Seismic interpreters routinely use the shape of an interpreted surface in developing prospects, in which the classic hydrocarbon trap is a ridge-shaped anticline. Carbonate buildups may appear as dome-shaped and karst collapse features as bowl-shaped. Differential compaction often results in valley shapes over shale-filled channels.
The interpretational value of a given shape is dependent on its depositional, diagenetic, and tectonic deformation context. If the channel fill is sand and the surrounding matrix shale, differential compaction can result in an incised valley appearing as a ridge, thereby providing a lithologic indicator. In flat-lying carbonates, joints will often be diagenetically altered and appear as valleys, while fracture intersections will appear as bowls. As always, the interpreter needs to be aware of the seismic data quality. In areas of limited lateral and vertical resolution, diffuse, or poorly-imaged faults may give rise to a recognizable shape anomaly. Care needs to be taken where velocity pull-up may induce deeper ridges and push-down deeper valleys on what might actually be flat structure.
Coupled with coherence, which delineates reflector edges, volumetric shape helps us rapidly recognize structural and stratigraphic style on horizontal and vertical slices. Pop-up blocks may appear as ridges bounded on both sides by low-coherence faults. Listric faults may be associated with a ridge-shaped roll-over anticline. Gas- and water-charged debris flow that can be drilling hazards may appear as high-coherence, dome shaped blocks.
Quantitative measures of reflector shape computed from uninterpreted seismic volumes are a byproduct of volumetric curvature. Volumetric curvature is now well-established in the interpretation community, with work flows developed to correlate healed fracture zones to ridges in shale plays to help guide hydraulic fracture stimulation programs. More recently, advances have been made in the volumetric quantification of pinch-outs and unconformities, providing images of both the magnitude and azimuth of reflector convergence. There is no “best” attribute. Rather, one should co-render mathematically independent attributes that are coupled through the underlying geology.
I will illustrate these concepts through application to land data volumes from North America.