兰德-短期水库调度的主动方法-深度不确定性下决策在加强预测知情水库调度中的潜在应用（英）-2025_24页_572kb

Summary

Context & Problem

Reservoir operators face complex decisions under short timeframes (days to hours) during potential flood events, balancing multiple objectives like flood mitigation and water supply while dealing with forecast uncertainty and model limitations.

Solution Approach: FIRO + DMDU

Forecast-Informed Reservoir Operations (FIRO) leverage weather and hydrological forecasts. Decision Making Under Deep Uncertainty (DMDU) addresses inherent uncertainties (forecast skill, model approximations) better than traditional approaches, especially for complex, high-stakes decisions. DMDU methods help develop robust strategies adaptable across various possible futures.

Key DMDU Methods

1. Scenario Planning: Create internally consistent futures to stress-test strategies and identify robust options.
2. Dynamic Adaptive Pathways (DAP): Develop flexible strategies that adjust over time based on changing conditions and predefined triggers.
3. Robust Decision Making (RDM): Run simulation models across many futures to stress-test strategies, quantify potential regret, and identify vulnerabilities.

Potential Benefits

• Improved flood mitigation without negatively impacting water supply.
• Increased water storage capacity.
• Greater flexibility in operational rules to account for evolving uncertainties.
• Optimized decision-making integrating probabilistic forecasts and managing multiple competing objectives.

Challenges & Requirements

• Addressing deep uncertainty where model parameters and future conditions are poorly known.
• Computational demands for methods like RDM.
• Need for flexible operational policies adapted to systematic uncertainty.
• Incorporating valuable probabilistic information in short-term, high-stakes decisions.
• Successful coproduction involving researchers and end-users (operational staff, regulators, public) throughout the process.

Recommendations

• Apply DMDU methods to develop robust operating rules.
• Explore techniques for real-time strategy optimization with minimal precomputation.
• Design effective communication and visualization tools.
• Prioritize the integration of DMDU principles into operational contexts to enhance decision-making and overall water management outcomes.
• Ensure operational approaches account for the skill and limitations of available forecast systems.