New Method for Estimating Periods in Hydrologic Series Data

Determination of dominant periods is a frequently encountered problem in stochastic analysis of hydrologic series data. In this paper, for mining important information about periods in hydrologic system, new idea for estimating periods has been put forward. New method, MSSA (main series spectral analysis method), has been developed. It is to reconstruct the main frequency signals of the original series firstly, and MESA (maximum entropy spectral analysis) is used to estimate periods of the new reconstructed series. Various period-estimating methods (including conventional ones and newly developed one) have been applied on same time series to compare which is better. Results show that conventional methods (FFT, MESA) are not as good as expected because of the influences of noises. However, these influences are not so strong when using new method (MSSA). It can be concluded that the new method would improve estimation results and some extent eliminate disturbance.