PREDICTING CHANGES TO DISINFECTANT BY-PRODUCTS IN TREATED WATER IMPACTED BY WILDFIRES IN CALIFORNIA

Wildfires are known to increase the transportation of sediment, metals, nutrients, and other contaminants downstream into receiving waters. Many studies have focused on local responses to wildfires through rigorous sampling programs, providing insight into the stream and river water quality impacts over time. This study takes a human-centered approach by investigating the impact of wildfires on treated drinking water. Studies have shown that wildfires increase the concentration of some regulated contaminants, like disinfection byproducts, nitrate, and metals in drinking water. The goal of this study is to better understand the complex relationships between catchment characteristics, wildfires, and precipitation, on changes to regulated contaminants. A random forest modeling approach, using more than 100 catchment-level variables derived from national datasets (StreamCat, Measuring Trends in Burn Severity, National Hydrography Dataset, and PRISM Climate Project), was trained to predict concentrations of disinfection by-products from more than 10 years of water quality tests from water treatment plants in California. Initial results from the model indicate that it is capable of explaining a high level of variability in the disinfection byproduct concentration data, particularly in catchments impacted by wildfires. Although this study focuses on disinfection byproducts in California specifically, it was constructed in a way that readily allows water managers across the United States to utilize it in post-wildfire contaminant prediction, and, more broadly, in the prediction of any regulated contaminant (e.g. exceedances of nitrate in groundwater used as drinking water). The model suggests that burn severity and precipitation are two of the most important variables for predicting disinfection byproduct concentrations, which is a useful tool to help inform post-fire water quality management.