The value of the radon-222 isotope for the prediction of groundwater residence times

The possible occurrence of health-sensitive bacteria in groundwater is directly related to their subsurface residence times. Several authors demonstrated the use of the dissolved radon isotope 222Rn (half-life 3.8 days) to determine mean aquifer residence time of freshly infiltrated surface water These authors based their studies on the following assumptions: (1) the average distribution of the 222Rn progenitor 226Ra is homogenous in the solid material on a macroscopic scale; (2) the radon activities, which reach steady-state after 15 to 20 days of groundwater residence, are representative of the entire flow path; (3) radon losses from the saturated into the unsaturated zone are constant with time; (4) the flow distance between infiltration and sampling location remains constant; and (5) the infiltrating water does not mix with older groundwater. The authors investigated field conditions, where rivers or canals form a constantly infiltrating boundary condition. They determined the steady-state 222Rn activity AS by choosing a distant representative sampling location where the aquifer residence time corresponds to at least the steady-state radon activity of 15 to 20 days. The infiltrating surface water body of our study site was a wetland pond, where the amount of inflowing surface water was controlled with gates. Previous to the sampling period, the infiltration was stopped for four weeks in order to uniquely obtain 'older' groundwater inside the test part of the aquifer. We suggest the use of a site-specific steady-state activity for these field conditions: by sampling the 222Rn activity sites before the start of the infiltration cycle, we measure a steady-state activity for each sampling location. Differences between sampling locations reflect the effects of geological heterogeneities on radon emanation along the respective flow path. The use of this site-specific steady-state activity further minimizes errors of calculated residence times. We further propose the use of an additional conservative tracer data, which permits the partitioning between older and freshly infiltrated groundwater in the sampling well. This allows adapting the fifth assumption of the cited authors, because mixing with older groundwater will occur in most field conditions, except if one does calculate residence time in the close vicinity of the infiltration boundary. The methods were studied in an aquifer, which consists of Quaternary alluvial deposits and provides a large part of the drinking water for the city of Basel. The study was part of a multidisciplinary project, where we tested the effects of wetland restoration on groundwater quality. In our field site we observed distinctive chloride concentration differences between existing groundwater and wetland (surface) water. Concentration distributions of the chloride anion are widely analyzed in subsurface hydrology due to the fact that dissolved chloride is considered a conservative tracer. The measured concentration of the surface water varied from 2.2 to 7.7 mg 1-1 for the samples taken during three periods of four weeks. The chloride content in the groundwater, however, varied between 21.2 and 27.5 mg 1-1. A simple analytic mixing model permits to calculate the percentage of recently infiltrated surface water at the sampling location. We assume steady-state flow conditions and a constant chloride concentration of the infiltration water and the older groundwater during the test. The hereby-calculated percentages are used to correct the measured 222Rn activities of the same samples. The calculated residence times tend to be the lowest for the wells beneath the wetland field (1 to 7 days), and the well located in the top layer 1 in 36 meter distance of the infiltration field (1 to 5 days). The migration to the well in the lowest part of the aquifer (layer 3), within almost the same distance to the infiltration (39 m), takes between 6 and 8 days. Migration from the infiltration field to the next cluster at 85 meter distance takes between 7 days and 13 days, respectively. The use of the proposed corrections of 222Rn activities to calculate more realistic residence times was tested by comparing the 222Rn residence times to corresponding values calculated with another method. For each sampling location, residence times were predicted with a numerical 3-D groundwater flow and transport model, which was calibrated on hydraulic head and chloride solute concentration. We found a good agreement between both methods especially in wells, where residence times were more than 5 days from the wetlands. (authors)