Mapping the extent of giant Antarctic icebergs with deep learning
Icebergs release cold, fresh meltwater and terrigenous nutrients as they drift and melt, influencing the local ocean properties, 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 b...
Published in: | The Cryosphere |
---|---|
Main Authors: | , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Copernicus Publications
2023
|
Subjects: | |
Online Access: | https://hdl.handle.net/10037/32655 https://doi.org/10.5194/tc-17-4675-2023 |
id |
ftunivtroemsoe:oai:munin.uit.no:10037/32655 |
---|---|
record_format |
openpolar |
spelling |
ftunivtroemsoe:oai:munin.uit.no:10037/32655 2024-02-11T09:58:44+01:00 Mapping the extent of giant Antarctic icebergs with deep learning Braakmann-Folgmann, Anne Christina Shepherd, Andrew Hogg, David Redmond, Ella 2023-11-09 https://hdl.handle.net/10037/32655 https://doi.org/10.5194/tc-17-4675-2023 eng eng Copernicus Publications The Cryosphere Braakmann-Folgmann, Shepherd, Hogg, Redmond. Mapping the extent of giant Antarctic icebergs with deep learning. The Cryosphere. 2023;17(11):4675-4690 FRIDAID 2215855 doi:10.5194/tc-17-4675-2023 1994-0416 1994-0424 https://hdl.handle.net/10037/32655 Attribution 4.0 International (CC BY 4.0) openAccess Copyright 2023 The Author(s) https://creativecommons.org/licenses/by/4.0 Journal article Tidsskriftartikkel Peer reviewed publishedVersion 2023 ftunivtroemsoe https://doi.org/10.5194/tc-17-4675-2023 2024-01-25T00:08:04Z Icebergs release cold, fresh meltwater and terrigenous nutrients as they drift and melt, influencing the local ocean properties, 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 operationally tracked 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 s. We evaluate the performance of our U-net and two state-of-the-art segmentation algorithms (Otsu and k-means) on 191 images. For icebergs larger than those covered by the training data, we find that U-net tends to miss parts. Otherwise, U-net is more robust in scenes with complex backgrounds – ignoring sea ice, smaller regions of nearby coast or other icebergs – and outperforms the other two techniques by achieving an F1 score of 0.84 and an absolute median deviation in iceberg area of 4.1 %. Article in Journal/Newspaper Antarc* Antarctic Iceberg* Sea ice The Cryosphere University of Tromsø: Munin Open Research Archive Antarctic The Cryosphere 17 11 4675 4690 |
institution |
Open Polar |
collection |
University of Tromsø: Munin Open Research Archive |
op_collection_id |
ftunivtroemsoe |
language |
English |
description |
Icebergs release cold, fresh meltwater and terrigenous nutrients as they drift and melt, influencing the local ocean properties, 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 operationally tracked 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 s. We evaluate the performance of our U-net and two state-of-the-art segmentation algorithms (Otsu and k-means) on 191 images. For icebergs larger than those covered by the training data, we find that U-net tends to miss parts. Otherwise, U-net is more robust in scenes with complex backgrounds – ignoring sea ice, smaller regions of nearby coast or other icebergs – and outperforms the other two techniques by achieving an F1 score of 0.84 and an absolute median deviation in iceberg area of 4.1 %. |
format |
Article in Journal/Newspaper |
author |
Braakmann-Folgmann, Anne Christina Shepherd, Andrew Hogg, David Redmond, Ella |
spellingShingle |
Braakmann-Folgmann, Anne Christina Shepherd, Andrew Hogg, David Redmond, Ella Mapping the extent of giant Antarctic icebergs with deep learning |
author_facet |
Braakmann-Folgmann, Anne Christina Shepherd, Andrew Hogg, David Redmond, Ella |
author_sort |
Braakmann-Folgmann, Anne Christina |
title |
Mapping the extent of giant Antarctic icebergs with deep learning |
title_short |
Mapping the extent of giant Antarctic icebergs with deep learning |
title_full |
Mapping the extent of giant Antarctic icebergs with deep learning |
title_fullStr |
Mapping the extent of giant Antarctic icebergs with deep learning |
title_full_unstemmed |
Mapping the extent of giant Antarctic icebergs with deep learning |
title_sort |
mapping the extent of giant antarctic icebergs with deep learning |
publisher |
Copernicus Publications |
publishDate |
2023 |
url |
https://hdl.handle.net/10037/32655 https://doi.org/10.5194/tc-17-4675-2023 |
geographic |
Antarctic |
geographic_facet |
Antarctic |
genre |
Antarc* Antarctic Iceberg* Sea ice The Cryosphere |
genre_facet |
Antarc* Antarctic Iceberg* Sea ice The Cryosphere |
op_relation |
The Cryosphere Braakmann-Folgmann, Shepherd, Hogg, Redmond. Mapping the extent of giant Antarctic icebergs with deep learning. The Cryosphere. 2023;17(11):4675-4690 FRIDAID 2215855 doi:10.5194/tc-17-4675-2023 1994-0416 1994-0424 https://hdl.handle.net/10037/32655 |
op_rights |
Attribution 4.0 International (CC BY 4.0) openAccess Copyright 2023 The Author(s) https://creativecommons.org/licenses/by/4.0 |
op_doi |
https://doi.org/10.5194/tc-17-4675-2023 |
container_title |
The Cryosphere |
container_volume |
17 |
container_issue |
11 |
container_start_page |
4675 |
op_container_end_page |
4690 |
_version_ |
1790594477549682688 |