Preprints
https://doi.org/10.5194/egusphere-2023-858
https://doi.org/10.5194/egusphere-2023-858
11 May 2023
 | 11 May 2023
Status: this preprint is open for discussion.

Mapping the extent of giant Antarctic icebergs with Deep Learning

Anne Braakmann-Folgmann, Andrew Shepherd, David Hogg, and Ella Redmond

Abstract. Icebergs release cold, fresh meltwater and terrigenous nutrients as they drift and melt, influencing the local ocean properties and encouraging sea ice formation and biological production. To locate and quantify the fresh water flux from Antarctic icebergs, changes in their area and thickness have to be monitored along their trajectories. While the locations of large icebergs are tracked operationally by manual inspection, delineation of their extent is not. Here, we propose a U-net approach to automatically map the extent of giant icebergs in Sentinel-1 imagery. This greatly improves the efficiency compared to manual delineations, reducing the time for each outline from several minutes to less than 0.01 sec. We evaluate the performance of our U-net and two state-of-the-art segmentation algorithms on 191 images. For icebergs, larger than covered by the training data, we find that U-net tends to miss parts. Otherwise, U-net is more robust to scenes with complex backgrounds, ignoring sea ice, smaller patches of nearby coast or other icebergs and outperforms the other two techniques achieving an F1 score of 0.84 and an absolute median deviation in iceberg area of 4.1 %.

Anne Braakmann-Folgmann et al.

Status: open (until 06 Jul 2023)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse

Anne Braakmann-Folgmann et al.

Video supplement

Segmentation maps of giant Antarctic icebergs Anne Braakmann-Folgmann https://doi.org/10.5281/zenodo.7875599

Anne Braakmann-Folgmann et al.

Viewed

Total article views: 280 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
181 94 5 280 2 1
  • HTML: 181
  • PDF: 94
  • XML: 5
  • Total: 280
  • BibTeX: 2
  • EndNote: 1
Views and downloads (calculated since 11 May 2023)
Cumulative views and downloads (calculated since 11 May 2023)

Viewed (geographical distribution)

Total article views: 282 (including HTML, PDF, and XML) Thereof 282 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 07 Jun 2023
Download
Short summary
In this study, we propose a deep neural network to map the extent of giant Antarctic icebergs in Sentinel-1 images automatically. While each manual delineation requires several minutes, our U-net takes less than 0.01 sec. In terms of accuracy, we find that U-net outperforms two standard segmentation techniques in most metrics and is more robust to challenging scenes including sea ice, coast and other icebergs. The absolute median deviation in iceberg area across 191 images is 4.1 %.