Abstract. The stochastic generation of discrete fracture networks (DFN) is a method for
modelling fracture patterns used to assess the in situ fragmentation in a
volume of rock. The DFN modelling approach is based on the
assumption that the natural fragmentation of rocks is a function of the length and
connectivity of the fractures within the considered volume of rock. Thus, in order to generate a site-specific DFN, the
primary geometric properties of the fracture surfaces within the
rock volume (especially orientation, size and fracture intensity as well as
the local spatial variability) must be defined as distribution functions (Elmo
et al., 2014). The required base statistics are usually obtained from fracture
analysis on boreholes, exposed rock surfaces or (to a limited extent) 3D
seismics (e.g. Bisdom et al., 2014; Bemis et al., 2014). We adopted a terrestrial close-range photogrammetry approach to capture
several outcrops and analyse fracture traces on the exposed rock surfaces, the chosen
workflow is based around the use of free and open-source software. Images were
acquired from several quarries in the Weschnitzpluton, a
granodioritic to quartz monzodioritic pluton in the Bergstrasse Odenwald
(e.g. Altherr et al., 1999) using a consumer-grade Nikon D5300
DSLR with fixed focal length instead of a drone or Lidar-system for legal
reasons, partially tree-lined outcrops and cost efficiency. Since point clouds
obtained from photogrammetry are inherently dimensionless, we used a spherical
target with compass and bubble level for scale and proper spatial orientation
(Froideval et al., 2019). The exact geolocation is not particularly important
for the task, so the use of GPS, total station or georeferenced ground control
points is not necessary. Dense point clouds were computed using the open
source SfM photogrammetry suite Meshroom (AliceVision, 2021), which can be
used for manual or semi-automatic detection of fracture surfaces and their orientation
(Schnabel et al., 2007) and to generate orthorectified images of the rock surface
to trace fracture lengths and nodes in a GIS (Nyberg et al., 2018). Our
investigations proved terrestrial photogrammetry to be a valuable and easily accessible tool in the
documentation of natural fracture patterns and a robust base for the
generation of DFN networks.