Causes of bias and uncertainty in fracture network analysis

Abstract

Fault and fracture networks are analysed to determine the deformation history and to help with such applications as engineering geology and fluidflow modelling. These analyses rely on quantifying such factors as length, frequency and connectivity. Measurements may, however, be influenced by a range of factors relating to resolution, geology, methods used and to the analyst(s). These factors mean that it can be difficult to obtain a single correct solution, with bias and uncertainty being introduced by different analysts, even for something as simple as counting the number of joint intersection points on a well-exposed bedding plane. These problems suggest there are significant issues in comparing databases, for example when using outcrop analogue data to model subsurface data. Our recommendation is that analysts and modellers should be aware of the potential pitfalls in their measurements of structures and, therefore, be more cautious with resultant analyses and models. We suggest that analysts assess their results by testing the reproducibility. Simple ways of doing this include: (1) checking for change in measurements (e.g., fracture frequencies) during the course of a study; (2) remeasuring part of the fracture network to check if the same results are obtained, and; (3) get one or more other analysts to blind-test the fracture network.

More information

Identifiers

ORCID for David J. Sanderson: orcid.org/0000-0002-2144-3527

Catalogue record

Export record

Altmetrics