Short high-accuracy tritium data time series for assessing groundwater mean transit times in the vadose and saturated zones of the Luxembourg Sandstone aquifer

• Published in: Hydrology and earth system sciences 2024-08, Vol.28 (15), p.3519-3547
• Main Authors: Gourdol, Laurent, Stewart, Michael K, Morgenstern, Uwe, Pfister, Laurent
• Format: Article
• Language: English
• Subjects: Accuracy; Analysis; Analytical errors; Anisotropy; Aquifers; Dating; Environmental tracers; Flow system; Geology; Groundwater; Groundwater age; Groundwater dating; Hydraulic properties; Hydrochemistry; Hydrogeology; Hydrologic cycle; Hydrology; Land use; Mean; Northern Hemisphere; Nuclear weapons; Parameter estimation; Parameter uncertainty; Precipitation; Sandstone; Sedimentary rocks; Southern Hemisphere; Time series; Tracers; Tracers (Chemistry); Transit time; Tritium; Uncertainty; Uncertainty analysis; Unsaturated flow; Vadose water; Water analysis; Water discharge; Water sampling; Water springs; Water, Underground; Weapons testing
• ISSN: 1607-7938; 1027-5606; 1607-7938
• DOI: 10.5194/hess-28-3519-2024

Among the manifold of environmental tracers at hand, tritium is the only one that can give information on groundwater age within the timescale of 100 years for the entire flow system, i.e., unsaturated and saturated. However, while in the Southern Hemisphere, a single water sample is sufficient for tritium-based young groundwater dating, several tritium measurements spanning multiple years are still needed in the Northern Hemisphere to disentangle the natural cosmogenic tritium input from that caused by the atmospheric thermonuclear weapons tests mainly carried out in the early 1960s. Although it is advised to focus tritium dating on sites where long chronicles of tritium data are available, in this study we tested the potential for short high-accuracy tritium data series (∼4 years) to date groundwater from 35 springs draining the Luxembourg Sandstone aquifer (central western Europe). We determined groundwater mean transit times using the lumped-parameter model approach in a Monte Carlo uncertainty estimation framework to provide uncertainty ranges inherent to the low number of tritium data at hand and their related analytical errors. Our results show that unambiguous groundwater mean transit time assessments cannot be determined solely based on such recent short tritium time series, given that several ranges of mean transit times appeared theoretically possible. Nonetheless we succeeded in discriminating groundwater mean transit times in the vadose and saturated zones of the aquifer through a stepwise decision process guided with several supplementary data. The mean transit time required for water to cross the vadose zone was estimated to be between 0.5±0.5 and 8.1±1.2 years depending on the spring, while for water to flow through the saturated zone, it varied from 5.7±2.4 to 18.9±4.6 years (median ± half of the 5–95 percentile range). Our findings are consistent with both the tritium measurements of individual springs and the hydrogeological context of the study area. We specifically corroborated the dating results using the known hydrogeological properties of the Luxembourg Sandstone aquifer, the hydrochemistry of the studied springs, and their discharge dynamics. When translated into water velocities (which average ∼12 and ∼170 m yr−1 for the vadose and the saturated zones, respectively), the tritium dating results mirrored the horizontal–vertical anisotropy of the aquifer's hydraulic properties caused by the bedded character of the Luxembourg Sandston