A data-driven approach to identifying PFAS water sampling priorities in Colorado, United States

BACKGROUND: Per- and polyfluoroalkyl substances (PFAS) are persistent chemicals used widely since the mid-20th century, with some linked to adverse health effects. OBJECTIVE: To integrate known and potential PFAS sources, environmental characteristics, and existing water sampling results into a PFAS risk prediction map to inform a Colorado Department of Public Health and Environment (CDPHE) sampling prioritization plan. METHODS: Random forest classification predicted groundwater contamination risk for PFOS and PFOA at unsampled locations using existing water sampling data and spatial covariates. Predictions were categorized into low (<5 ng/L), moderate (5–35 ng/L), and high (≥35 ng/L) risk. Variable importance and population characteristics informed sampling recommendations. RESULTS: Depending on category, sensitivity and precision ranged from 58% to 90%. The approach identified private wells in specific census blocks, schools, mobile home parks, and public water systems relying on groundwater as priority sampling locations, and revealed data gaps (limited sampling coverage and under-investigated source types). IMPACT: A data-driven, statewide prediction of PFOS/PFOA groundwater risk highlights opportunities to prioritize PFAS sampling and fill information gaps, supporting efficient protection of susceptible populations in Colorado.

Kelsey E. Barton, Peter J. Anthamatten, John L. Adgate, Lisa M. McKenzie, Anne P. Starling, Kevin Berg, Robert C. Murphy, Kristy Richardson