Fracture characterization and modeling from virtual outcrops

Abstract

Advances in virtual outcrop technologies and their introduction to fracture characterization allow extraction of fracture data from very large and inaccessible areas. The recent development of automated or semiautomated methods for fracture extraction aims to reduce or avoid tedious, time-consuming, and biased manual interpretation of fractures from virtual outcrops. We present a benchmarking exercise between a previously proposed automated fracture picking method, manual picking, and fieldwork methods. Comparison between the three methods highlighted their relative advantages and limitations. The automated fracture picking method provided excellent results in terms of fracture orientation, size, spatial distribution, and density. Fieldwork is complementary to fracture extraction from virtual outcrops, and it should focus on quality control of remote sensing data, poorly exposed areas, small-scale observations, diagenesis, timing of fracture development, building conceptual models, and linking fracture stratigraphy to rock properties. We propose a best practice for the use and integration of manual and/or automated fracture extraction from virtual outcrop and fieldwork data for fracture characterization and modeling from outcrop analogs. We consider integration of different methods as the best way to improve the modeling exercise while reducing operational costs and risks.