A multi-method for identifying and spatial modeling of groundwater nitrate natural background levels in an urbanized delta in China: Insight from Cl/Br ratios combining with KNN algorithm and EBK

Nitrate is one of the most widespread contaminants in groundwater, and natural background levels (NBLs) identification is essential for setting regional pollution thresholds and supporting groundwater quality management. Classical approaches show limitations in sample identification and spatial modeling, which make it difficult to depict the spatial heterogeneity and uncertainty of NBLs accurately. Therefore, a coupled multi-method approach in this study integrating Cl/Br ratio, k-nearest neighbor (KNN) algorithm, and empirical Bayesian kriging (EBK) was established to identify nitrate NBLs in various groundwater units of the Pearl River Delta (PRD), which has a large scale of urbanization, and to quantify their spatial heterogeneity and exceedance probability. Results showed that the alluvial-proluvial unit (Unit II) exhibited the highest NBLs-NO 3 - (21.15 mg/L) with relatively strong spatial continuity, the fissured unit (Unit III) showed low concentrations (11.45 mg/L) but significant spatial variability, while the coastal alluvial unit (Unit I) displayed the lowest concentrations (9.55 mg/L) with notable local fluctuations. These may be attributed to the relatively good permeability of the media in Unit II, the diversity of groundwater recharge sources, and the relatively oxidizing environment that limits nitrate attenuation. Unit III may be affected by discontinuities in groundwater flow paths controlled by irregular fractures. Unit I may be influenced by the low permeability of the deposits and the presence of localized areas with much organic matter that promote denitrification under reducing conditions. EBK modeling effectively depicted spatial variation and quantified uncertainty, thus evaluating the local probability of exceeding the nitrate natural background levels. This Cl/Br-KNN-EBK integrated approach provides a new perspective on a method for groundwater NBLs identification and can provide scientific support for groundwater contamination and environmental risk assessment.