Have you got Time for Depth?

27th June 2018

There is often an understandable disconnect between the roles of a processing and an imaging Geophysicist. The former tends to concentrate on achieving the best de-noise, de-multiple, de-ghosted product and the latter is more concerned with what the reflective properties of the imaged data can imply about the subsurface (velocity, anisotropy etc.).

In order to bridge this gap, we can attempt to pose questions such as 'what processes can have an impact on the velocity model?', 'what processes can adversely affect the image?' and 'how do you take velocities and change them for depth processing'. These items were discussed in Shearwater’s latest internal lunch and learn session titled ‘Have you got time for depth?’

The degree of success of a pre-migration process will affect the ability to estimate the parameter that we wish to invert, such as velocity. For example residual multiple or swinging noise may cloud the primary events. However there are tools and workflows that an imaging Geophysicist can call upon to minimise the impact through the use of a post migration conditioning sequence. The residual move-out (RMO) on a gather can be refined through the use of processes such as a high-cut filter, radon demultiple and spatial filtering to optimise S/N prior to picking. The RMO picking can be constrained though the use of criteria such as percentage envelopes, Gamma and quality factors, with any outlier picks addressed through high-grading.

The image below shows the net result from these steps, when we overlay our PSDM RMO picks upon non-conditioned (raw) gathers.

Likewise, S/N can be enhanced prior to the estimation of frequency variation in a dataset in order to accurately invert for Q. In short, the quality of a model is very much in the hands of the imaging Geophysicist, and challenging element of the data can be reasonably overcome.

There are however, certain pre-migration processes, which will, if not handled correctly, have a more direct and significant impact on the imaging and related inversion. Prism wave (‘double bounce’) energy is a good example of this, which can occur in select circumstances where strong model contrasts are coupled with structure (see image). Prism waves are often difficult to identify in the un-migrated domain, and signal can be mischaracterised as noise/multiple and removed accidentally, to the determent of the subsequent imaging. 

In summary, careful attention is required throughout both processing and model-building sequences in order to derive an optimum imaged result. 

By Andrew Woodcock, Depth Imaging Supervisor