小波方法在水资源趋势分析中的能力检验

通过Monte-Carlo模拟试验对小波多分辨分析法在水文时间序列趋势分析中的趋势识别能力进行了研究。结果表明,小波多分辨分析方法的趋势识别能力与时间序列成份组成有关,与序列的噪音分布类型无关,而分析中的分解尺度对于随机性影响大的序列,其值与变差系数成正相关。小波多分辨分析所具有的时频局部化特征,能够合理地反映出时间序列在时间域中的变化特点,其在频域中分析趋势比线性回归和多项式拟合等在时间域中分析趋势的方法更具有优势。另外,在趋势分析前应分析序列中合理的突变成分以及剔除序列中的周期性成分,否则小波趋势分析结果将不可靠。

Verification of the Power of the Wavelet Trend Analysis

This article researches on the power of the method of Multi-Resolution of wavelet when it used in analyzing trend of hydrologic time series.Through Monte-Carlo simulating approach,it is found that the tendency recognition capability of the analysis method relates to the components of the time series and has nothing to do with the sample noise distribution type.The scale of decompose of the wavelet in analytic processes correlated positively with the coefficient of variation of the time series.The local feature in the time-frequency domain of the Multi-Resolution of wavelet can determine reasonably the characteristics of time series in the time province.Analyzing trend in the frequency domain with Multi-Resolution of wavelet has preponderance over that with near regression and polynomial fit methods in the time domain.Moreover,the jump and the periodic ingredient of the sequence should be identified and eliminated from the series before analysis of the trend otherwise the results from wavelet trend analysis will be unreliable.