Multiscale multifractal detrended fluctuation analysis of earthquake magnitude series of Southern California

Abstract

The multifractal characteristics of magnitude time series of earthquakes that occurred in Southern California from 1990 to 2010 are studied in this work. A method for the scale division of the magnitude of these earthquakes based on empirical mode decomposition (EMD) and multifractal analysis is proposed. This method gains a new insight into measuring multifractal properties of the magnitude time series at multiple scales, and it reveals further information about the dynamic seismic behavior. By using EMD, a time series can be decomposed into mode time series that represent different time–frequency components. We find that time–frequency components show long-range correlation with different Hurst exponents by using R∕S analysis. Based on the different fractal structures of components, we consider three different scale series: Micro-, Mid- and Macro-scale subsequences, which are superposed and reconstructed by the components. The multifractal properties of the three scale subsequences are analyzed by using multifractal detrended fluctuation analysis (MF-DFA). The results show that the three different scale subsequences have various shapes of multifractal spectra and corresponding distinct properties. The Micro-scale subsequence singularity spectrum shows left-skewed, indicating a relative dominance of the lower Hurst exponent; the Mid-scale subsequence has a right-skewed singularity spectrum; the Macro-scale subsequence exhibits the most significant persistence and shows the strongest multifractality.