分解集成模型在径流预报领域的研究进展  

Research progress on decomposing ensemble models for streamflow forecasting

作　　者：徐腾[1,2] 刘家栋 南统超 鲁春辉 XU Teng;LIU Jiadong;NAN Tongchao;LU Chunhui(The National Key Laboratory of Water Disaster Prevention,Hohai University,Nanjing 210098,China;College of Water Conservancy and Hydropower Engineering,Hohai University,Nanjing 210098,China;Yangtze Institute for Conservation and Development,Nanjing 210098,China)

机构地区：[1]河海大学水灾害防御全国重点实验室,江苏南京210098 [2]河海大学水利水电学院,江苏南京210098 [3]长江保护与绿色发展研究院,江苏南京210098

出　　处：《水科学进展》2024年第6期1021-1032,共12页Advances in Water Science

基　　金：国家自然科学基金资助项目(42377046)。

摘　　要：气候变化致使流域水文状况发生较大改变,传统水文模型已不足以精准捕捉日径流序列的复杂动态过程。分解集成模型作为一种能有效处理非线性、非平稳径流预报的混合模型,尚缺乏系统的研究总结,在一定程度上限制了其在径流预报领域的发展与应用。因此,本文综述了分解集成模型的理论基础、应用发展及现实问题,分类总结了主流算法的应用流程,展望了其未来的研究方向。结果表明:①分解集成模型可以有效地处理非线性、非平稳的径流预报;②不同的分解集成模型在径流预报中的适用性有所差异,需根据具体情况选择合适的模型结构和参数设置;③分解集成模型面临模态混叠、边界效应和数据泄露等问题,可尝试通过建立更复杂的耦合模型和改进分解策略缓解。分解集成模型在径流预报中展现出了良好的性能和广阔的应用前景,但仍需通过进一步优化算法解决实际问题,以提升模型的准确性和可靠性,为应对复杂水文变化提供更加可靠的预报工具。Climate change has significantly impacted hydrological.conditions across diverse river basins.Traditional hydrological models fall short in accurately representing the complex dynamics of daily streamflow sequences.The decomposition ensemble model,designed to manage nonlinear and non-stationary streamflow forecasting effectively,has not been comprehensively studied or summarized.This gap limits its development and practical application in streamflow forecasting.This paper provides a detailed overview of the theoretical foundations,application developments,and practical challenges of the decomposition ensemble model.It systematically categorizes and summarizes mainstream algorithmic applications and outlines prospective research directions.The analysis reveals that:①The decomposition ensemble model excels at managing nonlinear and non-stationary streamflow forecasting.②The applicability of various decomposition ensemble models to streamflow forecasting varies,necessitating the selection of suitable model structures and parameter configurations for specific scenarios.③Decomposition ensemble models must address mode mixing and also tackle boundary effects and data leakage,challenges that could be mitigated through more sophisticated coupled models and advanced decomposition strategies.Despite the promising performance and broad potential applications of the decomposition ensemble model in streamflow forecasting,it is crucial to refine the algorithm and overcome practical challenges to improve the model's accuracy and reliability.Such enhancements will make it a more effective tool for managing complex hydrological variations.

关 键 词：分解集成模型 径流预报 水文模型 非平稳径流 混合模型 

分 类 号：TV11[水利工程—水文学及水资源]