FROM FIELD TO SCREEN: EXPERIMENTING WITH DIGITAL EXTRACTION OF GEOSPATIAL DATA

Advances in the digital acquisition of geospatial data have facilitated a transition from in-situ techniques (e.g., tape measures and Brunton compasses) to digital approaches. Using photogrammetry in tandem with the extraction of geospatial data from digitized models now enables data collection from structures that may otherwise be inaccessible. In this study, 3D computer models of meter-scale folds were created using an outcrop located near Kutztown, Pennsylvania. Individual folds were photographed and imported to Agisoft Metashape, where they were converted to models via photogrammetry. This process involves aligning photos, generating a dense cloud, and creating a mesh formed from the dense cloud. The resulting digital model was then processed using SolidWorks in preparation for use within Altair® HyperMesh®. With HyperMesh®, we can process the structure lines, construct the geological volume, and further discretize the volume into meshes (including nodes and elements). Then, based on the coordinates of the mesh, we utilized the MATLAB script developed by Welker et al. (2019) to extract geospatial data. The MATLAB script uses of the triangulated mesh to calculate normals (i.e. poles) to the mesh planes. Comparisons were then made between in-situ data (e.g., strike and dip of surfaces measured with a Brunton compass) and digitally extracted data from the same outcrop. Comparisons revealed that the quality of the model plays a large role in the accuracy of extracted digital data. Model quality originates with effective planning and execution of the photography stage of photogrammetry, where a consistent overlap, orientation, and camera model are crucial to model accuracy. Our first outcrop model was a poor representation of the outcrop’s real-world geometry, due in part to inconsistent camera orientation and the unanticipated effects of camera preprocessing (e.g., sharpening, reducing noise, etc.). Preprocessing caused issues with alignment, which led to distortion in the final model. By re-photographing the outcrop with careful attention to photo overlap and orientation, and using a camera without automatic photo corrections, subsequent models were far more representative of the outcrop. Preliminary results show that the geospatial output of digital models represents the real-world counterpart, and those digital models can offer advantages in data density, field safety, and convenience.