Document Type

Honors Thesis

Major

Earth Sciences

Advisor(s)

Seth Campbell

Committee Members

Annie Boucher, Peter Koons, Melissa Ladenheim, Kristin Schlid

Graduation Year

May 2020

Publication Date

Spring 5-2020

Abstract

Understanding glacial erosion rates is important because debris eroded by a glacier can impact glacier flow speeds, protect tidewater glaciers from rapid retreat, and impact the productivity of marine ecosystems. Traditionally, glacial erosion models rely on a rock’s inherent “erodibility”, typically presented as a constant, to predict how much debris will be eroded by the glacier. However, the erodibility of bedrock varies spatially as a function of its fracture density, fracture orientation, and lithology, so the notion of applying a constant erodibility term to a whole field site does not fully capture the actual bedrock dynamics of the system. In this work, I present a novel approach to quantify bedrock fracture density and orientation through the generation of a 3D Structure from Motion (SfM) model and the application of a series of machine learning algorithms. To test this approach, I quantified the fracture density of a glacial bedrock nunatak in the Juneau Icefield of Southeast (SE) Alaska. The spatial variation in fracture density across this nunatak was found to be highly variable. Bedrock in the SE region of this field site showed a relatively high fracture density (>20% fractured), whereas the central region of this field site showed a relatively low fracture density (0-10% fractured). Fracture orientations were shown to have a bimodal distribution, with the most common fracture orientations being approximately 0 and ± 90 degrees. This fracture density methodology and associated results can applied across the Juneau Icefield and other glacier systems to improve glacial bedrock erosion models.

Download

Included in

Glaciology Commons

Share