Changepoint Analysis of Natural Attenuation in Groundwater Improves Forecasts of Time to Attain Goal

Abstract

When a remedy is selected for groundwater contamination, the monitoring record is often evaluated to extract rate constants for attenuation of contaminants over time, and the rate constants are used to forecast a time when the concentrations will attain a clean-up goal. These evaluations typically assume the rate constant does not change. Data from 11 monitoring wells at the former Twin Cities Army Ammunition Plant (TCAAP) were evaluated to test whether this assumption was valid for this site. A previous evaluation at this site (based on data from 1987 through 1999) extracted rate constants that would bring the concentrations of trichloroethylene (TCE) in the 11 wells to the maximum contaminant level (MCL) on or before 2013. By 2020, only four wells had reached the MCL. Piecewise linear regressions were used to model the relationship between time elapsed and concentration of TCE for each well, and to identify any changepoints in the rate of attenuation over time. Each well had at least one changepoint with different TCE attenuation rates on either side of the changepoints. The slope of the most recent segment after the last changepoint provides the best information to forecast concentrations in the future. For four of the wells, that forecast indicated that concentrations of TCE would never reach the MCL. Piecewise linear regression analysis proved to be a valuable tool to detect changes in rate constants, and to update forecasts of the time required for groundwater concentrations to reach a cleanup goal.