The main objective of BERLIN is to provide an evidence-based foundation for a better understanding of the needs and gaps of adaptation policies by i) explicitly putting agents’ behaviors in the modeling loop to capture their intentions and preferences from observed data via Machine Learning; ii) projecting the coevolution of coupled human-natural systems under changing conditions considering co-designed storylines capturing climate risk perception and adaptation, and iii) determining the role of behavioral uncertainty in adaptation design in diverse Climate Change Hotspots, including semiarid regions, river deltas, and snow-dependent river basins.
Modeling behaviors via machine learning
We replace heuristic reservoir operating rules with data-driven behavioral representations by applying Inverse Reinforcement Learning to large-scale observational datasets. This provides an empirical foundation to simulate the evolution of coupled human–natural systems and quantify real-world trade-offs among competing water demands.
Modeling behaviors via social learning
We place stakeholders at the center of the project to understand how their needs shape reservoir operations and water management. Through co-designed participatory and social-learning processes, we explore climate risk perception and adaptation, generating insights to simulate future stakeholder behavior and strengthen resilient decision-making.
Evidence-based adaptation policies
We provide an evidence-based foundation for identifying locally grounded adaptation policies, informed by a deeper understanding of human preferences. We assess how behavioral uncertainty influences the evolution of multipurpose reservoir systems under changing climate, environmental, and socio-economic conditions.