Monitoring Forever Chemicals in Rural Water Supplies: Data Analytic Tools to Predict Water Pollution Using Existing Data

The overarching goal of the proposed project is to combine newly proposed prediction methods with existing data to develop a new approach to predict PFAS concentrations in the absence of direct monitoring. This will allow for the monitoring of water quality in areas where PFAS may harm human health, food supplies and livestock. The proposed objectives and activities to achieve this goal are:Build a national and publicly available PFAS database by collecting and compiling data from states and localities that implemented voluntary or mandatory PFAS monitoring. Compile and combine systematically collected data that may be correlated with PFAS concentrations. These include PFAS sources (e.g. wastewater treatment plants), water quality measures for contaminants regulated under the SDWA, and weather. Evaluate and quantify the performance of recently developed data analytic tools in econometrics and machine learning in predicting PFAS levels. Train and validate models in settings with PFAS observations to determine the combination of variables that best predict PFAS levels. Assess out of sample model performance spatially and temporally. Propose an approach that can be used by regulators to monitor PFAS in the absence of direct monitoring, and make clear the limitations and constraints of this approach. This approach can inform the targeted testing of PFAS testing in the riskiest locations, improving the cost-effectiveness of monitoring.Train one or two graduate students to pursue a research agenda that uses data analytics to improve monitoring of water pollution, and a career that combines applied econometrics with agricultural and resource economics. Expose undergraduate students to the field of and research in agricultural and resource economics.