Comparing probabilistic microbial risk assessments for drinking water against daily rather than annualised infection probability targets.

Some national drinking water guidelines provide guidance on how to define 'safe' drinking water. Regarding microbial water quality, a common position is that the chance of an individual becoming infected by some reference waterborne pathogen (e.g. Cryptsporidium) present in the drinking water should < 10(-4) in any year. However the instantaneous levels of risk to a water consumer vary over the course of a year, and waterborne disease outbreaks have been associated with shorter-duration periods of heightened risk. Performing probabilistic microbial risk assessments is becoming commonplace to capture the impacts of temporal variability on overall infection risk levels. A case is presented here for adoption of a shorter-duration reference period (i.e. daily) infection probability target over which to assess, report and benchmark such risks. A daily infection probability benchmark may provide added incentive and guidance for exercising control over short-term adverse risk fluctuation events and their causes. Management planning could involve outlining measures so that the daily target is met under a variety of pre-identified event scenarios. Other benefits of a daily target could include providing a platform for managers to design and assess management initiatives, as well as simplifying the technical components of the risk assessment process.

SCADA data and the quantification of hazardous events for QMRA.

The objective of this study was to assess the use of on-line monitoring to support the QMRA at water treatment plants studied in the EU MicroRisk project. SCADA data were obtained from three Catchment-to-Tap Systems (CTS) along with system descriptions, diary records, grab sample data and deviation reports. Particular attention was paid to estimating hazardous event frequency, duration and magnitude. Using Shewart and CUSUM we identified 'change-points' corresponding to events of between 10 min and >1 month duration in timeseries data. Our analysis confirmed it is possible to quantify hazardous event durations from turbidity, chlorine residual and pH records and distinguish them from non-hazardous variability in the timeseries dataset. The durations of most 'events' were short-term (0.5-2.3 h). These data were combined with QMRA to estimate pathogen infection risk arising from such events as chlorination failure. While analysis of SCADA data alone could identify events provisionally, its interpretation was severely constrained in the absence of diary records and other system information. SCADA data analysis should only complement traditional water sampling, rather than replace it. More work on on-line data management, quality control and interpretation is needed before it can be used routinely for event characterization.

Pathogen monitoring offers questionable protection against drinking-water risks: a QMRA (quantitative microbial risk analysis) approach to assess management strategies.

Risk mitigation provided by human monitoring and control over a water supply system has been consistently overlooked when estimating pathogen exposure to consumers. The Systems-Actions-Management (SAM) framework lends itself neatly to Quantitative Microbial Risk Assessment (QMRA) as one way to establish this link. The general premise is that an organisational protocol will influence how a human controller behaves, in turn influencing the system performance. For illustrative purposes, the framework was applied to a hypothetical water supply system to quantify the risk reduction offered by routine Cryptosporidium monitoring and the response to oocyst 'detects'. Our findings suggest that infrequent direct pathogen monitoring may provide a negligible risk barrier. The practice of sampling treated water to verify microbiological integrity is also dubious: oocyst densities were largely under-estimated, in part due to the spatial dispersion of oocysts in the waterbody, but predominantly from imperfect detection methods. The development of 'event-driven' monitoring schemes with barrier performance-based treatment verification methods, as promoted in new guidelines, is supported as a pressing issue to reduce the likelihood of undetected pathogen passage through a treatment plant.

Quantifying the impact of runoff events on microbiological contaminant concentrations entering surface drinking source waters.

Concentrations of microbiological contaminants in streams increase during rainfall-induced higher flow 'event' periods as compared to 'baseflow' conditions. If the stream feeds a drinking water reservoir, such periods of heightened pathogen loads may pose a challenge to the water treatment plant and subsequently a health concern to water consumers downstream. In order to manage this risk, it is desirable to first quantify the differences in surface water quality between baseflow and event conditions. The Event Mean Concentration (EMC) is a flow-weighted average concentration of a contaminant over the duration of a single event, proposed here as a standard parameter for quantifying the net effect of events on microbial water quality. Application of the EMC concept was assessed using flow and quality data for several events from an urbanised catchment. Expected mean EMCs were significantly larger than expected mean baseflow concentrations (p-value< or =0.012) for three microbial agents - Escherichia coil (13,000 [n = 7] v. 610 [n = 16] mpn/100 ml), Cryptosporidium (234 [n = 6] v. 51 [n = 16] oocysts/10 litres) and Campylobacter (48 [n = 5] v. 2.1 [n = 16] mpn/100ml). These parameter estimates were complemented by estimating data variability and uncertainty in the form of second-order random variables. As such the results are in a format appropriate for potential use as components in probabilistic risk assessments evaluating the effect runoff events have on drinking water quality.