Articles | Volume 17, issue 10
Research article
 | Highlight paper
19 Oct 2023
Research article | Highlight paper |  | 19 Oct 2023

Mapping Antarctic crevasses and their evolution with deep learning applied to satellite radar imagery

Trystan Surawy-Stepney, Anna E. Hogg, Stephen L. Cornford, and David C. Hogg


Total article views: 2,617 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
1,900 671 46 2,617 34 35
  • HTML: 1,900
  • PDF: 671
  • XML: 46
  • Total: 2,617
  • BibTeX: 34
  • EndNote: 35
Views and downloads (calculated since 31 Mar 2023)
Cumulative views and downloads (calculated since 31 Mar 2023)

Viewed (geographical distribution)

Total article views: 2,617 (including HTML, PDF, and XML) Thereof 2,543 with geography defined and 74 with unknown origin.
Country # Views %
  • 1


Latest update: 28 Feb 2024
This research is part of an exciting advancement in the field of glaciology, driven by machine learning. The study focuses on crevasse detection, a highly relevant topic from a scientific and logistic perspective. Crevasses may aid surface meltwater to penetrate through the ice thus impacting ice dynamics. Crevasses also pose a logistical challenge for fieldwork in the polar regions. In this study, the authors are able to automatically spot grounded crevasses using a Convolutional Neural Networks algorithm. One of the focus areas is the Thwaites Glacier, an area that has recently been subject to extensive scientific research due to its importance for the stability of the West Antarctic Ice Sheet.
Short summary
The presence of crevasses in Antarctica influences how the ice sheet behaves. It is important, therefore, to collect data on the spatial distribution of crevasses and how they are changing. We present a method of mapping crevasses from satellite radar imagery and apply it to 7.5 years of images, covering Antarctica's floating and grounded ice. We develop a method of measuring change in the density of crevasses and quantify increased fracturing in important parts of the West Antarctic Ice Sheet.