Automatic velocity picking - A heuristic approach to pick and velocity function validation
Philip W. Kitchenside · 55th EAEG Meeting · 1993
Picking is an intrinsically interactive procedure, and although a batch picker may prove to be a good substitute for manual picking, a subsequent interactive step will always be required. A primary goal of batch picking is therefore to reduce the time spent subsequently on QC and editing. In essence, the approach described here regards automatic picking as the identification of semblance maxima followed by a heuristic driven selection of those maxima likely to be good picks. The method assigns an overall quality factor to each peak. Using the quality factors as the basis for selection, velocity functions are then constructed from a set of "best" picks. Cumulative heuristic functions are used to guide the search towards the "best" velocity function, which is assumed to provide the picking at the current location. It is also used to define the velocity trend, which is input to the picking at the next location along the line. The method makes use of prior knowledge stored in a database, (velocity functions picked and QC'd during a previous iteration, and interpreted time horizon information). In addition, there is some scope for supervised learning.