Downloads: 123

Files in This Item:
File Description SizeFormat 
a52b0p75.pdf2.06 MBAdobe PDFView/Open
Title: 長期貯水池操作へのアンサンブル降水予報導入時における最適化モデルの放流決定過程に関する一考察
Other Titles: A Study on Water Release Decision Process of Optimization Models in the Application of Ensemble Forecast to Long-term Reservoir Operation
Authors: 野原, 大督  KAKEN_name
坪井, 亜美  KAKEN_name
堀, 智晴  kyouindb  KAKEN_id
Author's alias: NOHARA, Daisuke
HORI, Tomoharu
Keywords: 長期貯水池操作
long-term reservoir operation
ensemble forecast
water release optimization
Issue Date: 1-Jun-2009
Publisher: 京都大学防災研究所
Journal title: 京都大学防災研究所年報. B = Disaster Prevention Research Institute Annuals. B
Volume: 52
Issue: B
Start page: 753
End page: 764
Abstract: 長期貯水池操作へのアンサンブル予報の効果的な適用手法の構築に資するため,利水操作を対象にアンサンブル予報を用いた3種の最適放流決定モデルを構築し,予報の利用方法や各モデルにおける放流決定過程の違いが操作に与える影響について分析を行った。最適放流決定モデルには,いずれもDPの一種である決定論的DP,確率DP,及びSampling Stochastic DPを用いた。気象庁の1か月アンサンブル予報の降水量予報GPVを用いて吉野川流域早明浦ダムを対象に分析を行った結果,予測精度が比較的悪い場合でも予報を考慮することが有用であること,予報の分布情報を放流決定時に考慮する場合は分布の平均値だけでなく中央値にも注意する必要があることが示唆された。
Optimizing processes of three DP-based models are analyzed through the application to optimization of long-term reservoir operation, especially focusing operation for water use. One-month ensemble forecast of precipitation provided by JMA is introduced to long-term reservoir operation. Ensemble inflow scenario is then predicted from estimated precipitation scenario by use of regression model. Optimization of long-term reservoir operation which only focused on water utilization is conducted by use of predicted inflow scenario. Through the application to Sameura reservoir in the Yoshino River basin, it was shown that distribution characteristics of ensemble prediction such as relationship between normal and median value can be an index for decision of employment of optimization model, with consideration of prediction error tendency.
Appears in Collections:No.52 B

Show full item record

Export to RefWorks

Export Format: 

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.