10.1 How Are Local Extreme Sea Level Projections Affected by Distinct Storylines in Antarctic Ice Sheet Mass Loss?

Wednesday, 15 January 2020: 1:30 PM
158 (Boston Convention and Exhibition Center)
Daniel Gilford, Rutgers Univ., New Brunswick, NJ; and D. J. Rasmussen, R. Kopp, E. Ashe, R. DeConto, and D. Pollard

Handout (23.4 MB)

Global and local sea levels are projected to rise considerably this century, amplifying the frequency of extreme sea levels that can result in deadly and costly coastal flooding. But projections of future sea level rise are highly uncertain, due primarily to uncertainties in human decisions (through emissions pathways) and deep uncertainty in the mechanisms that could drive Antarctic ice-sheet mass loss. In some coastal regions, the differences between best and worst case scenario projections can be a meter of sea level rise or more depending on the ice-sheet physics considered. Understanding and differentiating these possible storylines in the Antarctic ice-sheet remains a research priority, especially to inform coastal adaption and decision-making. In this study we use a supervised machine learning technique to emulate an Antarctic ice-sheet model. The model is forced with two different emissions pathways and model run ensembles are constructed with different depictions of ice-sheet physics. In particular, these ensembles differ depending on whether or not marine ice-cliff instability (initiated by hydrofracturing driven by surface melt through atmospheric warming) is considered a primary loss mechanism. The emulated projections following these distinct physical and emission storylines in the Antarctic ice-sheet are integrated with a holistic sea level projection framework to produce localized sea level projections across the globe. We then calculate an amplification factor---an increase in expected frequency of extreme sea level events due to local sea level rise---associated with each storyline through 2100. While not a direct measure of coastal flooding (because it does not account for local factors such as terrain or infrastructure), the amplification factor is a useful quantification of how different storylines in the Antarctic ice-sheet influence local extreme sea levels. Leveraging our machine learning emulation technique and Bayesian inference, we find that paleoclimate estimates and modern day observations of Antarctic mass loss can inform the physical mechanisms driving Antarctic contributions to sea level rise. Associated extreme sea level projections are sensitive to both the emissions pathway and Antarctic ice-sheet physics. Emulation provides promising insight into the deep uncertainties in Antarctic ice-sheet mass loss. In particular, we show that these uncertainties make it difficult to distinguish which storyline our planet is following until at least mid-century. However, improved paleoclimate constraints or better understanding of ice-sheet physics is able to reduce this uncertainty.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner