The usage of open-source aerial images for the characterisation of a fracture network. Insights from a multi-scale approach in the Zagros Mts.

The characterisation of the fracture network is a fundamental step for modelling the circulation of different types of geofluids at multiple scales, including at the reservoir scale. Due to the difficulty of sampling the fracture network with a spatial continuity, fracture networks are often derived from a stochastic approach applied to ground truth parametrized datasets. Classical field methods for collecting and analysing fracture data sets (i.e., scan-lines or scan-areas) have limited sample area and the results might be affected by local factors (e.g., facies variations and faults). In poorly- to non-vegetated regions, field-derived data can be integrated with satellite and aerial images providing a “big-picture” of the study area. The availability of open-source, high-resolution aerial and satellite images (e.g., Google Maps, Bing Maps) coupled with the development of specific software and tools (e.g., NetworkGT; FracPaQ) allow to digitize and analyse very large and spatially continuous data sets of fractures at multiple scales for a single case study.

In this work, we test the use of Bing Maps imagery for the analysis of the fracture network affecting the Kuh-e-Asmari anticline, in the Zagros fold-and-thrust belt. The Kuh-e-Asmari anticline is considered as an outcropping analogue of fractured reservoirs and is in a scarcely vegetated area, covered by high resolution open-source aerial images (~2.6 m/pixel). We obtained three different image-derived fracture data sets by manually digitizing the fracture network at three different scales (1:50000, 1:5000, and 1:500) in QGIS. Each data set has been analysed using the NetworkGT plugin within QGIS. In detail, we analysed the orientation, length distribution, abundance, and topology of each fracture network data set, and we have compared the results with structural data from scan-lines performed in the field on the same anticline.

Results proved to be scale-dependent, with each scale having its pros and cons, as follows. The 1:50000 data set is the only spatially continuous data set, allows to rapidly map the main tectonic features (e.g., major faults and fracture corridors) but it is not accurate enough in terms of the definition of orientation sets and length analyses. The 1:5000 data set is potentially spatially continuous and allows to analyse fracture and connectivity distributions with high detail, by highlighting strongly fractured/connected elongated zones, possibly representing damage zones of known, previously mapped, fault strands. The 1:500 scale cannot guarantee the spatial continuity of the data set but, if applied to a small area, provide reliable results.

Results obtained in this work suggest that the manual interpretation of open-source aerial images is a viable way for the characterization of fracture networks in poorly vegetated areas only if the right scale is chosen. For the case study presented here we suggest interpreting the fracture network at 1:5000 or similar scales. A successful testing of semi-automatic or automatic algorithms for lineament detection is required to perform a spatially continuous interpretation and analysis at the highest possible resolution, and/or to analyse the fracture dataset at multiple scales (more than the 3 investigated here).