Interpretation of Structures in Polar Regions with Deep Learning Methods

Kurzfassung

Global climate change is rapidly changing the polar regions. In an effort to support monitoring these changes, this thesis develops deep learning methods for the remote sensing analysis of targets in these regions. Firstly, models for mapping glacier calving fronts are developed by rethinking how the task is encoded computationally. Secondly, the feasibility of detecting retrogressive thaw slumps in permafrost areas is established and made data-efficient through semi-supervised learning.