Master thesis defense by Dina Rapp

Title: Ice flow and machine learning

From the IPPC reports there is a clear difference between the projected ice melt and the actual ice melt. The current models do not capture the ice dynamics that explain this extra melt. This project aims to

explore seasonal patterns across the Greenland Ice-Sheet in a quantitative K-Means clustering, as well as by qualitatively investigating selected glaciers. Seasonal features of the whole grid are calculated and

used in combination with the clustering results. The project is heavily based on the articles Moon et al. (2014), Vijay (2019) and Vijay (2021), where three distinct seasonal velocity patterns are identified. Two

of them are linked to runoff and one to changes in the terminus. The two patterns related to runoff are also identified in the clustering performed in this project. The same interaction with runoff is observed.

The spatial and temporal distribution of the clusters is investigated and compared to the results from the mentioned articles. Similarities are found, and further insights are gained. Some patterns not identified by the mentioned articles are observed in this study.

 Vejledere: Christine Schøtt Hvidberg og Anne M. Solgaard (GEUS). Censor: Sebastian B. Simonsen (DTU)