Reconstructing the Antarctic ice sheet shape at the Last Glacial Maximum using ice core data

Turner, Fiona E.; Buck, Caitlin E.; Jones, Julie M.; Sime, Louise; Vallet, Irene Malmierca; Wilkinson, Richard D.

Abstract

The Antarctic ice sheet (AIS) is the Earth's largest store of frozen water; understanding how it has changed in the past allows us to improve our future projections of how it, and thus sea levels, may change. In this paper, we use previous reconstructions of the ice sheet, water isotope ratios from ice cores, and simulator predictions of the relationship between the ice sheet shape and isotope ratios to create a model of the AIS at the Last Glacial Maximum (LGM). We develop a prior distribution that captures expert opinion about the ice sheet, generate a designed ensemble of potential AIS shapes, run these through the climate model HadCM3, and train a Gaussian process statistical model of the link between ice sheet shape and water isotope ratios. In order to make the analysis computationally tractable, we develop a preferential principal component method that allows us to reduce the dimension of the problem in a way that accounts for the differing importance we place in previous reconstructions, allowing us to create a basis that reflects prior uncertainty. We use Markov Chain Monte Carlo (MCMC) to sample from the posterior distribution, finding shapes for which HadCM3 predicts water isotope ratios closely matching observations from ice cores. The posterior distribution allows us to quantify the uncertainty in the reconstructed shape, a feature missing in other analyses.

Citation

Turner, F. E., Buck, C. E., Jones, J. M., Sime, L., Vallet, I. M., & Wilkinson, R. D. (2023). Reconstructing the Antarctic ice sheet shape at the Last Glacial Maximum using ice core data. Journal of the Royal Statistical Society: Series C, 72(5), 1493-1511. https://doi.org/10.1093/jrsssc/qlad078

• Journal Article Type: Article
• Acceptance Date: Sep 11, 2023
• Online Publication Date: Sep 18, 2023
• Publication Date: 2023-11
• Deposit Date: Oct 13, 2023
• Publicly Available Date: Oct 16, 2023
• Journal: Journal of the Royal Statistical Society: Series C
• Print ISSN: 0035-9254
• Electronic ISSN: 1467-9876
• Publisher: Wiley
• Peer Reviewed: Peer Reviewed
• Volume: 72
• Issue: 5
• Pages: 1493-1511
• DOI: https://doi.org/10.1093/jrsssc/qlad078
• Public URL: https://nottingham-repository.worktribe.com/output/25956026
• Publisher URL: https://academic.oup.com/jrsssc/advance-article/doi/10.1093/jrsssc/qlad078/7276400