Pumping Optimization Under Uncertainty in an Island Freshwater Lens Using a Sharp-Interface Seawater Intrusion Model

Abstract

Pumping optimization under uncertainty is a powerful approach for the management of groundwater resources, and its implementation would be valuable in island aquifers where freshwater lenses are affected by seawater intrusion. Sharp-interface numerical models are especially well suited for the task as they offer fast simulation times, but to date they have not been used because of a lack of guidelines and due to the specific challenges associated with this approach. This study presents a methodology for pumping optimization under uncertainty for island freshwater lenses using a sharp-interface model (MODFLOW-SWI2) and demonstrates it for a real case (Magdalen Islands, Quebec, Canada). The total pumping in a well field was maximized while avoiding well salinization due to upconing. To do so, the sharp interface simulated below the pumping wells was corrected successively for cell-to-well upconing and for dispersion. Pumping optimization under uncertainty was then conducted using PESTPP-OPT, considering parameter and observation uncertainty, and was repeated for 23 reliability levels to illustrate a large range of risk-averse, tolerant and neutral stances. The maximum pumping was obtained as a function of risk of well salinization. This approach enabled quantification of the tradeoffs between pumping and risk, allocation of pumping amongst wells, and an examination of the ability of the well field to meet the water demand while maintaining an acceptable level of risk. Ultimately, this framework allows groundwater managers to select the final pumping scenario themselves, depending on their attitude toward risk.

Key Points

• A sharp-interface model enabled the application of computationally-efficient pumping optimization under uncertainty in a freshwater lens

• Limitations of the sharp-interface approach were overcome by using an analytical correction for dispersion below pumping wells

• A highly-transferrable framework was created for obtaining optimal pumping scenarios which reflect user-specified risk tolerance

Plain Language Summary

Fresh groundwater is an important source of drinking water in island environments, but its proximity to seawater makes it vulnerable to salinization. In these areas, fresh groundwater is often contained in a lens that overlies saline groundwater. Pumping within the freshwater lens to extract drinking water causes saltwater to migrate toward the pumping well, a phenomenon called saltwater upconing. As a result, water managers need to balance meeting the water demands of the local community with avoiding saltwater contamination of local drinking water supplies. Computer simulations of groundwater flow are commonly used to inform groundwater management decisions, but the uncertainties associated with these simulations are usually neglected, often resulting in the consideration of a single management solution. To improve water management strategies in freshwater lenses, we developed a new technique to evaluate the maximum amount of freshwater that can be safely pumped from the lens while recognizing the impact of simulation uncertainties on management solutions. Multiple pumping rate scenarios were generated, each associated with a different risk of well contamination between 0% and 100%. This method allows local water managers to then choose their preferred pumping scenario depending on the level of risk that they consider to be acceptable.