Conclusion

We have proposed an efficient approach to compute and apply adaptive time-frequency transform. The proposed SLTFT method allows us to better control the balance among time-frequency localizations, inverse transform, accuracy and computation cost. The revised streaming algorithm by adding an extra control parameter guarantee the adaptive time-frequency localization. Instead of the iterative strategy, the analytical solution for the nonstationary Fourier series estimation has the low computational complexity even when dealing with large-scale seismic data. Meanwhile, the custom frequency points in the least-squares problem makes the proposed method more flexible. The SLTFT provides a convenient time-frequency analysis domain with the adjustable resolution fitting for different kinds of nonstationary signals. Traditional seismic data processing tasks such as ground-roll attenuation, inverse-Q filtering, and multicomponent data registration become well defined in the SLTFT domain and allow for efficient and effective algorithms. Other possible applications of the proposed algorithm in seismic data analysis and processing may include seismic attenuation analysis, low frequency shadow detection, channel detection and so on.