Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks

A new algorithm for classification of sea ice types on Sentinel-1 Synthetic Aperture Radar (SAR) data using a convolutional neural network (CNN) is presented. The CNN is trained on reference ice charts produced by human experts and compared with an existing machine learning algorithm based on textur...

Full description

Bibliographic Details
Published in:Remote Sensing
Main Authors: Hugo Boulze, Anton Korosov, Julien Brajard
Format: Article in Journal/Newspaper
Language:English
Published: MDPI AG 2020
Subjects:
convolutional neural network
Sentinel-1
SAR
sea ice type
ice chart
Arctic
Science
Q
Sea ice
Online Access:https://doi.org/10.3390/rs12132165
https://doaj.org/article/881838aee22b4b5cafb7adde3a74dabf
id ftdoajarticles:oai:doaj.org/article:881838aee22b4b5cafb7adde3a74dabf
record_format openpolar
spelling ftdoajarticles:oai:doaj.org/article:881838aee22b4b5cafb7adde3a74dabf 2023-05-15T15:05:45+02:00 Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks Hugo Boulze Anton Korosov Julien Brajard 2020-07-01T00:00:00Z https://doi.org/10.3390/rs12132165 https://doaj.org/article/881838aee22b4b5cafb7adde3a74dabf EN eng MDPI AG https://www.mdpi.com/2072-4292/12/13/2165 https://doaj.org/toc/2072-4292 doi:10.3390/rs12132165 2072-4292 https://doaj.org/article/881838aee22b4b5cafb7adde3a74dabf Remote Sensing, Vol 12, Iss 2165, p 2165 (2020) convolutional neural network Sentinel-1 SAR sea ice type ice chart Arctic Science Q article 2020 ftdoajarticles https://doi.org/10.3390/rs12132165 2022-12-31T04:00:15Z A new algorithm for classification of sea ice types on Sentinel-1 Synthetic Aperture Radar (SAR) data using a convolutional neural network (CNN) is presented. The CNN is trained on reference ice charts produced by human experts and compared with an existing machine learning algorithm based on texture features and random forest classifier. The CNN is trained on two datasets in 2018 and 2020 for retrieval of four classes: ice free, young ice, first-year ice and old ice. The accuracy of our classification is 90.5% for the 2018-dataset and 91.6% for the 2020-dataset. The uncertainty is a bit higher for young ice (85%/76% accuracy in 2018/2020) and first-year ice (86%/84% accuracy in 2018/2020). Our algorithm outperforms the existing random forest product for each ice type. It has also proved to be more efficient in computing time and less sensitive to the noise in SAR data. The code is publicly available. Article in Journal/Newspaper Arctic Sea ice Directory of Open Access Journals: DOAJ Articles Arctic Remote Sensing 12 13 2165
institution Open Polar
collection Directory of Open Access Journals: DOAJ Articles
op_collection_id ftdoajarticles
language English
topic convolutional neural network
Sentinel-1
SAR
sea ice type
ice chart
Arctic
Science
Q
spellingShingle convolutional neural network
Sentinel-1
SAR
sea ice type
ice chart
Arctic
Science
Q
Hugo Boulze
Anton Korosov
Julien Brajard
Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks
topic_facet convolutional neural network
Sentinel-1
SAR
sea ice type
ice chart
Arctic
Science
Q
description A new algorithm for classification of sea ice types on Sentinel-1 Synthetic Aperture Radar (SAR) data using a convolutional neural network (CNN) is presented. The CNN is trained on reference ice charts produced by human experts and compared with an existing machine learning algorithm based on texture features and random forest classifier. The CNN is trained on two datasets in 2018 and 2020 for retrieval of four classes: ice free, young ice, first-year ice and old ice. The accuracy of our classification is 90.5% for the 2018-dataset and 91.6% for the 2020-dataset. The uncertainty is a bit higher for young ice (85%/76% accuracy in 2018/2020) and first-year ice (86%/84% accuracy in 2018/2020). Our algorithm outperforms the existing random forest product for each ice type. It has also proved to be more efficient in computing time and less sensitive to the noise in SAR data. The code is publicly available.
format Article in Journal/Newspaper
author Hugo Boulze
Anton Korosov
Julien Brajard
author_facet Hugo Boulze
Anton Korosov
Julien Brajard
author_sort Hugo Boulze
title Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks
title_short Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks
title_full Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks
title_fullStr Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks
title_full_unstemmed Classification of Sea Ice Types in Sentinel-1 SAR Data Using Convolutional Neural Networks
title_sort classification of sea ice types in sentinel-1 sar data using convolutional neural networks
publisher MDPI AG
publishDate 2020
url https://doi.org/10.3390/rs12132165
https://doaj.org/article/881838aee22b4b5cafb7adde3a74dabf
geographic Arctic
geographic_facet Arctic
genre Arctic
Sea ice
genre_facet Arctic
Sea ice
op_source Remote Sensing, Vol 12, Iss 2165, p 2165 (2020)
op_relation https://www.mdpi.com/2072-4292/12/13/2165
https://doaj.org/toc/2072-4292
doi:10.3390/rs12132165
2072-4292
https://doaj.org/article/881838aee22b4b5cafb7adde3a74dabf
op_doi https://doi.org/10.3390/rs12132165
container_title Remote Sensing
container_volume 12
container_issue 13
container_start_page 2165
_version_ 1766337400952324096

Similar Items