Data-driven surrogate modeling of high-resolution sea-ice thickness in the Arctic
A novel generation of sea-ice models with elasto-brittle rheologies, such as neXtSIM, can represent sea-ice processes with an unprecedented accuracy at the mesoscale for resolutions of around 10 km . As these models are computationally expensive, we introduce supervised deep learning techniques for...
Published in: | The Cryosphere |
---|---|
Main Authors: | , , , , , |
Format: | Text |
Language: | English |
Published: |
2024
|
Subjects: | |
Online Access: | https://doi.org/10.5194/tc-18-1791-2024 https://tc.copernicus.org/articles/18/1791/2024/ |
id |
ftcopernicus:oai:publications.copernicus.org:tc112651 |
---|---|
record_format |
openpolar |
spelling |
ftcopernicus:oai:publications.copernicus.org:tc112651 2024-09-15T18:34:12+00:00 Data-driven surrogate modeling of high-resolution sea-ice thickness in the Arctic Durand, Charlotte Finn, Tobias Sebastian Farchi, Alban Bocquet, Marc Boutin, Guillaume Ólason, Einar 2024-04-18 application/pdf https://doi.org/10.5194/tc-18-1791-2024 https://tc.copernicus.org/articles/18/1791/2024/ eng eng doi:10.5194/tc-18-1791-2024 https://tc.copernicus.org/articles/18/1791/2024/ eISSN: 1994-0424 Text 2024 ftcopernicus https://doi.org/10.5194/tc-18-1791-2024 2024-08-28T05:24:15Z A novel generation of sea-ice models with elasto-brittle rheologies, such as neXtSIM, can represent sea-ice processes with an unprecedented accuracy at the mesoscale for resolutions of around 10 km . As these models are computationally expensive, we introduce supervised deep learning techniques for surrogate modeling of the sea-ice thickness from neXtSIM simulations. We adapt a convolutional U-Net architecture to an Arctic-wide setup by taking the land–sea mask with partial convolutions into account. Trained to emulate the sea-ice thickness at a lead time of 12 h, the neural network can be iteratively applied to predictions for up to 1 year. The improvements of the surrogate model over a persistence forecast persist from 12 h to roughly 1 year, with improvements of up to 50 % in the forecast error. Moreover, the predictability gain for the sea-ice thickness measured against the daily climatology extends to over 6 months. By using atmospheric forcings as additional input, the surrogate model can represent advective and thermodynamical processes which influence the sea-ice thickness and the growth and melting therein. While iterating, the surrogate model experiences diffusive processes which result in a loss of fine-scale structures. However, this smoothing increases the coherence of large-scale features and thereby the stability of the model. Therefore, based on these results, we see huge potential for surrogate modeling of state-of-the-art sea-ice models with neural networks. Text Sea ice Copernicus Publications: E-Journals The Cryosphere 18 4 1791 1815 |
institution |
Open Polar |
collection |
Copernicus Publications: E-Journals |
op_collection_id |
ftcopernicus |
language |
English |
description |
A novel generation of sea-ice models with elasto-brittle rheologies, such as neXtSIM, can represent sea-ice processes with an unprecedented accuracy at the mesoscale for resolutions of around 10 km . As these models are computationally expensive, we introduce supervised deep learning techniques for surrogate modeling of the sea-ice thickness from neXtSIM simulations. We adapt a convolutional U-Net architecture to an Arctic-wide setup by taking the land–sea mask with partial convolutions into account. Trained to emulate the sea-ice thickness at a lead time of 12 h, the neural network can be iteratively applied to predictions for up to 1 year. The improvements of the surrogate model over a persistence forecast persist from 12 h to roughly 1 year, with improvements of up to 50 % in the forecast error. Moreover, the predictability gain for the sea-ice thickness measured against the daily climatology extends to over 6 months. By using atmospheric forcings as additional input, the surrogate model can represent advective and thermodynamical processes which influence the sea-ice thickness and the growth and melting therein. While iterating, the surrogate model experiences diffusive processes which result in a loss of fine-scale structures. However, this smoothing increases the coherence of large-scale features and thereby the stability of the model. Therefore, based on these results, we see huge potential for surrogate modeling of state-of-the-art sea-ice models with neural networks. |
format |
Text |
author |
Durand, Charlotte Finn, Tobias Sebastian Farchi, Alban Bocquet, Marc Boutin, Guillaume Ólason, Einar |
spellingShingle |
Durand, Charlotte Finn, Tobias Sebastian Farchi, Alban Bocquet, Marc Boutin, Guillaume Ólason, Einar Data-driven surrogate modeling of high-resolution sea-ice thickness in the Arctic |
author_facet |
Durand, Charlotte Finn, Tobias Sebastian Farchi, Alban Bocquet, Marc Boutin, Guillaume Ólason, Einar |
author_sort |
Durand, Charlotte |
title |
Data-driven surrogate modeling of high-resolution sea-ice thickness in the Arctic |
title_short |
Data-driven surrogate modeling of high-resolution sea-ice thickness in the Arctic |
title_full |
Data-driven surrogate modeling of high-resolution sea-ice thickness in the Arctic |
title_fullStr |
Data-driven surrogate modeling of high-resolution sea-ice thickness in the Arctic |
title_full_unstemmed |
Data-driven surrogate modeling of high-resolution sea-ice thickness in the Arctic |
title_sort |
data-driven surrogate modeling of high-resolution sea-ice thickness in the arctic |
publishDate |
2024 |
url |
https://doi.org/10.5194/tc-18-1791-2024 https://tc.copernicus.org/articles/18/1791/2024/ |
genre |
Sea ice |
genre_facet |
Sea ice |
op_source |
eISSN: 1994-0424 |
op_relation |
doi:10.5194/tc-18-1791-2024 https://tc.copernicus.org/articles/18/1791/2024/ |
op_doi |
https://doi.org/10.5194/tc-18-1791-2024 |
container_title |
The Cryosphere |
container_volume |
18 |
container_issue |
4 |
container_start_page |
1791 |
op_container_end_page |
1815 |
_version_ |
1810476000989937664 |