The geological interpretation of seismic image volumes depends greatly on the careful mapping of its faults. The exact location and throw of faults are an important ingredient in a geologist's reconstruction of the subsurface's historic sedimentary and structural processes. Additionally, faults potentially trap or leak hydrocarbons. Picking fault surfaces in seismic image volumes is time consuming and error prone, since a fault is easily overlooked. Additionally, the fault lines, which are picked in several two-dimensional image sections, have to be combined into fault surfaces of the image volume.
Current discontinuity attributes cannot replace the picking of faults by interpreters, but they offer an alternative view and they are quick, reproducible, and three-dimensional.
The quality of a discontinuity map ultimately is the ease with which it is interpreted reliably. To that end, a discontinuity attribute needs to suppress the image's sedimentary bedding elements and isolate the image's faults and other discontinuities. Furthermore, the image should resolve any discontinuity as well as possible. In general, the discontinuity image should be free of noise and high in contrast. Beyond such fundamental criteria, every interpreter may have personal preferences.
I found it difficult to develop an image processing algorithm without the guiding principles of physical laws.