Periodic Identification of Astronomical Time Series with Empirical Mode Decomposition and Wavelet Transform Analysis

Identification of dominant periodicities is a very important but difficult task in astronomical time series analysis. In the present paper, a new method of periodic identification is proposed in which empirical mode decomposition (EMD) and wavelet transform analysis (WTA) are used in combination. We firstly apply EMD method to decompose a time series into several intrinsic mode functions (IMFs), and then by using WTA approach to identify periodicities in each of IMFs, and finally all of the actual periodicities in astronomical time series can be obtained. Analyses of an observational data set indicate better performance of the proposed EMD-WTA method to identify periodicities. Compared with date compensated discrete Fourier transform and Lomb-Scargle period gram methods which are widely used presently, the EMD-WTA method not only can improve the periodic identifying capability of a time series, but also can improve overall periodic identification by being able to distinguish system noise, quasi-periodicities, and secular trend.