Coastal Groundwater Management : Seawater Intrusion Prevention, Artificial Recharge, and Climate Adaptation

Groundwater is experiencing substantial overdrafts and bearing ever-mounting freshwater demand. Existing groundwater management strategies are myopic and fail to orchestrate production and the operation of protection approaches, including managed aquifer recharge (MAR) and seawater intrusion barriers (SWIBs). Motivated by the urgency of sustainable groundwater management, we investigate how to optimize the joint operations of groundwater production, replenishment (via MAR), and protection (by injecting freshwater through SWIB). We model a central planner's decision on groundwater production, artificial replenishment, and freshwater injection quantities as a discrete-time, finite-horizon Markov decision process (MDP). We prove that the optimal groundwater production policy in the absence of MAR and SWIB follows a threshold policy that is independent of the groundwater level. However, under the joint operation of MAR and SWIB, the optimal groundwater production policy depends on the groundwater level and climate state. We calibrate our model by combining real data sets of freshwater demand, groundwater production, and the operations of MAR and SWIB in Orange County, CA. We find that SWIBs are more successful than MAR in enhancing groundwater operational flexibility, but their efficacy critically relies on natural replenishment quantities. In contrast, jointly operating MAR and SWIB achieves the lowest total costs, enlarges the operational flexibility, and stabilizes the operating range under varying climate conditions. Our results underscore the importance of MAR and SWIB in stabilizing groundwater levels, particularly when droughts become more acute or prolonged as a result of climate change. Nonetheless, the potential lack of water supply for MAR can significantly throttle the efficacy of SWIB and MAR in terms of operational flexibility. Our results indicate that groundwater planners should carefully evaluate the unit operating cost and the available amount of freshwater for MAR in practice. Our model and analyses also provide a handy tool for optimally automating the management of groundwater systems