Artificial Neural Networks: A New Landscape Characterization Toolbox

This project uses self-organizing maps, a form of artificial neural network techniques, to determine the role of glacial processes in uplift of the Transantarctic Mountains (TAM). The link between erosion and mountain building is known, but difficult to resolve because the processes are dynamically coupled. The TAM offer a unique opportunity to explore this relationship because they have been tectonically quiescent since the late Oligocene, but have undergone various forms of glaciation. Previous investigations of their landscape evolution have been based primarily on field inspection of landforms and surface features, a time intensive and logistically difficult undertaking. While these studies have documented the varying climatic influences on the TAM and the importance of fluvial, glacial, and desert processes in shaping the landscape, we are still left without an accounting of the impact of glacial erosion along the TAM as a whole. This project uses digital elevation models and self-organizing map analysis to provide quantitative metrics that characterize glacial landforms and determine the overall role of glaciation in mountain-building.

The broader impacts of this work are improving society's understanding of global climate change by demonstrating links between climate and landform.