the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Similar-Year Feature Construction Improves Deep-Learning Arctic Sea Ice Forecasts
Abstract. Seasonal Arctic sea ice prediction is important for understanding Arctic climate change and its impacts. Existing deep-learning models commonly use the multi-year mean climatology of the forecast period as a historical-evolution background. This provides stable seasonal information, but has limited ability to represent interannual differences in sea ice evolution. We propose a dynamic similar-year feature construction method based on sea ice spatial similarity and incorporate it into the SwinUNet forecasting framework. Using sea ice concentration (SIC) at forecast initialization and the Intersection over Union (IoU) of sea ice extent (SIE), the method identifies the most similar historical year since 1979 and extracts its SIC evolution over the forecast period as the historical-evolution feature. Continuous SIC and sea ice thickness (SIT) from the preceding four months are also included, combining recent conditions with historical same-period evolution. Using NSIDC SIC and PIOMAS SIT data from 1979 to 2024, four-month Arctic sea ice forecasts are conducted at pentad intervals. The Similarity scheme is compared with Continuous and Climatology features in SwinUNet, as well as Persistence and ECMWF SEAS5. Similarity generally outperforms Continuous and Climatology. Over the first 12 forecast periods, it reduces MAE by 5.41% and 6.97%, improves ACC by 1.63% and 4.03%, and increases BACC by 0.09% and 0.42%, respectively. Relative to Persistence and SEAS5, the annual-mean MAE decreases by 22.19% and 42.24%, while BACC improves by 4.67% and 8.43%. These results show that dynamic similar-year features can enhance SwinUNet skill without increasing model complexity.
- Preprint
(3117 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 21 Dec 2026)
-
CEC1: 'Comment on egusphere-2026-6462 - No compliance with the policy of the journal', Juan Antonio Añel, 02 Nov 2026
reply
-
AC1: 'Reply on CEC1', Mingfeng Wang, 05 Nov 2026
reply
Dear Juan A. Añel,
Thank you for the assessment. We have published a new version of the archive with the code, the software versions, the model input, the trained weights, and the test predictions:
https://doi.org/10.5281/zenodo.23940748
1. Software versions. README.md and requirements.txt list the versions used to run the experiments: Python 3.9.23, PyTorch 2.5.1+cu121, CUDA 12.1, cuDNN 9.1.0.70, NumPy 1.26.3, Matplotlib 3.9.4, and tqdm 4.67.1. The Swin-UNet is implemented in Swin-Unet-2m-similyear.py. The daily-merge script also uses xarray with the netCDF4 backend from the same environment.
2. Code. The archive contains the preprocessing scripts (build_sic_era5_piomas_nsidc.py and build_hou_data_fusion_nsidc.py), the Similarity training program (Swin-Unet-2m-similyear.py), and the evaluation script (plot_comparision3_2month.py).
3. Input data. fusion_dataset_nsidc_pentad.rar (520 MB) is the pentad dataset read by the model, from January 1979 through December 2025. It contains sea-ice concentration, PIOMAS thickness, the historical maximum ice mask, the monthly masks, and the pentad climatology. These processed fields are the inputs used in the experiments and are archived under this DOI.
4. Weights and model output. swinunet_best.pth (665 MB) is the trained Similarity checkpoint; the weights are stored in model_state_dict. test_predictions_with_dates_compact.npz (368 MB) contains the test-set predictions and targets used for the reported metrics, together with the dates and lead offsets. SIC in that file is stored as 254-level uint8 so that the archive stays within the repository size limit. README.md gives the keys and the code that restores the fields to float32 sea-ice concentration.Â
The Code and data availability section and the bibliography will be updated to this version DOI in the revised manuscript.
Kind regards,
Mingfeng WangCitation: https://doi.org/10.5194/egusphere-2026-6462-AC1 -
CEC2: 'Reply on AC1', Juan Antonio Añel, 05 Nov 2026
reply
Dear authors,
Thanks for addressing this issue so quickly. I have checked the repositories and we can consider now the current version of your manuscript in compliance with the code policy of the journal.
Juan A. Añel
Geosci. Model Dev. Executive Editor
Citation: https://doi.org/10.5194/egusphere-2026-6462-CEC2
-
CEC2: 'Reply on AC1', Juan Antonio Añel, 05 Nov 2026
reply
-
AC1: 'Reply on CEC1', Mingfeng Wang, 05 Nov 2026
reply
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 187 | 82 | 31 | 300 | 33 | 28 |
- HTML: 187
- PDF: 82
- XML: 31
- Total: 300
- BibTeX: 33
- EndNote: 28
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Dear authors,
Unfortunately, after checking your manuscript, it has come to our attention that it does not comply with our "Code and Data Policy".
https://www.geoscientific-model-development.net/policies/code_and_data_policy.html
First, the code of your model actually is composed of dependencies on a series of libraries and functions implemented on them. However, it is not possible to know that kind of algorithms and implementations of them are applied because you have not identified the libraries used, and do not provide their code in a repository. Therefore, at least you must clarify the versions of the software that is necessary to perform the experiments.Â
In addition, to access the data used in your work you cite sites that are not acceptable according to the policy of the journal. They do not fulfil GMD’s requirements for a persistent data archive because:
- They do not appear to have a published policy for data preservation over many years or decades (some flexibility exists over the precise length of preservation, but the policy must exist).
- They do not appear to have a published mechanism for preventing authors from unilaterally removing material. Archives must have a policy which makes removal of materials only possible in exceptional circumstances and subject to an independent curatorial decision,
- They do not appear to issue a persistent identifier such as a DOI or Handle for each precise dataset.
If we have missed a published policy which does in fact address this matter satisfactorily, please post a response linking to it. If you have any questions about this issue, please post them in a reply.
Also, you do not provide a repository containing the model output data generated in your work, neither the model weights, which we request in machine learning based works. Please, provide a repository containing them.
The GMD review and publication process depends on reviewers and community commentators being able to access, during the discussion phase, the code and data on which a manuscript depends, and on ensuring the provenance of replicability of the published papers for years after their publication. Please, therefore, publish your code and data in one of the appropriate repositories and reply to this comment with the relevant information (link and a permanent identifier for it (e.g. DOI)) as soon as possible. We cannot have manuscripts under discussion that do not comply with our policy. Â
Later, if the Topical Editor decides to continue with the review or publication process of your manuscript and you are requested to upload a new version of it, then The 'Code and Data Availability’ section of your manuscript must also be modified to cite the new repository locations, and corresponding references added to the bibliography.Â
I must note that if you do not fix this problem, we cannot continue with the peer-review process or accept your manuscript for publication in GMD.
Juan A. Añel
Geosci. Model Dev. Executive Editor