Integrating risk prioritization and interpretable machine learning to inform PFAS management in an urban aquatic system.

Journal: Journal of environmental management
Published Date:

Abstract

Per- and polyfluoroalkyl substances (PFASs) pose persistent challenges to urban environmental management due to their complex sources, ongoing substitution, and contrasting behavior across environmental media. This study assessed the contamination characteristics, spatial distribution, and risk prioritization of legacy and emerging PFASs in surface water and sediment of Shanghai, a representative megacity. A total of 38 PFASs were detected in both media, with mean concentrations of 120.97 ng/L in surface water and 33.90 ng/g in sediment. Surface water was dominated by short-chain and emerging PFASs, particularly PFBS, reflecting regulatory-driven substitution. In contrast, sediment was enriched in long-chain PFASs, with PFOA remaining predominant due to strong historical legacy effects. Spatial analysis revealed marked heterogeneity in PFAS contamination, with localized hotspots closely associated with urban activity intensity, river confluences, and point-source influences. Toxicological Priority Index (ToxPi) risk prioritization identified PFOA as the dominant risk driver in both surface water and sediment, while several emerging and short-chain PFASs contributed substantially to mixture-based risks in highly urbanized areas. Interpretable machine learning further indicated that precipitation and proximity to sewage outfalls are key spatial indicators governing PFAS variability in surface water, underscoring the central role of wastewater infrastructure and hydrological processes. Overall, effective PFAS management in megacities requires a media-specific and risk-prioritized approach that addresses both ongoing emissions and long-term sedimentary legacy contamination.

Authors

Keywords

No keywords available for this article.