Mapping PFAS Exceedance Risk in China's Surface Water: A Machine Learning Approach Informed by Source Distribution

Area of Science:

• Environmental Science
• Public Health
• Chemical Contamination

Background:

• Per- and polyfluoroalkyl substances (PFAS) are persistent pollutants causing widespread surface water contamination.
• China's significant role in fluorinated chemical production necessitates understanding nationwide PFAS distribution.
• Sparse monitoring data presents a challenge for assessing national-level PFAS contamination risks.

Purpose of the Study:

• To develop a predictive model for mapping PFAS exceedance risk in China's surface water.
• To identify high-risk areas and hotspots of PFAS contamination nationwide.
• To inform targeted mitigation strategies and optimize PFAS management.

Main Methods:

• Development of a Geographically Weighted Random Forest (GWR-RF) model.
• Integration of a comprehensive, spatially explicit inventory of over 280,000 potential PFAS sources.
• Generation of a 1 km resolution spatial map of PFAS exceedance risk.

Main Results:

• The GWR-RF model showed strong predictive performance (accuracy = 0.83, ROC-AUC = 0.91).
• Significant surface water areas in China are identified as high-risk for PFAS contamination.
• Hotspots are concentrated in eastern coastal plains and industrial provinces, affecting an estimated 80-90 million people.

Conclusions:

• Proximity to PFAS users and annual precipitation are key predictors of increased contamination risk.
• The study provides crucial scientific evidence for developing targeted PFAS mitigation and management strategies.
• Vulnerable populations in high-risk areas require prioritized protection efforts.