An integrated approach for the identification and assessment of the most critical chemical contaminant(s) at a drinking water intake has been developed. It involves the determination of a threshold or critical raw water concentration (CRWC) for target contaminants using the observed overall removal efficiency of a specific water treatment plant (WTP) and regulated drinking water concentrations for the target contaminants. The exceedance probability relative to the CRWC based on historical raw water quality monitoring data is then calculated. Finally, the integration of the raw water quality data and the overall efficiency of a particular WTP sequence allows for identification of the most critical contaminant(s) as well as an advance indication of which contaminants are most likely to challenge a plant. The proactive nature of this approach gives a utility the impetus and time to assess current treatment processes and potential alternatives. In addition, it was found that three- or four-parameter theoretical distributions are more appropriate than two-parameter probability distributions for the fitting of raw water quality data. This study reveals that the reliance on raw and/or treated water contaminant concentrations in isolation or on theoretical removals through treatment processes can, in some circumstances, be misguided.
Anomaly Detection for Raw Water Quality – A Comparative Analysis of the Local Outlier Factor Algorithm and the Random Forest Algorithms
