Predicting Calving Events in Antarctica using Machine Learning

Jacob Alexander Hay, Hamzeh Issa, Daniele Fantin, David Parkes, Jan Wuite, Amber A Leeson, Malcolm McMillan
Proceedings of the 7th Northern Lights Deep Learning Conference (NLDL), PMLR 307:131-143, 2026.

Abstract

Monitoring the calving dynamics of the Antarctic ice shelves is central to understanding a major driver for the changes to ocean levels on our planet. Several physical models have been proposed as calving laws, with varying predictive power. We propose an approach using Machine Learning (ML) to identify key variables and parameters that may be used in future models of the ice shelf calving dynamics. As part of an ongoing project, we have trained a U-Net on samples from a set of Gaussian Random Field-represented Essential Climate Variables (ECV). Ablation studies establish a few of the selected variables as having high correlation with calving events, with an F1 score above 0.9. Our first study site was the Larsen C Ice Shelf, on the northwest part of the Weddell Sea, where in 2017 there was a massive calving event. We have found strong correlations between the calving and the ice velocity leading up to this event, which may be further improved when accounting for basal melt rates in the area.

Cite this Paper

BibTeX
@InProceedings{pmlr-v307-hay26a, title = {Predicting Calving Events in Antarctica using Machine Learning}, author = {Hay, Jacob Alexander and Issa, Hamzeh and Fantin, Daniele and Parkes, David and Wuite, Jan and Leeson, Amber A and McMillan, Malcolm}, booktitle = {Proceedings of the 7th Northern Lights Deep Learning Conference (NLDL)}, pages = {131--143}, year = {2026}, editor = {Kim, Hyeongji and Ramírez Rivera, Adín and Ricaud, Benjamin}, volume = {307}, series = {Proceedings of Machine Learning Research}, month = {06--08 Jan}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v307/main/assets/hay26a/hay26a.pdf}, url = {https://proceedings.mlr.press/v307/hay26a.html}, abstract = {Monitoring the calving dynamics of the Antarctic ice shelves is central to understanding a major driver for the changes to ocean levels on our planet. Several physical models have been proposed as calving laws, with varying predictive power. We propose an approach using Machine Learning (ML) to identify key variables and parameters that may be used in future models of the ice shelf calving dynamics. As part of an ongoing project, we have trained a U-Net on samples from a set of Gaussian Random Field-represented Essential Climate Variables (ECV). Ablation studies establish a few of the selected variables as having high correlation with calving events, with an F1 score above 0.9. Our first study site was the Larsen C Ice Shelf, on the northwest part of the Weddell Sea, where in 2017 there was a massive calving event. We have found strong correlations between the calving and the ice velocity leading up to this event, which may be further improved when accounting for basal melt rates in the area.} }
Endnote
%0 Conference Paper %T Predicting Calving Events in Antarctica using Machine Learning %A Jacob Alexander Hay %A Hamzeh Issa %A Daniele Fantin %A David Parkes %A Jan Wuite %A Amber A Leeson %A Malcolm McMillan %B Proceedings of the 7th Northern Lights Deep Learning Conference (NLDL) %C Proceedings of Machine Learning Research %D 2026 %E Hyeongji Kim %E Adín Ramírez Rivera %E Benjamin Ricaud %F pmlr-v307-hay26a %I PMLR %P 131--143 %U https://proceedings.mlr.press/v307/hay26a.html %V 307 %X Monitoring the calving dynamics of the Antarctic ice shelves is central to understanding a major driver for the changes to ocean levels on our planet. Several physical models have been proposed as calving laws, with varying predictive power. We propose an approach using Machine Learning (ML) to identify key variables and parameters that may be used in future models of the ice shelf calving dynamics. As part of an ongoing project, we have trained a U-Net on samples from a set of Gaussian Random Field-represented Essential Climate Variables (ECV). Ablation studies establish a few of the selected variables as having high correlation with calving events, with an F1 score above 0.9. Our first study site was the Larsen C Ice Shelf, on the northwest part of the Weddell Sea, where in 2017 there was a massive calving event. We have found strong correlations between the calving and the ice velocity leading up to this event, which may be further improved when accounting for basal melt rates in the area.
APA
Hay, J.A., Issa, H., Fantin, D., Parkes, D., Wuite, J., Leeson, A.A. & McMillan, M.. (2026). Predicting Calving Events in Antarctica using Machine Learning. Proceedings of the 7th Northern Lights Deep Learning Conference (NLDL), in Proceedings of Machine Learning Research 307:131-143 Available from https://proceedings.mlr.press/v307/hay26a.html.

Related Material