Incorporating model uncertainty into the evaluation of interventions to reduce microcontaminant loads in rivers

Abstract

Models of microcontaminant fate and transport in wastewater treatment plants (WWTPs) and rivers have been developed and used to assist decision-making in the field of water management. These models come with parameter uncertainties that must be properly incorporated in the decision-making process. The main goal of this study is to evaluate how the magnitudes of key model parameter uncertainties influence the selection of end-of-pipe interventions (at WWTPs) designed to reduce the microcontaminant loads in rivers. We developed a model that describes the fate and removal of pharmaceuticals in WWTPs and the river network based on 3 key parameters: human pharmaceutical consumption and excretion (F) and the pharmaceutical degradation constants in WWTPs (k) and rivers (k). We modelled the fate and transport of diclofenac in the Llobregat River basin (NE Spain). We calibrated the model using a Bayesian approach, which resulted in an accurate prediction of measured diclofenac loads at 9 locations along the Llobregat River and at the influents and effluents of 2 WWTPs (R = 0.95). Using different scenarios, we evaluated three levels of uncertainty in the key model parameters. The first level of uncertainty corresponded to the reference distributions obtained from the Bayesian calibration. Then, for each parameter, we generated a narrower PDF (decreased uncertainty with respect to the reference) and a wider PDF (increased uncertainty). For each level of uncertainty, we evaluated increasing removal efficiencies of diclofenac at the WWTPs, from 38% to 98%. We assumed that removal efficiencies of up to 75% can be achieved by upgrading secondary treatment; beyond 75%, tertiary treatment is needed. The scenario analysis showed that achieving diclofenac removal efficiencies corresponding to tertiary treatment results in apparent concentration reductions (statistically significant differences relative to the reference situation), regardless of the level of uncertainty applied to the model parameters. However, upgrades in the secondary treatment resulted in apparent reductions only in the case of reduced uncertainty. We concluded that model uncertainty greatly influences the decisions that river basin authorities must make to reduce the microcontaminant loads released by WWTPs into rivers. In addition, we discussed research priorities to help reduce model uncertainty and thereby make more appropriate decisions.

Growth of the Decision Tree: Advances in Bottom‐Up Climate Change Risk Management

“…; Gimeno et al. ), and so are not able to adequately evaluate risks of extended contaminant plume exposure. Risk assessment in contaminant transport using two‐ or three‐dimensional model formulations have predominated in applications relating to the subsurface (e.g., Gharamti et al.…”

“…This indicates the importance of the two-dimensional CFD formulation, and a well-parameterized hydrologic model. Previous approaches to risk assessment or uncertainty analysis in problems of contaminant transport in surface water applications have been almost exclusively of the one-dimensional type (e.g., Hou et al 2014;Gimeno et al 2017), and so are not able to adequately evaluate risks of extended contaminant plume exposure. Risk assessment in FIGURE 6.…”

Spatial Modelling of Micro-pollutants in a Strongly Regulated Cross-border Lowland Catchment

“…First, it can be seen, that measured concentrations at the individual sites vary considerably for both conditions due to natural flow variability. For this reason, a maximum deviation of factor three is generally accepted as quality criterion for good agreement between monitoring and modelling results (Ort et al 2009;Verlicchi et al 2014;Gimeno et al 2017). This region is displayed as shaded area around the mean PEC in Fig.…”

Can source control of pharmaceuticals decrease the investment needs in urban wastewater infrastructure?

“…The model developed in Gimeno et al (2017) was used to describe the fate and removal of diclofenac and naproxen in the Llobregat river basin. The tool integrates 3 submodels: 1) a substance-human consumption and excretion model, which estimates the diclofenac and naproxen loads that reach the influents of WWTPs; 2) a WWTP model, which estimates the effluent loads; and 3) a river model, which estimates the loads in every river stretch.…”

“…Hence, the future of environmental protection involves the combination of source control with end-of-pipe measures (Eggen et al, 2014;Hillenbrand et al, 2014;van Wezel et al, 2017;Kümmerer et al, 2018). The cost and effectiveness of end-of-pipe measures to reach specific EQS has already been evaluated at a catchment level using model-based approaches (e.g., Gimeno et al, 2017;Kehrein et al, 2015;Ort et al, 2009). However, the effectiveness of source control measures in reducing needs for WWTP upgrading has been poorly addressed in the literature.…”