SICFormer: A 3D-Swin Transformer for Sea Ice Concentration Prediction
Sea ice concentration (SIC) is an important dimension for characterising the geographical features of the pan-Arctic region. Trends in SIC bring new opportunities for human activities in the Arctic region. In this paper, we propose a deep learning technology-based sea ice concentration prediction mo...
Published in: | Journal of Marine Science and Engineering |
---|---|
Main Authors: | , , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
MDPI AG
2024
|
Subjects: | |
Online Access: | https://doi.org/10.3390/jmse12081424 https://doaj.org/article/8335f7acb28e4b0087f63131e39ec9fe |
id |
ftdoajarticles:oai:doaj.org/article:8335f7acb28e4b0087f63131e39ec9fe |
---|---|
record_format |
openpolar |
spelling |
ftdoajarticles:oai:doaj.org/article:8335f7acb28e4b0087f63131e39ec9fe 2024-09-15T18:19:06+00:00 SICFormer: A 3D-Swin Transformer for Sea Ice Concentration Prediction Zhuoqing Jiang Bing Guo Huihui Zhao Yangming Jiang Yi Sun 2024-08-01T00:00:00Z https://doi.org/10.3390/jmse12081424 https://doaj.org/article/8335f7acb28e4b0087f63131e39ec9fe EN eng MDPI AG https://www.mdpi.com/2077-1312/12/8/1424 https://doaj.org/toc/2077-1312 doi:10.3390/jmse12081424 2077-1312 https://doaj.org/article/8335f7acb28e4b0087f63131e39ec9fe Journal of Marine Science and Engineering, Vol 12, Iss 8, p 1424 (2024) spatiotemporal prediction 3D-Swin Transformer sea ice concentration attention mechanisms Naval architecture. Shipbuilding. Marine engineering VM1-989 Oceanography GC1-1581 article 2024 ftdoajarticles https://doi.org/10.3390/jmse12081424 2024-09-02T15:34:38Z Sea ice concentration (SIC) is an important dimension for characterising the geographical features of the pan-Arctic region. Trends in SIC bring new opportunities for human activities in the Arctic region. In this paper, we propose a deep learning technology-based sea ice concentration prediction model, SICFormer, which can realise end-to-end daily sea ice concentration prediction. Specifically, the model uses a 3D-Swin Transformer as an encoder and designs a decoder to reconstruct the predicted image based on PixelShuffle. This is a new model architecture that we have proposed. Single-day SIC data from the National Snow and Ice Data Center (NSIDC) for the years 2006 to 2022 are utilised. The results of 8-day short-term prediction experiments show that the average Mean Absolute Error (MAE) of the SICFormer model on the test set over the 5 years is 1.89%, the Root Mean Squared Error (RMSE) is 5.99%, the Mean Absolute Percentage Error (MAPE) is 4.32%, and the Nash–Sutcliffe Efficiency (NSE) is 0.98. Furthermore, the current popular deep learning models for spatio-temporal prediction are employed as a point of comparison given their proven efficacy on numerous public datasets. The comparison experiments show that the SICFormer model achieves the best overall performance. Article in Journal/Newspaper National Snow and Ice Data Center Sea ice Directory of Open Access Journals: DOAJ Articles Journal of Marine Science and Engineering 12 8 1424 |
institution |
Open Polar |
collection |
Directory of Open Access Journals: DOAJ Articles |
op_collection_id |
ftdoajarticles |
language |
English |
topic |
spatiotemporal prediction 3D-Swin Transformer sea ice concentration attention mechanisms Naval architecture. Shipbuilding. Marine engineering VM1-989 Oceanography GC1-1581 |
spellingShingle |
spatiotemporal prediction 3D-Swin Transformer sea ice concentration attention mechanisms Naval architecture. Shipbuilding. Marine engineering VM1-989 Oceanography GC1-1581 Zhuoqing Jiang Bing Guo Huihui Zhao Yangming Jiang Yi Sun SICFormer: A 3D-Swin Transformer for Sea Ice Concentration Prediction |
topic_facet |
spatiotemporal prediction 3D-Swin Transformer sea ice concentration attention mechanisms Naval architecture. Shipbuilding. Marine engineering VM1-989 Oceanography GC1-1581 |
description |
Sea ice concentration (SIC) is an important dimension for characterising the geographical features of the pan-Arctic region. Trends in SIC bring new opportunities for human activities in the Arctic region. In this paper, we propose a deep learning technology-based sea ice concentration prediction model, SICFormer, which can realise end-to-end daily sea ice concentration prediction. Specifically, the model uses a 3D-Swin Transformer as an encoder and designs a decoder to reconstruct the predicted image based on PixelShuffle. This is a new model architecture that we have proposed. Single-day SIC data from the National Snow and Ice Data Center (NSIDC) for the years 2006 to 2022 are utilised. The results of 8-day short-term prediction experiments show that the average Mean Absolute Error (MAE) of the SICFormer model on the test set over the 5 years is 1.89%, the Root Mean Squared Error (RMSE) is 5.99%, the Mean Absolute Percentage Error (MAPE) is 4.32%, and the Nash–Sutcliffe Efficiency (NSE) is 0.98. Furthermore, the current popular deep learning models for spatio-temporal prediction are employed as a point of comparison given their proven efficacy on numerous public datasets. The comparison experiments show that the SICFormer model achieves the best overall performance. |
format |
Article in Journal/Newspaper |
author |
Zhuoqing Jiang Bing Guo Huihui Zhao Yangming Jiang Yi Sun |
author_facet |
Zhuoqing Jiang Bing Guo Huihui Zhao Yangming Jiang Yi Sun |
author_sort |
Zhuoqing Jiang |
title |
SICFormer: A 3D-Swin Transformer for Sea Ice Concentration Prediction |
title_short |
SICFormer: A 3D-Swin Transformer for Sea Ice Concentration Prediction |
title_full |
SICFormer: A 3D-Swin Transformer for Sea Ice Concentration Prediction |
title_fullStr |
SICFormer: A 3D-Swin Transformer for Sea Ice Concentration Prediction |
title_full_unstemmed |
SICFormer: A 3D-Swin Transformer for Sea Ice Concentration Prediction |
title_sort |
sicformer: a 3d-swin transformer for sea ice concentration prediction |
publisher |
MDPI AG |
publishDate |
2024 |
url |
https://doi.org/10.3390/jmse12081424 https://doaj.org/article/8335f7acb28e4b0087f63131e39ec9fe |
genre |
National Snow and Ice Data Center Sea ice |
genre_facet |
National Snow and Ice Data Center Sea ice |
op_source |
Journal of Marine Science and Engineering, Vol 12, Iss 8, p 1424 (2024) |
op_relation |
https://www.mdpi.com/2077-1312/12/8/1424 https://doaj.org/toc/2077-1312 doi:10.3390/jmse12081424 2077-1312 https://doaj.org/article/8335f7acb28e4b0087f63131e39ec9fe |
op_doi |
https://doi.org/10.3390/jmse12081424 |
container_title |
Journal of Marine Science and Engineering |
container_volume |
12 |
container_issue |
8 |
container_start_page |
1424 |
_version_ |
1810457194191126528 |