Deblending Seismic Interference in Ocean Bottom Node Data - a Deepwater Gulf of Mexico Subsalt Example

Presented at the 2017 SEG Annual Meeting in Houston


In a cost conscious exploration and production industry, where a premium is placed on efficient and productive seismic operations, no circumstance is perhaps more detrimental to marine seismic crew productivity than time sharing with other crews operating in proximity. High amplitude seismic interference (SI) from a nearby active seismic source can overwhelm the weaker reflections of interest generated by the primary seismic source and common practice has been to share the available shooting time when interfering energy exceeds prescribed thresholds. In lieu of time sharing, traditional processing approaches to minimize the effect of SI have taken the form of coherent noise suppression in the shot domain or incoherent noise suppression in common receiver, common offset or common midpoint domains. The incoherence in these domains results when the source activation times are not synchronized. A wide variety of interference mitigating processing tools have been applied over the years, ranging from inverse power weighted CMP stack (Lynn et al., 1987) or more recently F-X prediction filtering (Gulunay et al., 2004) or sparse tau-p inversion (Zhang and Wang, 2015).

The SI problem is not materially different from the problem of deblending data from planned simultaneous source surveys. Deblending is the separation of continuously recorded signals into the individual contributions from each source that would have been recorded had each source operated in isolation. Deblending can be viewed more as a signal separation problem than a noise attenuation problem. In a controlled simultaneous source experiment, the blending “noise”, the equivalent of SI, is identified based on its coherency when time referenced to the interfering source. In the case described in this paper, the application of traditional deblending to treat SI from another crew utilized knowledge of the source activation times from the other crew. Source positions are helpful, too, to construct coherency estimates and the source points must be spatially sampled adequately to be treated as an additional coherent energy source. When other sources are in proximity and timing and position data are shared, treating the SI as a deblending problem is not only an effective way of removing the interference from the primary data set, but also provides very useful additional data, usually long offset, that has value for imaging and velocity model building (Yang et al., 2016). When timing and positions of interfering sources are not available, the SI can still be mitigated during deblending (Baardman et al., 2014) but a coherent properly timed signal is not extracted from the SI.

In this paper, we demonstrate the successful mitigation of severe seismic interference in a deep water OBN survey from the Gulf of Mexico using an iterative coherency-constrained deblending procedure, (Mahdad et al., 2011).