基于分形特征的我国干旱灾害小波分析

小波分析可用于揭示旱灾时序数据的演变规律,这对其早期预警具有重要意义.利用我国近60年的的旱灾成灾率时序数据,在 探讨其分形及小波特征基础上,利用Mallat算法进行了多分辨小波分解与重构.结果显示:中国旱灾时序数据存在分形特征,其 最小嵌入维为6; 6层小波分解与重构能够清晰揭示其时序变化的多个周期,其中30-35年(中心32年)周期上能量密度最大.研究证明: 1)分维计算中的最小嵌入维作为小波分解的最大有效分解层数可行,通过对旱灾成灾率时序数据的分析也证实了该猜想,这拓展了分形 计算的应用. 2)我国旱灾成灾率受6个实质性状态变量驱动,具有多周期特征.目前仍处于旱灾成灾率的增长期.

Fractal and wavelet analysis on the 59 years time series of drought disaster ratio in China

The wavelet analysis can be used to disclose the evolution law of drought disaster ratio time series (DDRTS), which is important to the drought disaster early warning. Based on the wavelet and fractal analysis on the 59 years time series of drought disaster ratio in China, the original time series was decomposed and reconstructed using Mallat multi-resolution wavelet algorithm. The fractal analysis results showed that the Chinese DDRTS had obvious fractal characteristics, which minimum embedding dimension is six. The multi-resolution wavelet analysis results showed that the Chinese DDRTS had obvious multi-period characteristics, which the 30-35 years change period has maximum energy density. The following conclusions can be drawn. 1) The minimum embedding dimension of fractal analysis can be served as a maximum effective decompose level for the Mallat multi-resolution wavelet analysis, which is helpful to enlarge the application field of fractal analysis. 2) The evolution of Chinese DDRTS was driven by six independent state variables, and had obvious multi-period characteristic. As a whole, the drought disaster ratio remained to be in a growth period at present.