ABSTRACT
In this study, the combined use of a Bayesian mixing model (BMM), numerical model (random walk particle tracking-RWPT), and environmental tracers (δ18O-δD, 3H, and CFC) was applied to elucidate the probabilistic contribution of the recharge sources, flow path, and residence time of groundwater across the mountainous area of Jeju Island, South Korea. Especially, the BMM ability to estimate the variable recharge contributions to the aquifer by different elevations and seasons was investigated. The δ18O-δD isotopes showed that groundwater in the study area was primarily fed by precipitation during the wet season, and the BMM estimated that wet season recharge contributed to approximately 64% of the total. The BMM-based probabilistic estimation of recharge sources revealed a mixed contribution of source waters from different elevations. A notable difference in recharge flow path was observed between highland (>450 masl) and lowland (<400 masl) wells, where the inflow of source water from the regional flow was dominant in the former and both regional and local recharges served as significant groundwater sources in the latter. Evidence from age tracers (3H and CFC-12) also supported different recharge mechanisms between highland and lowland wells. A reasonable match between the BMM- and RWPT-derived recharge contributions (RMSE 0.02-0.06) was achieved within the uncertainty ranges, with RWPT being particularly useful for capturing different flow paths between highland and lowland wells. The dynamics revealed here provide important information for establishing an improved and informed groundwater management plan for the mountainous area of Jeju Island. Ultimately, this study highlights the advantageous integrated analysis of BMM, RWPT, and environmental tracer analyses to enhance the reliability of recharge area estimation and increase the collective understanding of complex hydrogeological systems in mountainous areas.
