Ren et al. (2024), Integrated Risk Management for Cascading Reservoirs Under Uncertainty using Networked Modelling

Description of Research Data and CodeThis repository contains the data and code associated with the research paper: Ren et al. (2024), Integrated Risk Management for Cascading Reservoirs Under Uncertainty using Networked Modelling, currently under review at Water Resources Research. OverviewTo investigate the risk interdependencies arising from hydraulic interactions in cascading reservoir systems, we developed a risk propagation model using Bayesian networks (see file: Risk_propagation_model). Building on this model, we employed EMODPS to create a robust operational model for the reservoirs (see file: Robust_operation_model). Our goal was to minimize the joint risks of insufficient hydropower output and ecological water shortages while formulating robust operating policies to mitigate system performance degradation in the face of uncertain future runoffs (generated from our runoff simulations, see file: runoff simulation). Additionally, we analyzed the relationship between overall risk and risk at individual reservoir sites using a scenario discovery algorithm to pinpoint scenarios that reveal vulnerabilities (see file: python_project_scenariodiscovery). AcknowledgmentsThis project builds upon the code developed by Giuliani et al. (2016) M3O-Multi-Objective-Optimal-Operations (https://mxgiuliani00.github.io/M3O-Multi-Objective-Optimal-Operations/), Hadka and Reed (2013) BORG MOEA (http://borgmoea.org/), and Kevin Patrick Murphy et al. (2007) Bayesian Network Toolbox (https://www.ipcc.ch/report/ar6/wg1/#InteractiveAtlas). We are grateful to the original authors for their contributions. While we have made modifications and extensions to the original code, we have not altered its license. Users should refer to the original repositories for more details and ensure compliance with the terms of the original authors' licenses.研究数据与代码说明 本仓库包含与研究论文《Ren等人（2024）：基于网络化建模的不确定性下梯级水库集成风险管理》相关的数据与代码，该论文目前正在《Water Resources Research》（《水资源研究》）期刊审稿中。 概述 为探究梯级水库系统中水力学交互作用所引发的风险相依性，我们构建了基于贝叶斯网络（Bayesian networks）的风险传播模型（详见文件：Risk_propagation_model）。在此模型基础上，我们采用EMODPS算法构建了梯级水库的鲁棒调度模型（详见文件：Robust_operation_model）。本研究的目标为：在应对由径流模拟生成的不确定未来径流（详见文件：runoff simulation）的场景下，最小化水电出力不足与生态缺水的联合风险，并制定鲁棒调度策略以缓解系统性能退化。 此外，我们采用情景发现算法（scenario discovery algorithm），分析了系统整体风险与单个水库站点风险之间的关联，以精准识别出揭示系统脆弱性的情景（详见文件：python_project_scenariodiscovery）。 致谢 本项目基于Giuliani等人（2016）开发的M3O-多目标最优调度（M3O-Multi-Objective-Optimal-Operations，https://mxgiuliani00.github.io/M3O-Multi-Objective-Optimal-Operations/）、Hadka与Reed（2013）开发的BORG多目标进化算法（BORG MOEA，http://borgmoea.org/），以及Kevin Patrick Murphy等人（2007）开发的贝叶斯网络工具箱（Bayesian Network Toolbox，https://www.ipcc.ch/report/ar6/wg1/#InteractiveAtlas）构建而成。我们对原作者的贡献致以诚挚谢意。 尽管我们对原始代码进行了修改与扩展，但并未变更其许可协议。用户应参考原始代码仓库以获取更多细节，并确保遵守原作者的许可条款。