An extensive analysis of water samples from wells across the eastern US links per- and polyfluoroalkyl substances (PFAS) in groundwater to the presence of other chemicals and various land uses. The data could help create models to predict what regions are high risk for PFAS contamination (Environ. Sci. Technol. 2022, DOI: 10.1021/acs.est.1c04795).
Exposure to PFAS has been linked to serious health conditions and is widespread. But identifying contaminated sites is slow and difficult because detecting the fluorinated compounds requires expensive instruments and specialized laboratory protocols to prevent sample contamination. To begin regulating PFAS as a drinking water contaminant, testing randomly sampled wells isn’t enough, says Cindy Hu, a data scientist at the research firm Mathematica who was not involved in the study. “We have to have a framework to know where to focus our resources.”
In the new study, conducted as part of the US Geological Survey’s National Water-Quality Assessment Project, USGS scientists led by Peter B. McMahon collected 254 well samples from five aquifer networks in the eastern US and tested them for 24 PFAS and a host of other chemical species. They also obtained land-use information, such as the sites’ proximities to urban and agricultural areas or to firefighter training sites, where PFAS-containing firefighting foams are commonly used. Using the massive amount of data, the researchers then built a model to find out how well various factors correlated to PFAS contamination.
The presence of tritium emerged as the unexpected top predictor of PFAS. The distance from each sampled well to a fire training site was the second-best predictor for contamination, followed by the amount of dissolved organic carbon, which could play a role in transporting PFAS in water.
“Tritium is oftentimes used as an indicator of age of groundwater,” says Andrea K. Tokranov, who led the modeling effort. Tritium—a radioactive isotope of hydrogen—in groundwater is linked to atmospheric fallout from nuclear weapons testing that started in the middle of the 20th century, so tritium concentrations can be used to track surface water that reached underground aquifers in 1953 or later. That time period happens to coincide with the introduction of PFAS in consumer goods. Thus, networks of wells that receive fresh surface water faster—and therefore have more modern groundwater—are more likely to have PFAS contamination, McMahon explains. For example, 92.9% of samples taken from a shallow well network in New England contained PFAS. All data can be found online at the National Water Information System.
The lack of data on PFAS has been a major obstacle for building robust models, Hu says, who last year developed a statistical model for predicting PFAS incidence in private wells in New Hampshire (Environ. Sci. Technol. 2021, DOI: 10.1021/acs.estlett.1c00264). One of the most significant contributions of the USGS study is that it expands the number of predictors available, she says. So existing data for some predictors, such as tritium levels in groundwater, could already be used to estimate PFAS risk, McMahon adds.