Multi-Objective Reservoir Management under Environmental Constraints: Hydropeaking and Thermal Impacts in Alpine Rivers

Multi-objective reservoir optimization plays a pivotal role in managing water scarcity and growing uncertainty in hydrology under climate change, especially in sensitive mountain environments. While these frameworks are effective in weighing competing uses such as hydropower production and water supply, they often provide a aggregated representation of environmental impacts. In particular, operationally driven alterations of flow regimes (hydropeaking) and downstream water quality, especially water temperature, are seldom incorporated as explicit objectives, despite representing some of the most critical stressors on riverine ecosystems.

Hydropeaking, arising from rapid sub-daily variations in turbine releases, is one of the most severe anthropogenic stressors in regulated Alpine rivers, impacting habitat availability, fish behavior and survival, and benthic communities. In parallel, reservoir operations substantially modify downstream water temperature through flow regulation and withdrawal, directly influencing dissolved oxygen, metabolic processes, and habitat suitability. Although these pressures operate through different mechanisms and timescales, both are directly controlled by reservoir management decisions.

To explicitly incorporate these ecosystem challenges, we develop a sub-daily simulation and optimization framework that integrates both hydropeaking and thermal dynamics directly into operational planning. Thermal dynamics are simulated using a one-dimensional, density-stratified Lagrangian model, which resolves the vertical thermal structure of the reservoir and its impact on release temperature with limited computational burden. Environmental objectives include minimizing (i) hydropeaking metrics that quantify the magnitude and frequency of sub-daily flow fluctuations, and (ii) downstream water temperature exceedance from natural conditions. These are optimized jointly with objectives related to hydropower revenue and irrigation reliability.

The framework is applied to the Ceresole reservoir (North-West Italy) using a closed-loop optimization approach. Policies are optimized using an Evolutionary Multi-Objective Direct Policy Search (EMODPS), permitting adaptive decision-making that responds dynamically to system states rather than prescribing pre-settled release trajectories.

Results show that extensive accounting for both hydropeaking and thermal objectives leads to tangibly different optimal operating strategies compared to traditional formulations, revealing clear trade-offs as well as non-obvious synergies between economic and ecological goals. The proposed framework provides a transparent and transferable approach for integrating operationally relevant environmental constraints into reservoir optimization, supporting more ecosystem-oriented hydropower management in Alpine river systems.