To what extent do fracture arrays conform to Andersonian fault models?

E.M. Andersons seminal work was founded on failure criteria developed by Coulomb, Navier and Mohr. Anderson applied these failure criteria to explain aspects of faulting in the crust, even though they were empirically derived from laboratory experiments on small rock samples that were subjected to uniaxial loading and that could not be tested beyond initial failure. Our study aims to test the degree to which fracture systems that develop in larger (outcrop-scale) rock volumes conform to Andersons predictions.

We have measured the geometry of over 25,000 fractures from more than 30 outcrop localities worldwide. The common characteristic of these outcrops is that the amount of fracturing is moderate to high. Fractures analysed include faults, joints and deformation bands. The fractures cut a range of host lithologies including carbonate, clastic, crystalline and mixed carbonate/clastic sequences. Tectonic regimes believed to have been active at the time of fracture formation include extensional, compressional and oblique settings. Finite strains vary from large total displacement across the outcrop, to areas of jointing in which total strain is extremely low. There is a wide range of ages of deformation, and some localities show evidence of reactivation: critically, however, a number of outcrops are in areas of active tectonics and contain fracture sets that cut rocks of Paleogene to late Neogene age.

The detailed morphology of the outcrop surface and exposed fractures was recorded using a terrestrial laser scanner (ground-based lidar). This technology provides very rich datasets in which the precise geometry of individual fractures can be observed and measured, at centimetre to kilometre scales. The fractures measured in outcrop include fracture traces (where the fracture intersects the surface of the outcrop) and exposed fracture surfaces (where the fracture forms a part of the outcrop surface).

Collectively, the datasets display a number of general characteristics:
Faults and joints are rarely planar. Most have variable dip and strike along their length (i.e. curved fracture geometries are typically helical rather than cylindrical sections).
Longer fractures have higher likelihood to consist of several short segments, but usually the segments adjacent to each other are not parallel and not co-planar.
Bulk orientation data from fractured outcrops typically show clusters (fracture sets) in which there is an inherent spread of orientation values. The amount of dispersion varies, but 20 or more is usual within each cluster. This spread in orientation values seems to be present for unimodal, bimodal and polymodal fracture sets. The variation in orientation is very much greater than the precision of the lidar equipment and the methods of analysis: the dispersal of the orientation data is not due to measurement error, or of small-scale localised extent, but should be regarded as the normal geometry for fracture sets.