Groundwater abstraction wells are commonly protected by zones of restricted land use. Such well protection areas typically cannot cover the entire well catchment, and numerous risk sources remain. Each risk source could release contaminants at any time, affect the well earlier or later, and thus put the quality of supplied water at risk. In this context, it seems fortunate that most well catchments are equipped with monitoring networks. Such networks, however, often grew historically while following diverse purposes that changed with time. Thus, they are often inadequate (or at least suboptimal) as reliable risk control mechanism. We propose to optimize existing or new monitoring networks in a multi-objective setting. The different objectives are minimal costs, maximal reliability in detecting recent or future contaminant spills, and early detection. In a synthetic application scenario, we show that (1) these goals are in fact competing, and a multi-objective analysis is suitable, (2) the optimization should be made robust against predictive uncertainty through scenariobased or Monte Carlo uncertainty analysis, (3) classifying the risk sources (e.g., as severe, medium, almost tolerable) is useful to prioritize the monitoring needs and thus to obtain better compromise solutions under budgetary constraints, and (4) one can defend the well against risk sources at unknown locations through an adequate model for the residual risk. Overall, the concept brings insight into the costs of reliability, the costs of early warning, the costs of uncertainty, and into the trade-off between covering only severe risks versus the luxury situation of controlling almost tolerable risks as well.

Collaboration between academics and practitioners promotes knowledge transfer between research and industry, with both sides benefiting greatly. However, academic approaches are often not feasible given real‐world limits on time, cost and data availability, especially for risk and uncertainty analyses. Although the need for uncertainty quantification and risk assessment are clear, there are few published studies examining how scientific methods can be used in practice. In this work, we introduce possible strategies for transferring and communicating academic approaches to real‐world applications, countering the current disconnect between increasingly sophisticated academic methods and methods that work and are accepted in practice. We analyze a collaboration between academics and water suppliers in Germany who wanted to design optimal groundwater monitoring networks for drinking‐water well catchments. Our key conclusions are: to prefer multiobjective over single‐objective optimization; to replace Monte‐Carlo analyses by scenario methods; and to replace data‐hungry quantitative risk assessment by easy‐to‐communicate qualitative methods. For improved communication, it is critical to set up common glossaries of terms to avoid misunderstandings, use striking visualization to communicate key concepts, and jointly and continually revisit the project objectives. Ultimately, these approaches and recommendations are simple and utilitarian enough to be transferred directly to other practical water resource related problems.

Groundwater wells are often protected by restricted land use within wellhead protection zones. Unfortunately, one cannot restrict land use in the entire catchment (especially in urban areas), and there is uncertainty in wellhead delineation. Thus, nearly all well catchments have an entire inventory of risk sources. Each of these risk sources may fail at any time, release contamination and affect the well earlier or later. In fact, most catchments are equipped with some form of monitoring network. Such networks, however, often grow historically, follow various purposes that changed over time, and thus are often suboptimal (if not even inadequate) for rigorous risk control. In this work, we propose a concept to plan monitoring networks through multi-objective optimization. The different objectives are minimal costs, maximal probability to detect all possible contaminants once they entered the aquifer, and earliest possible detection. Also, risk sources that are classified as severe versus medium or tolerable should be treated with different priorities. Therefore, we propose to treat detection probability and early-warning time as separate objectives for each risk class. The concept will allow catchment managers to obtain optimal monitoring networks for risk control, and to gain insight into the costs of certainty, the costs of early warning, and the costs of covering top risks versus the luxury situation of controlling even minor risks.

Optimal design of groundwater monitoring networks is challenging due to (1) conflicting objectives for assessing the performance of candidate monitoring networks, (2) uncertainty in system dynamics and hydrogeological context, and (3) the large decision space of possible monitoring‐well positions (also termed the search space). The immensity of the search space poses a significant challenge for modern multiobjective optimization tools. This study introduces two approaches that improve the efficiency and effectiveness of evolutionary multiobjective optimization tools when solving monitoring design problems. We show how a careful mathematical representation of the monitoring design search space and reductions of possible monitoring‐well positions enhance the solution and attainment of decision‐relevant multiobjective trade‐offs in monitoring quality. We demonstrate the value of our improved representation and reduction techniques on a three‐objective monitoring network design problem focused on urban source water protection (termed the U_Protect benchmarking problem). U_Protect abstracts a real‐world case study within an urban drinking‐water well catchment, including inaccessible and restricted areas for monitoring‐well installation, and random heterogeneities in the conductivity field. Our representation and reduction methods significantly enhance the effectiveness, efficiency, and reliability of the optimization. Our proposed framework shifts focus to the most impactful monitoring design decisions while also enhancing decision makers understanding of key performance trade‐offs. In combination, our proposed representation and reduction techniques have significant promise for enhancing the size and the scope of combinatorial monitoring problems that can be explored.

No abstract