Arash Massoudieh Christopher Green Kang-Kun Lee Dugin Kaown YeoJin Ju 2021 <div id="ab0005" class="abstract author" lang="en"><div id="as0005"><p id="sp0050">Aquifer flow systems near seawater interfaces can be complicated by density-driven flows and the formation of stagnation zones, which inevitably introduces uncertainty into groundwater age-dating. While age-dating has proved effective to understand the seawater intrusion and aquifer salinization process in coastal aquifers, further efforts are needed to propagate model and data uncertainty to the uncertainty associated with the inferred age distributions. This study was performed in a coastal aquifer located close to the Yellow Sea, South Korea, where there is a decreasing trend of groundwater levels due to recent heavy exploitation, raising a warning of induced seawater intrusion. We inferred the groundwater age distributions in wells around the intrusion zone and estimated the uncertainty associated with the inference based on multiple age tracers including<span>&nbsp;</span>³H, tritiogenic<span>&nbsp;</span>³He, radiogenic<span>&nbsp;</span>⁴He, CFC-11, CFC-12 and CFC-113 using Bayesian inference. We examined various models representing the age distributions including traditional parametric Lumped Parameter Models (LPMs) and two non-parametric “shape-free” models. The results showed that the mean ages at the study site ranged from 10.9 to 522.5 y. Complex, multimodal distributions of ages occurred near a seawater intrusion area and upland recharge zones, implying converging paths of a wide range of different ages in those regions. In particular, the age distributions estimated near the seawater intrusion interface were characterized by heavy-tailed mixing structures with elevated concentrations of<span>&nbsp;</span>⁴He. This likely indicates density-driven upward flow at the seawater intrusion interface, forcing old groundwater rich in<span>&nbsp;</span>⁴He into the shallow aquifer. The Bayesian inference estimated large uncertainties particularly for the old age distributions, which was attributed partly to the gradual accumulation of<span>&nbsp;</span>⁴He in groundwater. The Bayesian inference improved understanding of flow dynamics at a complex seawater interface and identified opportunities to further reduce uncertainty of old water age estimates that characterize upwelling groundwater near the interface.</p></div></div> application/pdf 10.1016/j.scitotenv.2020.141994 en Elsevier Complexity of groundwater age mixing near a seawater intrusion zone based on multiple tracers and Bayesian inference article