AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting
Abstract For maritime navigation in the Arctic, sea ice charts are an essential tool, which still to this day is drawn manually by professional ice analysts. The total Sea Ice Concentration (SIC) is the primary descriptor of the charts and indicates the fraction of ice in an ocean surface area. Natu...
Published in: | Scientific Reports |
---|---|
Main Authors: | , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Nature Portfolio
2023
|
Subjects: | |
Online Access: | https://doi.org/10.1038/s41598-023-32467-x https://doaj.org/article/4eb9a3f6b76442fc8804037efbe07f02 |
id |
ftdoajarticles:oai:doaj.org/article:4eb9a3f6b76442fc8804037efbe07f02 |
---|---|
record_format |
openpolar |
spelling |
ftdoajarticles:oai:doaj.org/article:4eb9a3f6b76442fc8804037efbe07f02 2023-08-20T04:04:36+02:00 AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting Andrzej Kucik Andreas Stokholm 2023-04-01T00:00:00Z https://doi.org/10.1038/s41598-023-32467-x https://doaj.org/article/4eb9a3f6b76442fc8804037efbe07f02 EN eng Nature Portfolio https://doi.org/10.1038/s41598-023-32467-x https://doaj.org/toc/2045-2322 doi:10.1038/s41598-023-32467-x 2045-2322 https://doaj.org/article/4eb9a3f6b76442fc8804037efbe07f02 Scientific Reports, Vol 13, Iss 1, Pp 1-10 (2023) Medicine R Science Q article 2023 ftdoajarticles https://doi.org/10.1038/s41598-023-32467-x 2023-07-30T00:40:02Z Abstract For maritime navigation in the Arctic, sea ice charts are an essential tool, which still to this day is drawn manually by professional ice analysts. The total Sea Ice Concentration (SIC) is the primary descriptor of the charts and indicates the fraction of ice in an ocean surface area. Naturally, automating the SIC chart creation is desired. However, the optimal representation of the corresponding machine-learning task is ambivalent and discussed in the community. In this study, we explore the representation with either regressional or classification objectives, each with two different (weighted) loss functions: Mean Square Error and Binary Cross-Entropy, and Categorical Cross-Entropy and the Earth Mover’s Distance, respectively. While all models achieve good results they differ as the regression-based models obtain the highest numerical similarity to the reference charts, whereas the classification-optimised models generate results more visually pleasing and consistent. Rescaling the loss functions with inverse class weights improves the performance for intermediate classes at the expense of open water and fully-covered sea ice areas. Article in Journal/Newspaper Arctic Sea ice Directory of Open Access Journals: DOAJ Articles Arctic Scientific Reports 13 1 |
institution |
Open Polar |
collection |
Directory of Open Access Journals: DOAJ Articles |
op_collection_id |
ftdoajarticles |
language |
English |
topic |
Medicine R Science Q |
spellingShingle |
Medicine R Science Q Andrzej Kucik Andreas Stokholm AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting |
topic_facet |
Medicine R Science Q |
description |
Abstract For maritime navigation in the Arctic, sea ice charts are an essential tool, which still to this day is drawn manually by professional ice analysts. The total Sea Ice Concentration (SIC) is the primary descriptor of the charts and indicates the fraction of ice in an ocean surface area. Naturally, automating the SIC chart creation is desired. However, the optimal representation of the corresponding machine-learning task is ambivalent and discussed in the community. In this study, we explore the representation with either regressional or classification objectives, each with two different (weighted) loss functions: Mean Square Error and Binary Cross-Entropy, and Categorical Cross-Entropy and the Earth Mover’s Distance, respectively. While all models achieve good results they differ as the regression-based models obtain the highest numerical similarity to the reference charts, whereas the classification-optimised models generate results more visually pleasing and consistent. Rescaling the loss functions with inverse class weights improves the performance for intermediate classes at the expense of open water and fully-covered sea ice areas. |
format |
Article in Journal/Newspaper |
author |
Andrzej Kucik Andreas Stokholm |
author_facet |
Andrzej Kucik Andreas Stokholm |
author_sort |
Andrzej Kucik |
title |
AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting |
title_short |
AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting |
title_full |
AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting |
title_fullStr |
AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting |
title_full_unstemmed |
AI4SeaIce: selecting loss functions for automated SAR sea ice concentration charting |
title_sort |
ai4seaice: selecting loss functions for automated sar sea ice concentration charting |
publisher |
Nature Portfolio |
publishDate |
2023 |
url |
https://doi.org/10.1038/s41598-023-32467-x https://doaj.org/article/4eb9a3f6b76442fc8804037efbe07f02 |
geographic |
Arctic |
geographic_facet |
Arctic |
genre |
Arctic Sea ice |
genre_facet |
Arctic Sea ice |
op_source |
Scientific Reports, Vol 13, Iss 1, Pp 1-10 (2023) |
op_relation |
https://doi.org/10.1038/s41598-023-32467-x https://doaj.org/toc/2045-2322 doi:10.1038/s41598-023-32467-x 2045-2322 https://doaj.org/article/4eb9a3f6b76442fc8804037efbe07f02 |
op_doi |
https://doi.org/10.1038/s41598-023-32467-x |
container_title |
Scientific Reports |
container_volume |
13 |
container_issue |
1 |
_version_ |
1774714977395933184 |