AWWA WSC67774

The Water Security (WS) initiative contamination warning system (CWS) pilot deployed in Cincinnati is comprised of five monitoring and surveillance components, one of which is online water quality monitoring. While real-time water quality monitoring has the potential to detect many contaminants of concern, they do not speak to the feasibility of discriminating water quality changes attributable to contamination from normal variability in the baseline. To address this aspect of the problem, change detection algorithms, referred to as event detection systems (EDS), were utilized in the Cincinnati CWS pilot. EDS tools must be trained to the specific water quality being monitored and then tuned to balance the trade-off between false alarms and false negatives (or undetected contamination events). To facilitate this training and tuning, a large-scale evaluation study was performed in which two EDS tools were challenged with thousands of datasets containing background data from the CWS pilot and simulated contamination incidents (Umberg and Allgeier, 2007). Contamination simulations were based on laboratory experimentation and theoretical analyses that demonstrate the correlation between contaminant concentration and a change in one or more water quality parameters, including: TOC, CL2, COND, ORP, pH, and TURB. The results of this study were not only instrumental in training and tuning the EDS tools for deployment at the CWS pilot, but also illustrate fundamental aspects of the performance capabilities of water quality monitoring and event detection. This paper presents a subset of the results from this large-scale evaluation study, focusing on the ability of the two EDS tools under study to detect different contaminant types. Includes 9 references, tables, figures.