Multi-Task Spatiotemporal Deep Learning Based Arctic Sea Ice Prediction ...
Important natural resources in the Arctic rely heavily on sea ice, making it important to forecast Arctic sea ice changes. Arctic sea ice forecasting often involves two connected tasks: sea ice concentration at each pixel and overall sea ice extent. Instead of having two separate models for two fore...
Main Authors: | , , , , , , , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | unknown |
Published: |
UMBC HPCF
2021
|
Subjects: | |
Online Access: | https://dx.doi.org/10.13016/m2tk7h-gil8 https://mdsoar.org/handle/11603/23253 |
id |
ftdatacite:10.13016/m2tk7h-gil8 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.13016/m2tk7h-gil8 2023-08-27T04:07:12+02:00 Multi-Task Spatiotemporal Deep Learning Based Arctic Sea Ice Prediction ... Bourne Jr., Jamal Hu, Michael Kim, Eliot Kruse, Peter Lama, Skylar Ali, Sahara Huang, Yiyi Wang, Jianwu 2021 https://dx.doi.org/10.13016/m2tk7h-gil8 https://mdsoar.org/handle/11603/23253 unknown UMBC HPCF This item is likely protected under Title 17 of the U.S. Copyright Law. Unless on a Creative Commons license, for uses protected by Copyright Law, contact the copyright holder or the author. UMBC High Performance Computing Facilty HPCF CreativeWork article 2021 ftdatacite https://doi.org/10.13016/m2tk7h-gil8 2023-08-07T14:24:23Z Important natural resources in the Arctic rely heavily on sea ice, making it important to forecast Arctic sea ice changes. Arctic sea ice forecasting often involves two connected tasks: sea ice concentration at each pixel and overall sea ice extent. Instead of having two separate models for two forecasting tasks, in this report, we study how to use multi-task learning techniques and leverage the connections between ice concentration and ice extent to improve accuracy for both prediction tasks. Because of the spatiotemporal nature of the data, we designed two novel multi-task learning models based on CNNs and ConvLSTMs, respectively. We also developed a custom loss function which trains the models to ignore land pixels when making predictions. Our experiments show our models can have better accuracies than separate models that predict sea ice extent and concentration separately, and that our accuracies are better than or comparable with results in the state-of-the-art studies. ... Article in Journal/Newspaper Arctic Sea ice DataCite Metadata Store (German National Library of Science and Technology) Arctic |
institution |
Open Polar |
collection |
DataCite Metadata Store (German National Library of Science and Technology) |
op_collection_id |
ftdatacite |
language |
unknown |
topic |
UMBC High Performance Computing Facilty HPCF |
spellingShingle |
UMBC High Performance Computing Facilty HPCF Bourne Jr., Jamal Hu, Michael Kim, Eliot Kruse, Peter Lama, Skylar Ali, Sahara Huang, Yiyi Wang, Jianwu Multi-Task Spatiotemporal Deep Learning Based Arctic Sea Ice Prediction ... |
topic_facet |
UMBC High Performance Computing Facilty HPCF |
description |
Important natural resources in the Arctic rely heavily on sea ice, making it important to forecast Arctic sea ice changes. Arctic sea ice forecasting often involves two connected tasks: sea ice concentration at each pixel and overall sea ice extent. Instead of having two separate models for two forecasting tasks, in this report, we study how to use multi-task learning techniques and leverage the connections between ice concentration and ice extent to improve accuracy for both prediction tasks. Because of the spatiotemporal nature of the data, we designed two novel multi-task learning models based on CNNs and ConvLSTMs, respectively. We also developed a custom loss function which trains the models to ignore land pixels when making predictions. Our experiments show our models can have better accuracies than separate models that predict sea ice extent and concentration separately, and that our accuracies are better than or comparable with results in the state-of-the-art studies. ... |
format |
Article in Journal/Newspaper |
author |
Bourne Jr., Jamal Hu, Michael Kim, Eliot Kruse, Peter Lama, Skylar Ali, Sahara Huang, Yiyi Wang, Jianwu |
author_facet |
Bourne Jr., Jamal Hu, Michael Kim, Eliot Kruse, Peter Lama, Skylar Ali, Sahara Huang, Yiyi Wang, Jianwu |
author_sort |
Bourne Jr., Jamal |
title |
Multi-Task Spatiotemporal Deep Learning Based Arctic Sea Ice Prediction ... |
title_short |
Multi-Task Spatiotemporal Deep Learning Based Arctic Sea Ice Prediction ... |
title_full |
Multi-Task Spatiotemporal Deep Learning Based Arctic Sea Ice Prediction ... |
title_fullStr |
Multi-Task Spatiotemporal Deep Learning Based Arctic Sea Ice Prediction ... |
title_full_unstemmed |
Multi-Task Spatiotemporal Deep Learning Based Arctic Sea Ice Prediction ... |
title_sort |
multi-task spatiotemporal deep learning based arctic sea ice prediction ... |
publisher |
UMBC HPCF |
publishDate |
2021 |
url |
https://dx.doi.org/10.13016/m2tk7h-gil8 https://mdsoar.org/handle/11603/23253 |
geographic |
Arctic |
geographic_facet |
Arctic |
genre |
Arctic Sea ice |
genre_facet |
Arctic Sea ice |
op_rights |
This item is likely protected under Title 17 of the U.S. Copyright Law. Unless on a Creative Commons license, for uses protected by Copyright Law, contact the copyright holder or the author. |
op_doi |
https://doi.org/10.13016/m2tk7h-gil8 |
_version_ |
1775347987802750976 |