Using spatial data for sampling prioritization and prediction of PFAS chemicals in fish tissue in the Columbia River Basin

Abstract

Background and Aim U.S. tribes and states are challenged with identifying and characterizing the extent of per- and polyfluoroalkyl substances (PFAS) contamination and human exposure to PFAS, as well as developing effective and cost-efficient ways to prioritize site investigation. In the Columbia River Basin (CRB), particularly in and around the region’s tribal lands, the availability of PFAS occurrence data is limited and exposure is less characterized than in other parts of the country. This study aims to develop and pilot a predictive modeling workflow for prioritization of sampling locations that show potential for PFAS exposures in vulnerable populations in the CRB. Methods and Results While previous studies have mainly developed predictive models for PFAS in groundwater, this study develops and evaluates predictive models for PFAS in fish tissue. Understanding PFAS levels in fish is particularly important in CRB because it can be a major component of tribal and indigenous peoples’ diet. Existing spatial data in the region are leveraged including locations of known PFAS sources or contaminated sites and potential, yet unconfirmed, sources such as PFAS-related industry facilities, mining and refinery sites, airports, and landfills. Machine learning and statistical models employ these spatial data to predict PFAS concentrations in fish tissue in natural waters throughout the CRB and will be evaluated using empirical fish tissue occurrence data. Conclusions Model predictions help to identify areas in which high PFAS concentrations are likely to occur in fish tissue but sampling has not yet confirmed. This workflow for targeted prioritization of sampling investigations and the identification of contaminated natural resources will help U.S. tribes and states in the Columbia River Basin more efficiently manage waste, initiate clean-up activities, and identify opportunities for source reduction, thereby promoting healthy and resilient communities. Keywords PFAS, spatial analysis, predictive modeling, fish tissue, Columbia River Basin