本文系统地提出了基于连续小波变换和离散小波变换的小波互相关分析方法,以及小波互相干性分析方法。在介绍水文序列小波互相关系数求解方法的基础上,提出小波互相关度定量指标,用于描述两时间序列在整体时间域上的互相关关系;并进一步提出绘制小波互相关系数等值线图的方法,以实现对时间序列互相关关系进行“时频综合分析”的目的。对黄河利津站和花园口站54年(1950-2003年)年径流序列进行分析,结果显示:时滞为0时序列在时间尺度为3、7、11和20年时具有良好的互相关关系,较传统方法仅揭示两序列在整体时间域上的线性互相关程度具有优越性,应用其可实现由整体到局部认识时间序列互相关关系的目的。能够分析和定量描述非平稳时间序列在特定时间尺度和指定时滞下的互相关关系,可克服传统互相关分析方法的局限,具有更好的灵活性和适用性。
The continuous wavelet transform (CWT) based and discrete wavelet transform (DWT) based wavelet cross-correlation (WCC) methods are developed respectively for hydrologic series analysis and the wavelet coherence analysis method is also suggested for analysis in frequency domain. The wavelet cross-correlation degree is defined to describe the cross-correlation between two specific series in the whole time domain. The method for drawing wavelet cross-correlation coefficient contour map is given, by which the integrated time-frequency analyses of series cross-correlation can be carried out. The proposed methods are applied to analyze the measured annual runoff series of the Yellow River in the period from 1950 to 2003 at the Huayuankou Station and Lijin Station. The results indicate that the two runoff series exhibit well cross-correlated under zero time delay and four time scales. While the traditional cross-correlation analysis method can only reflect their linear cross-correlation in the whole time domain. It is concluded that the pro- posed methods are better than the traditional method because they can describe the cross-correlation of non-stationary time series under any time scale and any time delay.