (RBF) has a long history as a process for generating safe water for human consumption in Europe. During industrialization in the 19th century drinking water facilities in England, the Netherlands and Germany started using bank filtered water due to the increasing pollution of the rivers. The systematic production of bank filtrates started around 1870-1890 (BMI 1975, 1985). Since then, RBF and in case of insufficient quantity, artificial groundwater recharge (AR) have been generally applied as a first barrier within the drinking water treatment chain. The most common and widely used method for artificial groundwater recharge (AR) are infiltration ponds (Asano, 2007). These simple surface spreading basins provide added benefits of treatment in the vadoze zone and subsequently in the aquifer. Advanced pretreatment of the infiltration water by coagulation, and advanced post-treatment of the recharged water, e.g. with activated carbon or ozonation became necessary in many cases after the 1960’s as the quality of the source water further decreased. Today the water supply of many European cities and densely populated areas relies on riverbank filtration or artificial recharge. Following Castany (1985), in France, the proportion of bank-filtered water reaches approximately 50% of the total drinking water production (Doussan et al., 1997). In the Netherlands 13% of drinking water is produced from infiltration of surface water, such as bank filtration and dune infiltration (Hiemstra et al., 2003). In Germany riverbank filtration and artificial groundwater recharge are used in the valleys of the rivers Rhein, Main, Elbe, Neckar, Ruhr, and in Berlin along the Havel and Spree (Grischek et al., 2002). In Berlin 75% of the drinking water is derived from riverbank filtration and artificially recharged groundwater (Schulze, 1977). Riverbank filtration is also applied in the United States as an efficient and low cost drinking water pre-treatment technology (Ray et al., 2002), also to improve the removal of surface water contaminating protozoa. In most applications, MAR is intended to act as a buffer in terms of water availability (quantity) and water quality. In general, the level of knowledge of natural treatment systems, notably in aquifers, is not as high as in engineered systems, because the biogeochemical environment in aquifers that modify water quality for sure, will vary in space and time (Dillon et al. 2008). The heterogeneity of the system, strengthens its buffer potential on the one hand, but makes it more difficult to describe and control on the other hand. Key parameters that determine the quantitative storage capacity of the system are the specific hydrogeology of the aquifer (e.g. transmissivity and porosity) and the clogging potential at the entry point of the recharge water (infiltration pond, well or river bank). Clogging occurs due to physical, chemical and biochemical processes and needs to be regarded carefully as it may reduce the systems performance substantially. From literature it is known, that increased clogging reduces the oxidation state of the clogging layer. At a bank filtration site at Lake Tegel, Berlin, it was observed that intensity and spatial distribution of clogging strongly depends on the extent and thickness of the unsaturated zone. Geochemical observations suggest, that atmosperic oxygen induces redox processes which lead to a reduction of the clogging layer (Wiese & Nützmann 2008). This is possibly due to the complex interaction of hydrochemical and biological processes within the uppermost centimetres of the aquifer (Hoffmann et al., 2006). If these processes are likewise found in AR system, they may be influenced as to minimize basin-cleaning efforts. This needs to be further investigated. Water quality aspects of MAR are governed by i) the quality of the infiltrated / injected water ii) physical straining of particulate and particle-bound substances, iii) adsorption and desorption, iv) biogeochemical degradation / deactivation processes within the aquifer, iv) the geochemical composition of the aquifer, and v) the quality of the ambient groundwater. The process most important for MAR applications is usually the physical straining of particulate and particlebound substances, lessening the effort for subsequent drinking water treatment. In Berlin, e.g. disinfection of drinking water can usually be avoided due to complete removal of pathogens during underground passage of up to 6 months. Cyanobacterial toxins (e.g. microcystins) that are primarily cell-bound are efficiently removed as well (Grützmacher et al. 2007). On the other hand there is still a lack of understanding under which circumstances microcystins or other cyanobacterial toxins like cylindrospermopsin (currently observed in growing quantities in Germany) are released, thus becoming potentially more mobile in the subsurface. Adsorption to the aquifer matrix contributes to the elimination of organic substances and heavy metals. Although this does not remove the substances completely, peak loads – e.g. from oil spills – are retarded and maximum concentrations reduced. In addition, sorption prolongs the detention time in the aquifer which multiplies the time for biodegradation. Biological degradation in the subsurface is responsible for the elimination of dissolved organic carbon (usually resulting from natural organic matter, NOM) and organic trace substances that occur at varying extent. Investigations have shown that the redox potential in the aquifer is decisive for the degree of elimination (Stuyfzand, 1998; Massmann et al. 2007). Due to increasingly sensitive analytical methods trace organics present in surface waters (e.g. pharmaceutical residues) have been detected in many MAR systems e.g in Berlin and the Netherlands (Massmann et al, 2007, Stuyfzand et al. 2007). Advanced numerical models including reactive flow and transport can simulate the complex interactions between the hydrogeochemical environment and degradation of trace organics (Greskowiak et al. 2006). However, so far this has only been applied for a limited number of compounds at very few sites. Further research is needed to apply these methods for risk assessment. A second method for predicting the removal of organic micropollutants is the more statistically based approach of linking substance properties (molecular weight, number of double bonds, number of aromatic rings, etc.) to biodegradation via quantitative structure-activity relationship (QSAR) type models. This has been applied successfully to other water treatment methods – a transfer to MAR is lacking so far. As MAR is a technology that relies on the interaction of natural processes framework conditions like climate and hydrogeology play an important role. There is a need for testing the transferability from central European conditions to other regions, and for an assessment, how temperature changes affect the system’s elimination capacity. With ongoing climate change, reducing precipitation in some regions of Europe and increasing peak flow events in others, MAR is the ideal technology to act as a buffer for quantity and quality. The European Water Supply and Sanitation Platform (www.wsstp.org) for example has identified MAR as a technology potentially fit for future challenges.