Statistical Analysis of Lineaments and Their Relation to Fracturing, Faulting, and Halokinesis in the East Texas Basin

Lineament analysis is part of a broad spectrum of structural studies employed to determine the tectonic stability of the East Texas Basin. Such information is necessary to assess the suitability of East Texas salt domes as possible repository sites for the storage of high-level nuclear wastes. A sequence of statistical operations was designed to identify and assess the significance of lineament preferred orientation by means of a variety of statistical tests or parameters, including vector summation, length weighting, X2, F, and t testing, the Bernshtein accuracy criterion, and an index of preferred orientation. Black-and-white aerial photographs, at scales between 1:17,400 and 1:25,500, and band-5 Landsat imagery were analyzed. Well-defined, northeast-trending and northwest-trending lineament populations are present throughout the East Texas Basin. The northeast trend, comprising two peaks oriented at 045° and 055°, corresponds to the orientation of the Mexia-Talco peripheral fault zone, to subsurface faults in the center of the basin, and to some lithologic contacts. The northwest trend comprises two peaks oriented at 310° and 325°. Both the northeast and northwest trends are thought to result from preferential directions of fracture induced by interference folding at depth. This folding is caused by halokinesis and is reflected in the regional gravity field. The Elkhart - Mount Enterprise fault zone has exerted little noticeable effect on the regional lineament pattern, mainly because of its subparallel orientation. Areas above shallow salt domes, particularly those in the southern part of the basin, are associated with higher lineament densities and lower preferred orientation of lineaments than are non-dome areas or areas above deep salt diapirs; this probably reflects radial and concentric fault and fracture patterns above the shallow domes. Analysis of computer-generated, geologically meaningless sets of “lineaments” strongly suggests that confidence levels of 99 percent are necessary to exclude randomly generated peaks, and that the significance of orthogonal pairsets has been exaggerated in the literature.