Recharge of waste water in an unconsolidated poorly sorted alluvial aquifer is a complex process, both physically and hydrochemically. The aim of this paper is to analyse and conceptualise vertical transport mechanisms taking place in an urban area of extensive wastewater infiltration by analysing and combining the water balance, the microbial (Escherichia coli) mass balance, and the mass balance for dissolved solutes. For this, data on sediment characteristics (grain size, organic carbon, reactive iron, and calcite), groundwater levels, and concentrations of E. coli in groundwater and waste water were collected. In the laboratory, data on E. coli decay rate coefficients, and on bacteria retention characteristics of the sediment were collected via column experiments. The results indicated that shallow groundwater, at depths of 50 m below the surface, was contaminated with E. coli concentrations as high as 10⁶ CFU/100 mL. In general, E. coli concentrations decreased only 3 log units from the point of infiltration to shallow groundwater. Concentrations were lower at greater depths in the aquifer. In laboratory columns of disturbed sediments, bacteria removal was 2-5 log units/0.5 cm column sediment. Because of the relatively high E. coli concentrations in the shallow aquifer, transport had likely taken place via a connected network of pores with a diameter large enough to allow bacterial transport instead of via the sediment matrix, which was inaccessible for bacteria, as was clear from the column experiments. The decay rate coefficient was determined from laboratory microcosms to be 0.15 d^{- 1}. Assuming that decay in the aquifer was similar to decay in the laboratory, then the pore water flow velocity between the point of infiltration and shallow groundwater, coinciding with a concentration decrease of 3 log units, was 0.38 m/d, and therefore, transport in this connected network of pores was fast. According to the water balance of the alluvial aquifer, determined from transient groundwater modelling, groundwater flow in the aquifer was mainly in vertical downward direction, and therefore, the mass balance for dissolved solutes was simulated using a 1D transport model of a 200 m column of the Quaternary Alluvium aquifer. The model, constructed with PHREEQC, included dual porosity, and was able to adequately simulate removal of E. coli, cation-exchange, and nitrification. The added value of the use of E. coli in this study was the recognition of relatively fast transport velocities occurring in the aquifer, and the necessity to use the dual porosity concept to investigate vertical transport mechanisms. Therefore, in general and if possible, microbial mass balances should be considered more systematically as an integral part of transport studies.

Because of heterogeneity among members of a bacteria population, deposition rates of bacteria may decrease upon the distance bacteria are transported in an aquifer. Such deposition rate decreases may result in retained bacteria concentrations, which decrease hyper-exponentially as a function of transport distance, and may therefore significantly affect the transport of colloids in aquifers. We investigated the occurrence of hyper-exponential deposition of Escherichia coli, an important indicator for fecal contamination, and the causes for such behavior. In a series of column experiments with glass beads of various sizes, we found that attachment of E. coli decreased hyper-exponentially, or, on logarithmic scale in a bimodal way, as a function of the transported distance from the column inlet. From data fitting of the retained bacteria concentration profiles, the sticking efficiency of 40% of the E. coli population was high (α=1), while the sticking efficiency of 60% was low (α=0.01). From the E. coli total population, an E. coli subpopulation consisting of slow attachers could be isolated by means of column passage. In subsequent column experiments this subpopulation attached less than the E. coli total population, consisting of both slow and fast attachers. We concluded that the main driver for the observed dual mode deposition was heterogeneity among members of the bacteria population. Intra-population may result in some microbes traveling surprisingly high distances in the subsurface. Extending the colloid filtration theory with intra-population variability may provide a valuable framework for assessing the transport of bacteria in aquifers.

Though coliform bacteria are used worldwide to indicate fecal pollution of groundwater, the parameters determining the transport of Escherichia coli in aquifers are relatively unknown. We evaluated the occurrence of both straining and attachment of E. coli ATCC25922 in columns of ultra-pure, angular, saturated quartz sand. The column experiments were conducted over a wide range of porous medium sizes, column heights, input concentrations, and pore water flow velocities. Straining and attachment were examined by modelling the breakthrough curves (with HYDRUS 1D). In addition, model output was compared with measured strained and attached bacteria via column extrusion experiments (in which sand was extruded from the column and placed in excess water) and flow reversal experiments (in which the pore water flow direction was reversed, thereby dislodging strained bacteria). Our model consisted of an attachment rate coefficient and a straining rate coefficient; both of these decreased with transport distance. The straining rate coefficient also decreased in a Langmuirian way, in response to the filling of available pore space, which in turn depended on influent bacteria concentration, quartz grain diameter, and transport distance. The maximum strained fraction was 25-30% of total bacteria mass applied to the column; the maximum attached fraction was 30-35%. The fit between modelled and measured (strained and attached) bacteria masses was acceptable, as was the sensitivity of the model output to fitted parameter values. Our results lead to a new description for the time-dependent mass balance of strained bacteria, which entails using three fitting parameters. The results also imply that column experiments in combination with retention profiles (or various column lengths) are not enough to explain the retention processes in a column. Column extrusion and flow reversal experiments provide vital additional information on the occurrence and magnitude of straining. Our straining model could be of assistance in evaluating the importance of straining and in incorporating the straining process in bacteria transport modelling.