Abstract: Multi-Attribute Computer-Aided Seismic Stratigraphy: Semi-Automated Prediction of Depositional Environments

Samuel D. Leroy

Multi-attribute seismic displays give us new freedom and flexibility in finding stratigraphic information in seismic data. By using multiple seismic attributes together, they allow automated classification and prediction of depositional environments. Used in conjunction with well data and traditional seismic stratigraphic techniques, this method helps produce a low cost and rapid picture of reservoir geometry at the prospect and regional scales.

Actual implementation of these methods is most easily done using inexpensive image-analysis software that includes similarity classification functions. Two approaches have been taken: (1) statistical description, using a multi-dimensional statistical space, such as reflection strength vs. frequency versus amplitude, and (2) spatial filtering to produce a similarity classification based on frequency and reflection character.

Both approaches are relatively simple. Statistical classification requires that primary trace attributes be extracted from the seismic data set, but is ultimately the more powerful tool. Spatial filtering at its simplest implementation requires only a paper copy of a seismic line to start with, although classification accuracy can be increased when several different kinds of displays are input. Processing which can be used with spatial filtering includes "true" amplitude, expanded trace, instantaneous phase, and frequency.

Results from these different methods tend to support each other, indicating that the natural signal we are getting back from the rocks is strong; nature has already separated the sediments into distinct physical units with differing cyclicity and lithologic contrasts.

AAPG Search and Discovery Article #90986©1994 AAPG Annual Convention, Denver, Colorado, June 12-15, 1994