Deep multi-scale learning for automatic tracking of internal layers of ice in radar data
Abstract In this study, our goal is to track internal ice layers on the Snow Radar data collected by NASA Operation IceBridge. We examine the application of deep learning methods on radar data gathered from polar regions. Artificial intelligence techniques have displayed impressive success in many p...
Published in: | Journal of Glaciology |
---|---|
Main Authors: | , , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Cambridge University Press (CUP)
2020
|
Subjects: | |
Online Access: | http://dx.doi.org/10.1017/jog.2020.80 https://www.cambridge.org/core/services/aop-cambridge-core/content/view/S0022143020000805 |
id |
crcambridgeupr:10.1017/jog.2020.80 |
---|---|
record_format |
openpolar |
spelling |
crcambridgeupr:10.1017/jog.2020.80 2024-04-07T07:53:42+00:00 Deep multi-scale learning for automatic tracking of internal layers of ice in radar data Rahnemoonfar, Maryam Yari, Masoud Paden, John Koenig, Lora Ibikunle, Oluwanisola 2020 http://dx.doi.org/10.1017/jog.2020.80 https://www.cambridge.org/core/services/aop-cambridge-core/content/view/S0022143020000805 en eng Cambridge University Press (CUP) http://creativecommons.org/licenses/by/4.0/ Journal of Glaciology volume 67, issue 261, page 39-48 ISSN 0022-1430 1727-5652 Earth-Surface Processes journal-article 2020 crcambridgeupr https://doi.org/10.1017/jog.2020.80 2024-03-08T00:34:13Z Abstract In this study, our goal is to track internal ice layers on the Snow Radar data collected by NASA Operation IceBridge. We examine the application of deep learning methods on radar data gathered from polar regions. Artificial intelligence techniques have displayed impressive success in many practical fields. Deep neural networks owe their success to the availability of massive labeled data. However, in many real-world problems, even when a large dataset is available, deep learning methods have shown less success, due to causes such as lack of a large labeled dataset, presence of noise in the data or missing data. In our radar data, the presence of noise is one of the main obstacles in utilizing popular deep learning methods such as transfer learning. Our experiments show that if the neural network is trained to detect contours of objects in electro-optical imagery, it can only track a low percentage of contours in radar data. Fine-tuning and further training do not provide any better results. However, we show that selecting the right model and training it on the radar imagery from the start yields far better results. Article in Journal/Newspaper Journal of Glaciology Cambridge University Press Journal of Glaciology 67 261 39 48 |
institution |
Open Polar |
collection |
Cambridge University Press |
op_collection_id |
crcambridgeupr |
language |
English |
topic |
Earth-Surface Processes |
spellingShingle |
Earth-Surface Processes Rahnemoonfar, Maryam Yari, Masoud Paden, John Koenig, Lora Ibikunle, Oluwanisola Deep multi-scale learning for automatic tracking of internal layers of ice in radar data |
topic_facet |
Earth-Surface Processes |
description |
Abstract In this study, our goal is to track internal ice layers on the Snow Radar data collected by NASA Operation IceBridge. We examine the application of deep learning methods on radar data gathered from polar regions. Artificial intelligence techniques have displayed impressive success in many practical fields. Deep neural networks owe their success to the availability of massive labeled data. However, in many real-world problems, even when a large dataset is available, deep learning methods have shown less success, due to causes such as lack of a large labeled dataset, presence of noise in the data or missing data. In our radar data, the presence of noise is one of the main obstacles in utilizing popular deep learning methods such as transfer learning. Our experiments show that if the neural network is trained to detect contours of objects in electro-optical imagery, it can only track a low percentage of contours in radar data. Fine-tuning and further training do not provide any better results. However, we show that selecting the right model and training it on the radar imagery from the start yields far better results. |
format |
Article in Journal/Newspaper |
author |
Rahnemoonfar, Maryam Yari, Masoud Paden, John Koenig, Lora Ibikunle, Oluwanisola |
author_facet |
Rahnemoonfar, Maryam Yari, Masoud Paden, John Koenig, Lora Ibikunle, Oluwanisola |
author_sort |
Rahnemoonfar, Maryam |
title |
Deep multi-scale learning for automatic tracking of internal layers of ice in radar data |
title_short |
Deep multi-scale learning for automatic tracking of internal layers of ice in radar data |
title_full |
Deep multi-scale learning for automatic tracking of internal layers of ice in radar data |
title_fullStr |
Deep multi-scale learning for automatic tracking of internal layers of ice in radar data |
title_full_unstemmed |
Deep multi-scale learning for automatic tracking of internal layers of ice in radar data |
title_sort |
deep multi-scale learning for automatic tracking of internal layers of ice in radar data |
publisher |
Cambridge University Press (CUP) |
publishDate |
2020 |
url |
http://dx.doi.org/10.1017/jog.2020.80 https://www.cambridge.org/core/services/aop-cambridge-core/content/view/S0022143020000805 |
genre |
Journal of Glaciology |
genre_facet |
Journal of Glaciology |
op_source |
Journal of Glaciology volume 67, issue 261, page 39-48 ISSN 0022-1430 1727-5652 |
op_rights |
http://creativecommons.org/licenses/by/4.0/ |
op_doi |
https://doi.org/10.1017/jog.2020.80 |
container_title |
Journal of Glaciology |
container_volume |
67 |
container_issue |
261 |
container_start_page |
39 |
op_container_end_page |
48 |
_version_ |
1795669797559599104 |