A PARTIALLY AUTOMATED WORKFLOW TO ANALYZE FAULT SCARP MORPHOLOGY USING UAV-ACQUIRED IMAGERY

Morphometric dating of fault scarps could provide quick insight into active tectonics of a region, if morphometric age can be confidently determined. Descriptions and measurements from a large number of fault scarps are required in order to advance our understanding of morphometric dating. A standardized approach to analyzing fault scarp data will permit easy comparison of data to identify factors that govern the evolution of fault scarps. We propose such a workflow for fault scarp data generated from structure-from-motion (SfM) photogrammetry with images acquired by UAV.

Overlapping images captured with an UAV and commercial SfM software enable us to produce point clouds representing the ground surface and orthorectified images of geomorphic and structural features. We present UAV-SfM-derived datasets from bedrock fault scarps in southwestern Iceland and northeastern California. From the point cloud, we extract topographic profiles perpendicular to the fault scarp at set intervals along the strike. We use curve-fitting algorithms to extract quantitative information, such as the curvature, from the scarp profile. Along strike variations in scarp morphology can be assessed visually from these profiles, and the degree of variation is quantified using a supervised k-nearest-neighbor classification algorithm. Using this algorithm, we train the classifier with topographic profiles of known morphologic classes (e.g. vertical free face or talus slope), then apply it to an entire scarp, thus automatically assigning class membership to profiles of unknown morphologic classes. Additionally, we characterize talus size and orientation both in the field and using measurements taken from the SfM model. Using the point cloud, we also estimate the volume of talus produced, a useful proxy for the degree of scarp degradation. To characterize local bedrock structure, we use orthorectified images of the scarp face in order to map discontinuities and assess fracture patterns and densities. We can thus compare a scarp’s shape with a variety of factors that might affect scarps, such as geologic unit, position of the scarp profile along the fault segment, and distribution of joints and fractures in the faulted material, to ultimately untangle how scarp morphology evolves through time.