Scholarlink

<p>The combined effects of global warming and intensive human activities have disrupted the fundamental assumption of hydrological stationarity, resulting in significant changes to runoff sequences that were traditionally derived under stationary hydro-meteorological conditions. To investigate optimal water resource planning for irrigation systems under stochastic runoff conditions, this study focuses on a reservoir–pump station irrigation system and develops an optimized water allocation model. This study employs Copula functions and Gibbs sampling to simulate stochastic runoff, thereby generating a large set of runoff samples. For three representative hydrological year types (<i>P</i> = 50%, 75%, and 95%), six machine learning models were used to extract operational rules for the irrigation system under varying hydrological conditions. The SHAP framework, a post-hoc interpretability technique, was used to quantify feature interactions and their contributions to model predictions. The results show that the discrepancy between optimization outcomes from the mathematical model and the simulated operational rules ranges from 3% to 10% across representative hydrological years, thereby confirming the practical applicability of the extracted rules for guiding optimal decision-making in reservoir–pump station irrigation systems. This study provides theoretical support for optimizing complex irrigation systems and promoting the sustainable management of scarce water resources.</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

Investigation into the Management Protocols for “Reservoir-pumping Station” Irrigation Infrastructure during Characteristic Hydrological Years Amidst Variable Runoff Conditions

  • 期刊论文

  • 发表日期：

    2026年5月16日

查看全文
  • Cong Wei,
  • Wenpeng Zhao,
  • Jing Xu,
  • Haiyu Liu,
  • Zhicheng Zhong

摘要

The combined effects of global warming and intensive human activities have disrupted the fundamental assumption of hydrological stationarity, resulting in significant changes to runoff sequences that were traditionally derived under stationary hydro-meteorological conditions. To investigate optimal water resource planning for irrigation systems under stochastic runoff conditions, this study focuses on a reservoir–pump station irrigation system and develops an optimized water allocation model. This study employs Copula functions and Gibbs sampling to simulate stochastic runoff, thereby generating a large set of runoff samples. For three representative hydrological year types (P = 50%, 75%, and 95%), six machine learning models were used to extract operational rules for the irrigation system under varying hydrological conditions. The SHAP framework, a post-hoc interpretability technique, was used to quantify feature interactions and their contributions to model predictions. The results show that the discrepancy between optimization outcomes from the mathematical model and the simulated operational rules ranges from 3% to 10% across representative hydrological years, thereby confirming the practical applicability of the extracted rules for guiding optimal decision-making in reservoir–pump station irrigation systems. This study provides theoretical support for optimizing complex irrigation systems and promoting the sustainable management of scarce water resources.