Decision Support System for Integrated Reservoir Operation Using Artificial Intelligence Based on Operational Ensemble Meteorological Forecasts

2018 Fiscal Year Final Research Report

• Project/Area Number: 16K06510
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Hydraulic engineering
• Research Institution: Kyoto University
• Principal Investigator: Nohara Daisuke 京都大学, 防災研究所, 助教 (00452326)
• Research Collaborator: GOURBESVILLE Philippe ; KIM Young-Oh
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: アンサンブル予報 / ダム / 弾力的操作 / 人工知能 / 意思決定支援

Outline of Final Research Achievements

A decision support system for integrated reservoir operation considering operational ensemble meteorological forecasts was developed using artificial intelligence techniques. A method to estimate basin rainfall prediction and its uncertainty by use of statistical relationship between pressure distributions and basin rainfall estimated from historical flood records in the target river basin. An optimal reservoir operation strategy is then estimated from predicted conditions of pressure distributions and rainfall by ensemble forecasts for integrated reservoir operation such as prior release operation considering its effects and risks in both the flood and drought managements.

Free Research Field

水資源工学

Academic Significance and Societal Importance of the Research Achievements

本研究の学術的な特色は，アンサンブル予報に含まれる複雑な情報を高速に処理する上で親和性が高いことに着目して，ダム操作時の予報データの活用に不可欠な大量の情報処理を人工知能技術や情報工学的手法を駆使して支援する手法を開発する点にある．また，アンサンブル予報の降水量予測値がそのままでは精度が芳しくないことに鑑み，予報に含まれる気圧場などの予測データを活用することで，降水予測精度の改善を図るとともに，流域の過去の出水時の経験を活用しながら，ダムの弾力的操作を行うことによって治水・利水管理の高度化を図る手法を検討しており，実管理者にも受け入れられやすい予測の利用方法を提案している．