Eliot Jager: Ice sheet modelling: The quest for reliable forecasts

Abstract: Accurately projecting ice sheet mass loss is critically important for society, as it directly impacts coastlines worldwide. However, despite recent improvements in ice sheet models, few models accurately reproduce historical mass loss, and their predicted outcomes appear overconfident. This inability to replicate the past raises serious concerns about the models' prognostic skill.

To address this, we present a framework for generating robust projections of the sea-level rise contribution from Upernavik Isstrøm, a Greenland tidewater glacier. We demonstrate how: (1) incorporating parameterizations based on known physics improves the reconstruction of past behavior; (2) rigorous uncertainty quantification, while increasing the stated uncertainty, provides more honest and reliable forecasts; and (3) leveraging observational data is key to proving the efficacy—not just the application— of Bayesian calibration. Collectively, these steps form an essential pathway toward more reliable ice sheet forecasts.