IceTFT v1.0.0: interpretable long-term prediction of  Arctic sea ice extent with deep learning

Mu, Bin; Luo, Xiaodan; Yuan, Shijin; Liang, Xi

doi:https://doi.org/10.5194/gmd-16-4677-2023

Articles | Volume 16, issue 16

https://doi.org/10.5194/gmd-16-4677-2023

© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/gmd-16-4677-2023

© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 16, issue 16

Model description paper

|

18 Aug 2023

Model description paper |

| 18 Aug 2023

IceTFT v1.0.0: interpretable long-term prediction of Arctic sea ice extent with deep learning

Bin Mu, Xiaodan Luo, Shijin Yuan, and Xi Liang

Data sets

Sea Ice Index, Version 3 Fetterer, F., K. Knowles, W. N. Meier, M. Savoie, and A. K. Windnagel. https://doi.org/10.7265/N5K072F8

Boulder Monthly Means: Snowfall National Oceanic and Atmospheric Administration Physical Sciences Laboratory, Boulder Climate and Weather Information https://doi.org/10.5281/zenodo.7533097

JRA-55: Japanese 55-year Reanalysis, Monthly Means and Variances Japan Meteorological Agency https://doi.org/10.5065/D60G3H5B

Model code and software

IceTFT: 1.0.0 Xiaodan Luo https://doi.org/10.5281/zenodo.7409157

Short summary

To improve the long-term forecast skill for sea ice extent (SIE), we introduce IceTFT, which directly predicts 12 months of averaged Arctic SIE. The results show that IceTFT has higher forecasting skill. We conducted a sensitivity analysis of the variables in the IceTFT model. These sensitivities can help researchers study the mechanisms of sea ice development, and they also provide useful references for the selection of variables in data assimilation or the input of deep learning models.