SPECTRAL DECOMPOSITION OF SEISMIC DATA WITH IMPROVED SYNCHROSQUEEZING TIME-FREQUENCY TRANSFORM

Spectral decomposition is a novel signal analysis tool for seismic data. Resolution of traditional time-frequency transform methods is limited by Heisenberg uncertainty principle. By assigning complex coefficients along frequency or scale axis, synchrosqueezing algorithm is a way to sharpen time-frequency representation towards its ideal representation. Whereas, traditional synchrosqueezing algorithm is not very suitable for signal which contains strong frequency modulated modes. Time-frequency respresentation needs to be further sharped to meet the needs of seismic signal analysis and interpretation. In this paper, we introduce an improved synchrosqueezing algorithm named second-order synchrosqueezing transform into seismic spectral decomposition. With computation of second-order derivatives of the phase of STFT, we can obtain an invertible and sharper time-frequency representation than traditional synchrosqueezing algorithm. The method is applied to synthetic signal and field seismic data. Results show its effectiveness.