Document Type : Original Research Paper
Authors
Insitute of Geophysics, University of Tehran, Iran
Abstract
All the earthquakes are transient and nonstationary. For lack of alternatives, in seismology as well as earthquake engineering, most data are still processed by using the Fourier analysis. The most difficulties in the Fourier spectral analysis are associated with nonlinearity and non-stationary nature of the data. Such methods cannot reveal the detailed information in the dispersion properties, the wave form deformation, and the energy-frequency distribution. In this study, a new technique based on the time-domain empirical mode decomposition has been explained, which enables us to analyze both short-term information and long-term structures in seismic waves. It provides insight into long term memory and local time behavior of seismic signals. Oscillation modes of Earthquakes and Explosions are compared to each others, relationships between each group of the same data and differences between different data are obtained based on the produced IMFs. With respect to the advanced methods such as discrete stochastic non-Markov process it has been shown that this technique gives better identification. Plotting the maximum frequency of different IMFs via position of them, gives an effective identification tool. In this study first the technique is explained and then the obtained results are illustrated and discussed.
Keywords