Subglacial erosion and the form of a glacier's bed are controls of a variety of processes, notably glacier mass balance change. Quarrying dominates glacial landscape evolution in hard bedded environments. Bedrock fracture is a control on the rate of quarrying. The specific nature of this controlling relationship is not well parametrized at the regional scale within a glacial forefield. In order to parametrize such a relationship, quarried surfaces must be differentiated from unquarried surfaces at a wide scale. We create Structure from motion (Sfm) models to provide a dataset for this classification to be performed digitally. Quarried surfaces are described using engineering parameters in comparison to unquarried surfaces in order to determine the feasibility of the implementation of machine learning (ML) to make such a classification. Significant differences are found in curvature as well as textural and anisotropic roughness parameters between quarried and unquarried surfaces, showing the potential of ML to segment a digital model of a glacial forefield and identify quarried surfaces.
"},{"label":"Degree","value":["Bachelor of Arts with Departmental Honors"]},{"label":"Author-supplied Keywords","value":["Quarrying","Surface roughness","Curvature"]},{"label":"Administrative Note","value":"Drupal Node ID: 33893