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OA北大核心CSTPCD

[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 anal-ysis methods.[Methods]In this study,two groups of artificial sequences with known period and trend char-acteristics 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.