A New Method of Periods´ Identification in Hydrologic Series Based on EEMD

Identification of dominant periods is a very important but difficult task in hydrologic time series data analysis. In this paper, for improving the results of periods´ identification, a new method, called EEMD-MESA (ensemble empirical mode decomposition-maximum entropy spectral analysis), has been proposed, whose main idea is identifying the main intrinsic mode functions (MIMFs) in hydrologic series firstly, and then by using MESA to identify periods in each MIMFs, all periods in the hydrologic series can be gotten finally. By applying to an observed runoff series, advantages of the new method have been verified. Analyses results show that EEMD-MESA is as better as MSSA but much better than other methods (FFT and MESA); While compared with MSSA, EEMD-MESA is more convenient and time-saving. Therefore, the EEMD-MESA method would be more applicable to practical hydrologic works.