Fracture aperture calculation
To normalize CT-scan datasets, we fit a Gaussian function to the
distribution of CT numbers to obtain a CT-number mean
(m x) and standard deviation
(s x), where x is either S1, S2, or S3. To compare
datasets acquired at different stages, we shifted the CT-numbers of
datasets S2 and S3 bym S1-m S2 andm S1-m S3, respectively. We
added a value of 1 to each voxel, cropped each image to 718x718 pixels
around the sample center, and assigned a value of 0 to pixels with a
distance >718/2 from the sample center. We binarized the
datasets to assign each voxel to either solid rock or air by applying a
threshold calculated ast x=m x-2.5s x. Voxels with CT-number equal to or greater
than t x were assumed to represent rock and
assigned a value of 255. Voxels with CT-number lower thant x and greater than zero were assumed to be air
and assigned a value of 128.
To obtain a FADP of a binarized dataset, we calculated: 1) The Euclidian
distance of each voxel in the fracture. This is achived by a) performing
an iterative image morphological erosion assigning approximated
distances of each fracture voxel from the fracture rim; and b)
calculating the Euclidian distance of each voxel within the fracture
from the closest voxel representing rock; 2) The skeleton of the
fracture (SK) consisting in the voxels that are within the fracture and
report the maximum Euclidian distance from the fracture rim into respect
the 26 surronding voxels. Such a device extracts the center surface
while preserving the topology and Euler number, also known as the Euler
characteristic of the objects (Kerschnitzki et al., 2013; Lee et al.,
1994). Finally, the FADP was calculated at each SK location by doubling
the Euclidian distance recorderded in such voxels.