A data-driven method for time-varying wavelet extraction based on the local frequency spectrum

Abstract

Seismic wavelet extraction always plays a central role in high-resolution seismic processing. Conventional methods assume that seismic data are stationary when a constant wavelet is considered, which ignores the time-varying characteristics of seismic wavelets. In reality, seismic data are nonstationary because of attenuation, scattering, and other physical processes during propagation, which means that the frequency spectrum of seismic signal changes from shallow to deep formations. We have developed a time-varying wavelet extraction method by using a highly energy-concentrated time-frequency representation technique. Time-varying wavelets are generated according to the local frequency spectrum at every instant. In addition, because the estimations of parameters for wavelet extraction are fully data-driven, the results of the proposed method are more accurate and suitable for the nonstationary nature of actual seismic data. Synthetic tests indicate the reliability and robustness of the proposed method, even under noise contamination. By applying the time-varying wavelet extracted using the proposed method to seismic inversion on a field data example, we obtain the deconvolution result with improved resolution and a better fit to the well-log reflectivity compared to that by using conventional wavelet extraction methods.