Historical, seasonal, and spatial variations in chlorination disinfection byproducts within Seoul's drinking water distribution system: A decade-long assessment

Abstract

Seoul, the most densely populated metropolitan city, provides drinking water to 9.4 million people and 4.1 million households. This study investigated the historical, seasonal, and spatial variations of chlorine disinfection byproducts (DBPs) in drinking water distribution systems, focusing on the primary contributors, from 2014 to 2024. Seasonal variations were found in total trihalomethanes (THMs) concentrations, with the highest levels in Summer (34.2 +/- 9.6 mu g L-1). THMs, particularly chloroform, contributed the most to seasonal fluctuations. Haloacetonitriles (HANs) and chloral hydrate (CH) were also slightly higher during Summer, whereas haloacetic acids (HAAs) showed the opposite trend, with the highest levels observed in Winter (5.9 +/- 3.0 mu g L-1). Spatial analysis revealed that DBP concentrations at consumer taps were 1.2-8.0 times higher than at water treatment plants, likely due to extended water age and further reactions between chlorine and natural organic matter in distribution pipes and storage tanks. Elevated water temperatures were strongly correlated with THM levels, while stagnation predominantly influenced CH concentrations. Machine learning models, particularly ensemble methods, were employed to predict DBP concentrations. The models showed moderate accuracy for THMs (R-2 = 0.60) but limited accuracy for HANs (R-2 < 0.12). Feature importance analysis highlighted water temperature as the most critical factor for THM prediction, while pipe distance was significant for HAAs, HANs, and CH. These findings underscore the need to address seasonal and operational factors in DBP management. As global temperatures rise, drinking water utilities must adopt innovative strategies to mitigate DBP formation while ensuring drinking water safety. This study provides valuable insights for enhancing risk management in drinking water distribution systems.