A Multisynchrosqueezing-Based S-Transform for Time-Frequency Analysis of Seismic Data

Abstract

Time-frequency analysis (TFA) technique has a powerful capacity to characterize non-stationary signals. In this paper, a highly energy concentrated TFA method, called multisynchrosqueezing-based S-transform (MSSBST), is proposed for the analysis of seismic data. Herein, we combine S-transform (ST) and multisynchrosqueezing framework by making full use of an iterative reassignment procedure to concentrate time-frequency energy in a stepwise manner. Furthermore, we derive a series of formulas about MSSBST and its inverse transform, which means that the MSSBST allows for signal reconstruction from its time-frequency representation (TFR). The numerical analysis shows that the proposed method not only can effectively enhance the time-frequency energy concentration but also can offer better performance in characterizing non-stationary signals compared with the short-time Fourier transform (STFT), ST and synchrosqueezing S-transform (SSST). Field examples further demonstrate its potential in depicting spectral anomalies related to hydrocarbon reservoir, thus, facilitating seismic interpretation.