Automated β-D-glucuronidase activity monitoring for identification of combined sewer overflows driving microbial water quality degradation

• Automated on-site monitoring of GUS activity was used to identify intermittent CSOs driving water quality degradation at drinking water intakes. • A Random Forest model using CSO, physico-chemical and hydrometeorological data accurately predicted GUS activity. • CSO discharge duration and cumulative effects from multiple CSOs contributed to peak GUS activity. • Electrical conductivity at drinking water intakes was associated with CSO events and GUS activity peaks. • GUS activity monitoring was more reliable than a microbial threat index and routine E. coli monitoring for identifying priority intermittent contaminant sources. The large number and variability of Combined Sewer Overflow (CSO) discharges in urban areas represent an important source of pollution, contributing to the deterioration of receiving water quality, particularly where it serves drinking water supplies. In the context of increased precipitation intensity because of climate change, there is a need to properly identify CSOs having the greatest impact on source water quality either individually or simultaneously while considering cumulative effects. Automated on-site monitoring (AOM) of β-D-glucuronidase (GUS) activity was implemented for 1.5 years at two urban drinking water intakes (DWIs) to assess the impact of upstream CSO discharges on source water quality. Prolonged peaks of fecal contamination were observed during early spring, typically lasting an average of 2-3 consecutive days (longest up to 6-12 days), and were primarily associated with snowmelt and/or rainfall events. GUS activity at the DWIs was highly correlated with intermittent CSO events. The discharge characteristics of a subset of CSOs with regards to overflow duration and total number of simultaneous overflows emerged as primary drivers of fecal contamination peaks. The most problematic CSOs affecting DWIs were identified based on the 95 th percentile GUS activity peaks. This novel approach using high frequency microbial water quality data provides essential information for targeting mitigation strategies towards the development of effective source water protection actions.