Detecting Multiple Timescales Regime Shifts by Haar Wavelet

摘要

This new methodology based on Haar wavelet transform can effectively identify multiple timescales regime shifts,quantitatively describe their corresponding confidence levels,and is strongly resistant against random signal components.The regime shifts are satisfactorily revealed in the constructed data series used for assessment of the proposed method.Preliminary application of the method to examine the Southern Oscillation Index indicated clearly that the abrupt changes such as the 1976 shift at multiple scales and a significant shift that existed around 1998 at least in 5-year scale,feature the climate variation in the Pacific during the recent 70 years.