Semi-automatic mapping of fault rocks on a Digital Outcrop Model, Gole Larghe Fault Zone (Southern Alps, Italy)

Andrea Bistacchi and Silvia Mittempergher. ( 2016 )
in: 2016 RING Meeting, ASGA

Abstract

The use of Digital Outcrop Models (DOMs) based on LIDAR or photogrammetry greatly increased the mass of data available from field surveys. However, the manual interpretation of such datasets for quantitative structural analysis is still a time consuming process, limiting the effective usage of very large datasets. We present a semi-automatic workflow for mapping lineaments, such as fault-fracture traces, on DOMs, with the aim of speeding up the interpretation of field data. The DOMs are reconstructed with photogrammetric techniques using a large number of high resolution digital photographs processed with VisualSFM (Surface From Motion) software. The SFM algorithm links directly each pixel in the images to the corresponding point on the DOM. This linkage allows to process the high resolution images using image analysis techniques, and to project the extracted features on the DOMs without losing resolution. We developed a MATLABĀ® toolbox which extracts the fault-fracture traces from images in three steps: (i) fault traces from images: after preprocessing, faults are extracted using an edge detection algorithm. After several tests, we found that a shearlet/wavelet-based algorithm as the most effective in discriminating linear elements (faults and fractures) from the wall rock texture; (ii) fault trace connection: after vectorization, collinear segments are automatically connected; (iii) supervised validation: relevant fault segments are manually selected, discarding spurious results. The vector fault-fracture traces are then topologically corrected and projected on the DOM. The method has been tested on DOMs of fault and fractured zones in both crystalline (tonalite) and sedimentary (limestone) host rocks. The best results are obtained in outcrops of fractured limestone, because of its relatively homogeneous background texture. However, the shearlet-based edge detection algorithm proved to be effective in discriminating linear features also in strongly textured rocks as tonalite.

Download / Links

BibTeX Reference

@INPROCEEDINGS{,
    author = { Bistacchi, Andrea and Mittempergher, Silvia },
     title = { Semi-automatic mapping of fault rocks on a Digital Outcrop Model, Gole Larghe Fault Zone (Southern Alps, Italy) },
 booktitle = { 2016 RING Meeting },
      year = { 2016 },
 publisher = { ASGA },
  abstract = { The use of Digital Outcrop Models (DOMs) based on LIDAR or photogrammetry greatly increased
the mass of data available from field surveys. However, the manual interpretation of such datasets for
quantitative structural analysis is still a time consuming process, limiting the effective usage of very large
datasets. We present a semi-automatic workflow for mapping lineaments, such as fault-fracture traces, on
DOMs, with the aim of speeding up the interpretation of field data.
The DOMs are reconstructed with photogrammetric techniques using a large number of high
resolution digital photographs processed with VisualSFM (Surface From Motion) software. The SFM
algorithm links directly each pixel in the images to the corresponding point on the DOM. This linkage
allows to process the high resolution images using image analysis techniques, and to project the extracted
features on the DOMs without losing resolution. We developed a MATLABĀ® toolbox which extracts the
fault-fracture traces from images in three steps: (i) fault traces from images: after preprocessing, faults are
extracted using an edge detection algorithm. After several tests, we found that a shearlet/wavelet-based
algorithm as the most effective in discriminating linear elements (faults and fractures) from the wall rock
texture; (ii) fault trace connection: after vectorization, collinear segments are automatically connected;
(iii) supervised validation: relevant fault segments are manually selected, discarding spurious results. The
vector fault-fracture traces are then topologically corrected and projected on the DOM.
The method has been tested on DOMs of fault and fractured zones in both crystalline (tonalite) and
sedimentary (limestone) host rocks. The best results are obtained in outcrops of fractured limestone,
because of its relatively homogeneous background texture. However, the shearlet-based edge detection
algorithm proved to be effective in discriminating linear features also in strongly textured rocks as
tonalite. }
}