Editors’ Highlights are summaries of recent papers by AGU’s journal editors.
Source: Journal of Advances in Modeling Earth Systems
The amount that Antarctic ice shelves melt now and in the future is considered to be a source of “deep uncertainty” in critically important projections of global sea level. This is partly because measurements of melt from the underside of ice shelves are extremely rare, due to the remote locations involved and the inaccessibility of ice shelf cavities. Model-based simulations are therefore essential, but traditional ocean models are computationally expensive to run.
To tackle this problem, Burgard et al. [2023] develop a new method that uses neural networks and a deep learning approach to emulate what ocean models predict. Although the emulated melt is not yet a perfect match to the traditional models, the approach nonetheless represents an important step forward in how these challenging environments might be modeled. That is because statistical models are typically much faster to run than traditional models, which means that this advance has the potential to accelerate and refine future projections of melting, reducing the uncertainty around future sea level rise.
Citation: Burgard, C., Jourdain, N. C., Mathiot, P., Smith, R. S., Schäfer, R., Caillet, J., et al. (2023). Emulating present and future simulations of melt rates at the base of Antarctic ice shelves with neural networks. Journal of Advances in Modeling Earth Systems, 15, e2023MS003829. https://doi.org/10.1029/2023MS003829
—Nicholas Golledge, Associate Editor, JAMES