An algorithm for the determination of multi-period structure of time series

Wavelet analysis is widely used in meteorology and hydrology. In this paper, a data analysis method is proposed to explore the multi-period characteristics, in which its oscillation amplitude, the periodic oscillation trend is given with a quantitative description, based on the 55 years of temperature data(1951–2005) for Heilongjiang Jixi weather station and the Matlab generated data by employing the modified Morlet wavelet transform. The advantage of using the modified Morlet wavelet instead of classical Morlet wavelet is presented. Additionally, the paper addresses the issues including determining the main, the second and other periods with the wavelet variation and determining the scale population from the smallest to largest scales.