next up previous print clean
Next: Conclusions Up: Application to 3D real Previous: Data regularization

Migration after regularization

The next step after regularizing the coverage of the 3D subset is to apply 3D migration to the partial stack. Although at this stage any wave-extrapolation technique can be applied to the regularly sampled subset, I chose Kirchhoff migration for consistency in comparing the results of imaging before and after regularization.

Figure mig910 compares the results of migrating the 3D subset using different imaging flows. The Figure represents a depth slice (910 m) where differences are most noticeable between the results. The migration of the oversampled survey indicates a complex morphology of a meandering river system marked by ramification of the major channel. The output of migrating the irregularly sampled subset is very noisy and distorted by strong artifacts that make the interpretation of the channels difficult. The result of migrating the NMO-stack is smoother but displays poor resolution and the channels are also unresolved. Applying AMO to the sparsely sampled data produced aliasing noise in the partial stack and therefore degraded the quality of the migration. The result of migration after data-space ICO also failed to produce a good image of the river channel. In contrast, migration after regularization with model-space ICO unveiled much of the details in the image and resolved the different branches of the channel. Given the dominantly flat geology of the survey, it was expected that migration after NMO should provide a good image. Such result can be observed on the 920m depth slice which marks the floor of the river channel mig920. A plausible explanation for this phenomenon is that the morphology of the river system becomes more complex towards the top of the deposition sequence. This results into diffractions from the edges of levees and from possible barrier islands. While these diffractions were nicely preserved by ICO, they were destroyed during stacking by NMO.


next up previous print clean
Next: Conclusions Up: Application to 3D real Previous: Data regularization
Stanford Exploration Project
1/18/2001