STFT-based multisynchrosqueezing transform using a second-order signal model for seismic data analysis

Since time-frequency analysis (TFA) technique can reveal the local properties of seismic signals, it has been widely applied in seismic data analysis. Short-time Fourier transform (STFT) is a valuable tool for analyzing non-stationary signals in geophysics, and researchers have utilized it to solve various geophysical problems including spectral decomposition, seismic data interpolation and signal filtering. In this paper, a novel time-frequency method named short-time Fourier transform based multisynchrosqueezing transform using a second-order signal model (FMSST2) is introduced to analyze seismic data. The FMSST2 combines the multisynchrosqueezing framework using an iterative reassignment procedure and a second-order signal model to concentrate the energy in the time-frequency map. Moreover, The FMSST2 allows for signal reconstruction with a high accuracy. Two synthetic examples are employed to validate the effectiveness of the FMSST2 method, and the results show that the FMSST2 method does a good job in terms of energy-concentration and noise robustness compared to some classic TFA methods such as the STFT, STFT-Based synchrosqueezing transform (FSST) and multisynchrosqueezing transform (MSST). Applications on field data further demonstrate the potential of the FMSST2 method in characterizing hydrocarbon reservoir, making it a promising time-frequency resolution enhancement tool in seismic data analysis.