Time-frequency post-processing: A theoretical review and its application to reservoir characterization

Time-frequency analysis is a powerful tool for revealing features hidden in non-stationary signals. To address the energy diffusion commonly observed in the traditional time-frequency spectrum, time-frequency post-processing methods have been developed. They significantly enhance the energy concentration of the time-frequency spectrum by reassigning the original spectral coefficients, thereby exhibiting broad application potential in signal processing. This paper provides a systematic review of the theoretical progress in time-frequency post-processing methods and analyzes their internal connections, providing a comprehensive and clear overview. Through numerical examples, this paper compares and analyzes the differences in time-frequency results between synchrosqueezing methods and synchroextracting methods, intuitively demonstrates the time-frequency localization capabilities of various estimators, and provides a reference for method selection. In addition, this paper discusses the application value of time-frequency post-processing methods in reservoir characterization. The synthetic seismic signals and field seismic examples demonstrate that instantaneous frequency-based post-processing methods are more sensitive to the frequency response of thin interbeds, while group delay-based post-processing methods have advantages in identifying seismic reflection interfaces. These results offer new insights and technical perspectives for addressing challenges in reservoir characterization. Finally, this paper outlines future research directions for time-frequency post-processing methods from the perspectives of multi-channel seismic signal processing, time-frequency spectral features, and deep learning integration.