Datapages, Inc.Print this page

A Novel Technique for Fracture Data Analysis Using Three-Dimensional Signal Processing and Statistical Cluster Analysis: Outcrop Examples from the Grand Canyon, U.S.A. and Gulf of Suez, Egypt


Pigott, John D.1, Louis C. Niglio2, Andy Rich3 (1) University of Oklahoma, Norman, OK (2) Minerals Management Service, New Orleans, LA (3) Samson Investment Company, Tulsa, OK


Cluster analysis and signal processing techniques can effectively discriminate clear sig­nals from noise when applied to fracture populations, i.e. delineate spatially and temporally discrete fracture populations. Fracture analysis was performed in outcrop from 11 field sta­tions along the Gulf of Suez margin, Republic of Egypt, and 29 stations within the Grand Canyon. The data were separated into spatial and temporal populations by 3D cluster analy­sis, transferred into the frequency domain, 3D deconvolved, and analyzed statistically both among and between stations.

For the Gulf of Suez, three populations of para-coulombic fracturing are apparent: a Mesozoic inherited northeast to west-northwest coupled set, a Paleogene to Neogene north­northeast coupled northwest set, and a Neogene northwest set. For the Grand Canyon, clus­ter analysis indicates three dominate bi-modal fracture patterns and identifies three domi­nating sets of NE, WNW, NW, and a weakly developed E-W set.

Analysis of paleostresses for the Gulf of Suez appear to be related to specific tectonic events predating, coeval with, and post-dating the crustal extension of the Gulf of Suez with overprinting by the sinistral opening of the Gulf of Aqaba. For the Grand Canyon, NE and NW fracture sets appear related to bending of cover rocks above crystalline basement trends while the WNW fracture set may have formed under a post-Permian regional tectonic stress.

The 3D signal processing and cluster analysis technique appears to be robust and to hold great promise for the potential use of fracture mechanical stratigraphy and interpreted paleostress analysis even in noisy data.