SEISMIC SEQUENCE ATTRIBUTES EXTRACTION THROUGH JOINT TIME-FREQUENCY REPRESENTATION

Abstract

Seismic signals have always created great interest for seismologists and engineers world- wide. These signals represent the inelastic response of the structures and ground, which are characterised by ground motion of nite and short duration. Earthquake ground motions are inherently non stationary in nature. For better understanding about the characterisation of local site e ects subject to earthquake, reliable earthquake signals processing is essential. The traditional method for analysis of seismic signal is Fourier transform (FT), which is more challenging because of its limitations as cannot provide simultaneous time and frequency localizations thus is not useful for analyzing nonsta- tionary signals. Joint Time-Frequency method has a potential to greatly facilitate the extraction of time- frequency information from the signals. This representation unfolds the temporal information and maps a time series into 2-D quantity of time and frequency with e ective characterization of the time-frequency image. In this thesis the Spectral decomposition of seismic signal has been performed through some well known techniques like STFT, CWT,ST and some recently developed techniques as KF,SST. To reduced spectral smearing and to achieve a higher time-frequency localization,an ST based synchrosqueezing method is established, which is refered as Stockwell-Synchrosqueezing transform(S-SST).Illustrated its performance by applying this on synthetic signal as well as on real time seismic signals. Adoption of this innovative joint time-freqyency method will be helpful for earthquake record analysis in the near future.