Multi-scale variance partitioning reveals hidden regional connectivity in groundwater contamination: Implications for drinking water security

Abstract

Rural communities dependent on groundwater face increasing contamination risks, yet large-scale assessments of actual drinking water sources remain rare. This study pioneers a novel geostatistical framework to quantify contamination patterns using 2349 groundwater wells that serve as primary drinking water sources for populations in unsupplied areas of Chungcheongnam-do, South Korea. Our integrated analysis revealed that NO₃⁻-N is the most pressing concern with 16.1 % of wells exceeding the national drinking water standard. Our analysis revealed unprecedented spatial contamination architecture. Nitrate demonstrated spatial coherence extending 62 km, vastly exceeding the <20 km ranges observed for trace elements. Variance partitioning quantified that neighboring wells contribute 38–40 % to nitrate variability at any location, indicating substantial hydraulic interconnection across the regional aquifer system. Local Indicators of Spatial Association identified six agricultural townships as contamination hotspots where mean nitrate concentrations reach 64.7 mg L⁻¹—over six times the safe drinking water limit. These hotspots exhibited 68 % cropland coverage compared to 33 % in non-hotspot areas. These findings transform understanding of groundwater contamination from local to regional phenomena, necessitating watershed-scale management rather than well-by-well remediation to protect rural drinking water supplies.