EMD及其扩展方法在水文学中的研究进展及应用综述  

Advances and applications of empirical mode decomposition and its variants in hydrology:A review

作　　者：陈云飞[1,2,3] 刘祖钰[1,2,3] 刘秀花 贺军奇 郑策[1,2,3] 马延东 CHEN Yunfei;LIU Zuyu;LIU Xiuhua;HE Junqi;ZHENG Ce;MA Yandong(School of Water and Environmental,Chang’an University,Xi’an 710054,China;Key Laboratory of Subsurface Hydrology and Ecological Effects in Arid Region of Ministry of Education,Xi’an 710054,China;Key Laboratory of Ecological Hydrology and Water Security in Arid Areas,Ministry of Water Resources,Xi’an 710054,China;Shaanxi Academy of Forestry,Xi’an 710082,China)

机构地区：[1]长安大学水利与环境学院,西安710054 [2]旱区地下水与生态效应教育部重点实验室,西安710054 [3]水利部旱区生态水文与水安全重点实验室,西安710054 [4]陕西省林业科学院,西安710082

出　　处：《灌溉排水学报》2025年第2期101-112,共12页Journal of Irrigation and Drainage

基　　金：国家自然科学基金项目(42372288,41877179,41901034,42202279);中央高校基本科研业务费项目(300102292904)。

摘　　要：受气候变化、生态演替以及人类活动的影响,水文序列蕴含了大量多界面交互叠加信息,使其过程表现出高度非线性与非平稳特征。因此,如何挖掘、分析水文序列中的隐藏信息及局部时变特征一直是水文学领域研究的热点和难点。经验模态分解(EMD)自提出以来便迅速受到各界学者们的广泛关注,近年来在水文领域中的应用更突显了其处理非线性、非平稳数据的优越性。本文梳理了EMD的基本理论、方法特性及现存问题,总结了5种发展较为成熟且应用广泛的EMD/类EMD扩展方法,包括希尔伯特-黄变换、集合经验模态分解、多元经验模态分解、极点对称模态分解及变分模态分解。随后通过水文序列的多时空尺度分析、趋势检验和模型预测3个方面,综述了EMD及其扩展方法的应用研究现状。最后,对其在水文学中的应用研究进行了展望,并就EMD的理论框架、水文变异性分析及区域多时空尺度研究提出了具体建议。Hydrological series are influenced by climate change,ecological succession,and human activities,containing complex,multi-layered,and interactive information that reflects highly non-linear and non-stationary characteristics.Effectively extracting and analyzing hidden information,while understanding temporal variation in hydrological data,remains a challenge.Empirical Mode Decomposition(EMD)has garnered increasing attention due to its ability to analyze non-linear and non-stationary data.This paper reviews the theory and application of EMD in hydrology,including its advantages and limitations.The review explores extensions and adaptations of EMD aimed at improving hydrological sequence analysis.We provide a comprehensive overview of the fundamental theory,methodological characteristics,and current challenges of EMD,covering five EMD-based methods:Hilbert-Huang Transform(HHT),Ensemble Empirical Mode Decomposition(EEMD),Multivariate Empirical Mode Decomposition(MEMD),Extreme Point Symmetric Mode Decomposition(ESMD),and Variational Mode Decomposition(VMD).The pros and cons of each method are analyzed.Additionally,we review the progress of EMD and its variants,particularly in the context of multi-spatiotemporal scale analysis,trend detection,and predictive modeling of hydrological series.Key insights are drawn from the use of EMD in detecting trends,identifying temporal variation features,and improving model predictions for hydrological data across different spatial and temporal scales.Despite the advancements in application of EMD and its variants in hydrology,challenges remain,including issues related to method robustness and efficiency in handling large-scale datasets.We conclude by offering recommendations for future research,including advancing EMD theory and enhancing techniques for analyzing hydrological variability at multi-temporal and multi-spatial scales under changing environmental conditions.

关 键 词：数据挖掘 多时空尺度分析 水文变异性分析 趋势检验 水文预报 

分 类 号：P349[天文地球—水文科学]