Wavelet-crosscorrelation-based interferometric redatuming in 4D seismic

摘要

Interferometric virtual source (VS) redatuming crosscorrelates downgoing waves with the corresponding upgoing waves to convert records from surface-source gathers to virtual gathers at the buried receiver locations. The assumption is that the VS records show the reduced effects of overburden complexity and therefore provide improved seismic repeatability and image quality. In practice, however, the repeatability and quality of the redatumed data can be degraded by one-sided illumination and inadequate preprocessing. In the presence of complex nearsurface structure for land redatuming, the intricate shallow structure and highly variable weathering layers introduce surface multiples, ground-roll/scattering noise, and interference among different wave modes that additionally contaminates the down-and upgoing wavefields. To address these issues, we have developed a new interferometric redatuming method based on crosscorrelation in the wavelet domain. Specifically, by transforming time-offset data into the local time-scale (TS) and local TS-wavenumber domains, we have designed and applied filters on the amplitude of the cross spectra in each domain to suppress the aforementioned artifacts and noise, while maintaining accurate kinematic information by retaining the original phase spectra. The new procedure involves forward wavelet transform of the source records, crosscorrelating the wavelet coefficients, filtering in the corresponding domains, and then inverse wavelet transformation to produce the VS records. Components associated with scattering noise and ground roll can be effectively suppressed because our designed filters tackle their features accurately in the local TS and local TS-wavenumber domains, respectively. We tested the method on data from 13 time-lapse surveys that included a significant repeatability problem across a 17-month survey gap, and we compared the results with results from the conventional VS methods. Our VS method produced the best improvement