The comparison and application of time-frequency analysis methods to seismic signal

Aiming at seismic prospecting signals,the comparisons of some effective time-frequency analysis methods for nonstationary signal are studied,including short time Fourier transform(STFT),wavelet transform(WT),S transform(ST),Wigner distribution(WVD),smoothed pseudo-Wigner distribution(SPWVD),cone-kernel time-frequency distribution(CKD) and adaptive optimum kernel time-frequency representation(AOK).Numerical methods and applications are implementd on base of their principles.Firstly,time-frequency representations of the above methods and testing computations are implemented with a modeling unstable seismic signal and time-frequency attributes are extracted.The comparisons of the results of above methods are studied according to the time-frequency localization precision and cross-terms suppressing.Secondly,the above methods are applied to real 2D seismic data and seismic attributes extraction for instantaneous frequency and instantaneous bandwidth.On base of the results,it can be concluded that AOK time-frequency representation is the best method among the above methods according to the time-frequency localization precision,cross-terms suppressing,sequences detection validity and worthing to do more research for seismic attributes extraction and spectral decomposition.