基于希尔伯特-黄变换与小波分析的降雨序列多时间尺度研究  

Multi-time Scale Study of Rainfall Series Based on Hilbert-Huang Transform and Wavelet Analysis

作　　者：贺军奇[1,2,3] 郭鑫佳 陈云飞 刘秀花[1,2,3] 高万德[1,2,3] 龙婷 He Junqi;Guo Xinjia;Chen Yunfei;Liu Xiuhua;Gao Wande;Long Ting(School of Water and Environment,Chang′an University,Xi′an 710054,China;Key Laboratory of Subsurface Hydrology and Ecological Effects in Arid Region,Ministry of Education,Chang′an University,Xi′an 710054,China;Key Laboratory of Ecological Hydrology and Water Security in Arid Areas,Ministry of Water Resources,Chang′an University,Xi′an 710054,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

出　　处：《水土保持研究》2024年第2期130-141,共12页Research of Soil and Water Conservation

基　　金：国家自然科学基金项目(41877179,41901034,42001033);陕西水利科技计划项目(2019slkj-18);中央高校基本科研业务费项目(300102292904)。

摘　　要：[目的]对比分析不同方法在多尺度周期计算性能方面的差异,为区域水文周期分析方法的选择提供更多参考。[方法]利用2组已知周期和趋势特征的人工序列,对基于多元经验模态分解(Multivariate Empirical Mode Decomposition,MEMD)改进的希尔伯特—黄变换(Hilbert-Huang Transform,HHT)与小波分析在多尺度周期计算性能差异上进行了测试、对比与分析;并在此基础上,使用改进的HHT(后称HHT)进一步分析了毛乌素沙地11个站点39年降雨序列的多时间尺度特征。[结果]对于人工序列,小波分析(对称延拓)结果偏大,与预设周期存在显著差异(Sig.=0.003<0.05);而HHT和小波分析(周期延拓)的周期结果与预设周期没有显著差异,且HHT精度(Sig.=0.142>0.05)要略高于周期延拓小波(Sig.=0.109>0.05)。对于实测降雨序列,HHT显示毛乌素沙地的降雨主周期存在明显区域差异:腹地站点降雨主周期为5.5 a,而边缘站点为2.7~3.3 a;分解后的残差分量表明了沙地降雨在时间上呈上升趋势,而空间上呈现由东到西逐渐减少的分布状态。[结论]HHT因其自适应性,无需考虑众多参数的选取,在保持精度的同时可以提取更多水文信息,在区域降雨序列多尺度分析上更具优势,因此具有重要的实际应用价值。[Objective]To Compare and analyze the differences in multi-scale periodic calculation performance between different methods can provide more reference for the selection of regional hydrological periodic analysis methods.[Methods]In this study,two groups of artificial sequences with known period and trend characteristics were used to test,compare and analyze the difference between Hilbert-Huang Transform(HHT)based on Multivariate Empirical Mode Decomposition(MEMD)and wavelet analysis in multi-scale period calculation performance.On this basis,the improved HHT was used to analyze further the multi-time scale characteristics of the 39-year rainfall series at 11 stations in Mu Us Sandy Land.[Results]For artificial sequences,the wavelet analysis(symmetric extension)results are relatively large,and there is a significant difference from the preset period(Sig.=0.003<0.05).The periodic results of HHT and wavelet analysis(periodic extension)show no significant difference from the preset period,and the accuracy of HHT(Sig.=0.142>0.05)is slightly higher than that of periodic extension wavelet(Sig.=0.109>0.05).For the measured rainfall series,HHT shows significant regional differences in the main period of rainfall in the Mu Us Sandy Land:the main period of rainfall at the hinterland station is 5.5 years,while at the edge station,it is 2.7~3.3 years.The decomposed residual component indicates that the rainfall at this region showed an upward trend in time and a gradually decreasing distribution state in space from east to west.[Conclusion]This study shows that HHT can extract more hydrological information while maintaining accuracy due to its data-driven,without considering the selection of many parameters,and has more advantages in the multi-scale analysis of regional rainfall series,thus it has important practical application value.

关 键 词：多元经验模态分解 HHT 小波分析 降雨周期 毛乌素沙地 

分 类 号：P333[天文地球—水文科学]