Scholarlink

<p>The frequency of disastrous floods has significantly increased across the world in recent decades. Severe damage caused by floods in regions where river flow is regulated by reservoirs has caused the operation of reservoirs to be hotly debated. This paper presents a forensic optimal operation approach (FOOA) and a goal-attained approach (GAA) to investigate historical operation (HO) of reservoirs under floods to examine whether operational decisions contributed to increased flood damages. FOOA combines a reservoir inflow forecasting method with a simulation-optimization model to minimize the maximum release from the most downstream reservoir under floods using support vector machine (SVM) and the Genetic Algorithm (GA). GAA applies observed historical reservoir inflows to the simulation-optimization model for flood control. GAA and FOOA are developed in this study for the forensic investigation of Seimare and Karkheh reservoirs’ operation during the 2019 floods, which was one of the three largest floods that have occurred in Karkheh basin, Iran, in the past 70 years. This paper’s results show that historical reservoir operation intensified floods, as established by the finding that the average vulnerability of the reservoirs system to downstream damages under HO was 65 and 92% worse compared to FOOA’s and GAA’s. This study presents a methodology for improved decision-making to reduce future flood impacts, and advances the current understanding of how operational errors may magnify flood hazards.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning and optimization applied to the assessment of flood hazard magnification by reservoir operation

  • 期刊论文

  • 发表日期：

    2026年4月21日

查看全文
  • Manizhe Zarei,
  • Omid Bozorg-Haddad,
  • Hugo A. Loáiciga

摘要

The frequency of disastrous floods has significantly increased across the world in recent decades. Severe damage caused by floods in regions where river flow is regulated by reservoirs has caused the operation of reservoirs to be hotly debated. This paper presents a forensic optimal operation approach (FOOA) and a goal-attained approach (GAA) to investigate historical operation (HO) of reservoirs under floods to examine whether operational decisions contributed to increased flood damages. FOOA combines a reservoir inflow forecasting method with a simulation-optimization model to minimize the maximum release from the most downstream reservoir under floods using support vector machine (SVM) and the Genetic Algorithm (GA). GAA applies observed historical reservoir inflows to the simulation-optimization model for flood control. GAA and FOOA are developed in this study for the forensic investigation of Seimare and Karkheh reservoirs’ operation during the 2019 floods, which was one of the three largest floods that have occurred in Karkheh basin, Iran, in the past 70 years. This paper’s results show that historical reservoir operation intensified floods, as established by the finding that the average vulnerability of the reservoirs system to downstream damages under HO was 65 and 92% worse compared to FOOA’s and GAA’s. This study presents a methodology for improved decision-making to reduce future flood impacts, and advances the current understanding of how operational errors may magnify flood hazards.