Climate change intensifies extreme weather events, stressing water resources and demanding optimal reservoir management for competing needs.

Exploring advanced forecast-based control strategies, Lake Como serves as a multi-objective case study.

First, this seminar details a novel algorithm using probabilistic ensemble forecasts to improve decision-making, validated by blocked k-fold cross-validation.

Second, it couples neuro-evolution deep reinforcement learning with time-varying sensitivity analysis to optimize network topologies and evaluate the importance of each forecast across the year and across different compromise solutions.