A new process allows prestack migration to be simplified into the three steps of trace gathering, normal moveout, and stacking. This new method has many advantages that range from reduced processing times to more accurate velocity analysis.
The objective of seismic processing is to re-Iocate the sampled reflection energy, back to the original reflector position. This has traditionally been accomplished by a sequence of processing steps that required common midpoint (CMP) gathering, normal moveout (NMO), stacking, and post-stack migration. This traditional path requires the estimation of two velocity models, one for the NMO process, and one for migration. The incorporation of dip moveout (DMO) into the processing sequence has allowed the estimation of velocities that match the geology, and improved the focusing of dipping events in areas with complex structure. These methods that use two velocity models may be replaced with prestack migration in which one velocity model is used. Now the data is moved in one step from the input trace to the output migrated position. A number of pre-stack migration methods that incorporate velocity analysis are available. Typically, 2-D prestack migration is performed on source records, or constant offset sections to preserve some form of prestack geometry for velocity analysis.
The new pres tack migration process begins by gathering the energy of the input data into common scatter point (CSP) gathers. The CSP gather is similar in function to the common mid point (CMP) gather as they both define the output trace location, and both contain input traces that are sorted with offset. However, the CSP gather contains many more traces than the CMP gather. The CMP gather only contains traces in which the source and receiver are equidistant and in opposite directions from the CMP location. In contrast, the CSP gather contains energy from all the input traces within the prestack migration aperture. This gather is based on the assumption that each point in the geological structure is regarded as a scatter point, from which the source energy may be scattered to any receiver. The CSP gathers are formed when samples from each input trace are assigned an equivalent offset, an offset that is based on the distance of the scatter point from the source and receiver.
A typical 2-D CMP gather will contain less than a quarter live traces, with the maximum offset limited to the source-receiver offset. A super gather of a number of CMP's is required to fill the empty offsets for velocity analysis. In contract, a CSP gather from a 2-D line may contain thousands of traces, with high fold in each offset bin, and contain offsets that are far beyond the range of the source-receiver offset. A 3-D CSP gather may contain many thousands of traces.
The CSP gather contains some energy from all the input traces that would migrate to the output trace. Similarly, each input trace will contribute some energy to all CSP gathers within its migration aperture. These input traces are sorted by offset, and added into offset bins for each CSP gather. Note that there is no time shifting of the input samples.
When the CSP gathers have been formed, each CSP gather may be scaled and filtered, or processed similarly to CMP gathers. Conventional algorithms such as noise and multiple removal, or velocity analysis, may also be used on the CSP gathers. Velocity analysis performed on the CSP gather will contain a more accurate velocity discrimination than those derived from NMO gathers. These improved discrimination results from using only one CSP gather, the high fold in the offset bins within the gather, and offsets that are much larger than the source-receiver offset. Note that the noise and multiples in the CSP gather now apply to the final prestack migration offsets, and not the offset CMP gathers in conventional processing.
After the CSP gathers have been formed, all that remains to complete the new pres tack migration process is conventional NMO and stacking. If desired, more complex forms of NMO may be evaluated and applied to the data in the CSP gather.
The quality of the CSP gather is dependent on the definition used for the equivalent offset. A number of options are available. A simple but effective definition is based on prestack Kirchhoff time migration. This method equates the travel times of the raypaths from the original source and receiver positions, with the ray path travel times of co-located source and receivers positioned at the equivalent offset. This definition also allows converted waves to be processed in a conventional manner.
The use of CSP gathers offers a number of other potential benefits. Many of the input traces are combined into the offset bins of the CSP gather, with a fold in each bin that may be in the ten's for 2-D data, and much higher for 3-D data. NMO is applied once for all the traces within the offset bin, with an arithmetic saving that is proportional to the fold in the offset bin. Other potential benefits of the new method may result in improved static solutions and better field designs.
About the Author(s)
Born and raised in Perth, Western Australia, John Bancroft's family emigrated to Calgary in the mid sixties. John completed his B.Sc. and M.Sc. at the University of Calgary and obtained his Ph.D. from Brigham Young University, all in Electrical Engineering. His first five professional years were spent working with the National Technical Institute for the Deaf, at Rochester Institute of Technology, on projects that ranged from speech analysis to cochlea implants.
In 1980 he returned to Calgary and commenced his career in the geophysical industry. He spent a number of years at Geo-X and Veritas working on geophysical research and advanced seismic applications. For the last four years, John has been a consultant to the industry. He is currently an Adjunct faculty member of the Department of Geology and Geophysics, and a Senior Research Geophysicist with the CREWES Project, at the University of Calgary.
John has a great deal of experience in developing software for seismic processing. He has written software that includes: 3-D acquisition and processing, surface consistent amplitude deconvolution and statics, modelling, migration and inversion. Migration has been his specialty for over twelve years, and he has written most of the major algorithms. He has developed a course to teach the practical aspects of migration, and is a SEG instructor.
John's interests include his family, observing hydraulic wave motion from a sandy vantage point, and studying the atmospheric standing waves above the foothills and Rockies. He is actively involved in the Scouting movement, and is a member of the ASEG, CSEG, SEG, EAEG, IEEE, and APEGGA.