Search

Combined sewer overflows (CSOs) are a source of microbial contamination of drinking water intakes located downstream from their discharge. To safeguard the quality of the source water, it is essential to evaluate the risk levels associated with these municipal structures. This study compares two risk assessment approaches to test their applicability for assessing the risk of CSOs to drinking water intakes in a highly urbanized watershed. The first approach was based on a deterministic equation that combines the characteristics of an overflow structure allowing the risk to be rated as very low, low, medium, high, or very high. The second probabilistic risk assessment approach yielded findings that are probabilistically distributed across the five levels of risk. This approach was developed by constructing a novel Bayesian network to probabilistically link the different factors defining the exposure of water intakes to the hazards of CSOs. The comparison between the results of these two approaches highlighted the importance of simultaneously considering many scenarios for assessing the risk of contamination of source waters. It was possible to use the Bayesian network rather than the deterministic equation, which only supports one scenario at a time. It was also shown that the deterministic approach often overestimated risk levels for CSO outfalls close to the water intake. This occurred because the assessment process emphasized the distance factor between the discharge point and the water intake, while neglecting other crucial characteristics of the overflow, such as duration and frequency. In particular, the deterministic approach tended to underestimate risk for CSOs associated with low overflow frequencies as it did not support scenarios of overflow duration, unlike the probabilistic approach. The validation and sensitivity analysis of the Bayesian model revealed that the population residing in the CSO's drainage basin, along with the frequency and duration of the overflows, exerted the greatest influence on the resulting risk levels. These factors outweighed other variables utilized in the risk assessment, including vulnerability of the drinking water intake, the type of overflow recorder, pipe diameter, and variables defining the exposure of the water intake to the discharge. In the context of implementing action plans, the Bayesian network is estimated as a cost-effective technique as it prioritized overflow structures needing special attention in a highly urbanized watershed, where the same CSOs were deterministically rated as having the same risk level. The results also demonstrated the effectiveness of the Bayesian model in addressing data gaps faced by water managers and stakeholders. The Bayesian model proved capable of assessing risks with uncertainties for CSOs, even with limited input data available. These findings can assist managers in identifying problematic structures by considering various scenarios, unlike the deterministic approach, which left almost half (n = 42) of the study site's overflow structures unassessed due to data limitations.

Environmental concentrations of 25 pesticides in superficial water were employed to conduct an ecological risk assessment (ERA) in a mixed land-use basin utilized as a drinking water source. A deterministic risk assessment (RQ) was utilized to evaluate the chronic risk to aquatic biota, while a probabilistic risk assessment (PRA) approach was applied to assess the acute and chronic risk in the most sensitive species and at the community level. A high risk was identified for insecticides (pyrethroids, organophosphates and organochlorines). RQs ranged from 4.0e -4 (2,4-D) to 105.3 (ethion) considering median concentrations and from 8.0e -4 (2,4-D) to 230 (p,p'-DDT) considering extreme concentrations. Temporal variation in ΣRQs showed the highest risk during spring and summer months, which is related to the crop calendar and land use in the Laguna del Cisne basin. For PRA, the probability of exceeding the hazardous concentration HC5 (5th percentile) was higher for the most sensitive species in chronic exposure, especially for cypermethrin (38.9 %), permethrin (25.6 %), and chlorpyrifos (16.6 %). In the case of acute exposures, the probability of surpassing HC5 was higher for the entire freshwater biota, with the highest values observed for bifenthrin (28.3 %), cypermethrin (25.5 %), permethrin (11.75 %), and ethion (11.1 %). The advantages and disadvantages of PRA for assessing pesticide ecological risk were compared with the conventional deterministic RQs approach, highlighting that PRA offers improvements over the deterministic risk assessment, especially for organophosphate pesticides. Additionally, PRA provides a more comprehensive evaluation of risk for both short and long-term exposure, has the potential to incorporate others available toxicity data (e.g., LD50, Daily Intake), and utilizes different hazardous concentrations, such as HC5, HC10, and HC50. Our findings emphasize the urgent need to establish a national regulatory framework to evaluate and mitigate pesticide risks in aquatic ecosystems, especially in drinking water source like Laguna del Cisne., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023 Elsevier B.V. All rights reserved.)