Groundwater dating with sulfur hexafluoride: Methodology and field comparison with tritium and hydrodynamic methods

Full text: The tritium method was regarded for several decades to be the simplest and most convenient tracer method for determining the ages of young groundwaters. However, with the tritium levels declining in the atmosphere, certain modifications of the tritium method were proposed, such as 3He/3H and 3He+3H methods. Also some anthropogenic atmospheric trace gases have been adopted as dating tools for young groundwaters, for instance 85Kr and freons (F-11, F-12 and F-113). Another atmospheric trace gas of anthropogenic origin (SF6) has recently been shown to a reliable tool for dating young groundwaters. The SF6 concentration in the atmosphere increases monotonically since about 1960 to the present concentration of about 5 pptv, which corresponds to about 2.2 fmol/l in water being in equilibrium with the atmosphere at the temperature of 10 deg. C. In the present study, performed partly under the BASELINE Project carried out within the EU Fifth Framework Programme, a simple methodology of SF6 analyses in groundwater samples has been developed and interpretation procedures tested. The current detection limit of our analytical system is approximately 0.02 fmol/l. Samples are collected in glass bottles (ca. 2.4 l) without contact with the atmosphere and are transported to the laboratory where the dissolved gases are extracted using the headspace method. The extracted SF6 is measured using gas chromatograph equipped with an electron capture detector. Two confined Polish sandy aquifers of Tertiary age were investigated within the project. The hydrochemistry (major, minor and trace components) and environmental tracers (3H, 14C, 13C, 2H, 18O, F-11, F-12, F-113 and SF6) were measured. Tritium and SF6 ages were estimated from time records of the tracers using the lumped-parameter approach. The hydrodynamic modelling was performed with the Visual-MODFLOW, flow times estimated with the MODPATH, and the space distributions of tritium and SF6 modelled with the MT3D. The presented work summarizes the results obtained so far in the Bogucice Sands aquifer, which belongs to the medium-size aquifers in Poland with the area of 176 km2. It consists of sands and sandstones of a Badenian deltaic basin, with high carbonate material contents (3-10 % in sands and 25-29 % in sandstones). The recharge is mainly at the outcrops in the south, and by downward leakage near the outcrops and in some areas of extensive withdrawal. The general flow direction is towards the Vistula valley in the north, with discharge by upward leakage through the confining layers of clays and claystones. The hydraulic conductivity values found from pumping tests performed within the aquifer boundaries decrease in the direction of flow from approximately 8.8 x 10-5 to 2.3 x 10-6 m/s. In general, tritium and SF6 contents show similar space distributions, and their measurable concentrations roughly correspond to the position of the redox boundary. In some wells in spite of roughly constant tritium concentration of about 8 T.U. observed for three years, no measurable SF6 was found. That discrepancy is easily explained by one of the fitted models (dispersion model with the mean age of 158 years and the dispersion parameter of 0.5). Similarly, in another well, the SF6 data allowed to identify the most adequate model from those fitted to the tritium data. However, in general, the SF6 data yield lower ages, which is qualitatively justified by shorter travel times of that tracer through the unsaturated zone. The initial hydrodynamic and transport modelling yielded in some wells much lower and in other wells much higher SF6 and tritium concentrations than those measured. This means that hydrodynamic and transport models do not yield unique solutions unless calibrated with the tracer data. Calibration of the numerical models with the tracer data in the area of young waters is under way, but the analysis of the data available to date shows that all the main discrepancies can be explained by known or probably existing complexities of the aquifer and confining layers. For the two most distant wells, where the stable isotope and 14C data indicate a glacial age, the initial modelling with the MODPATH yielded 2 ka. However, glacial ages are easily obtainable in the model by adequate changes in the hydraulic conductivity of the permeable and semi-permeable layers. Tracer data explain the existence of three different hydrochemical zones. Young waters with distinct tritium and SF6 contents are aerobic, and of HCO3-Ca-and HCO3-SO4-Ca types. In that zone slightly increased Na and Cl contents as well as the highest concentrations of SO4 and nitrates are observed due to anthropogenic influences. Older waters are deeper, under the confining cover, in the zone of anaerobic conditions. Dissolved oxygen, nitrate and uranium contents are reduced, and Fe, Mn and NH4 contents increase in that zone. In the third zone, waters are of glacial age and of HCO3-Ca-Na or HCO3-Na types with TDS values higher than 1 g/l and Na content higher than 200 mg/l, due to either admixtures of older waters, or still going on freshening of marine sediments. (author)