What Antarctic sea-level rise estimates to 2050 should be used for decision-making?

We focus on the Antarctic contribution to sea level by 2050, intending to improve sea-level rise estimates for decadal decision-making purposes. We compare ISMIP6 2100 ice sheet model-derived projections and data-driven estimates from 2015-2050. We find that models divide into two categories of response based on their initialisation approach, with spin-up-style models generally showing little response to forcing (relative to their control) over this period even under a high emissions scenario, while data assimilation models suggest increased change in the Amundsen Sea Embayment and parts of East Antarctica, and accelerating ice loss along the Siple Coast. We suggest a lower surface mass balance in the forced simulations than the control simulations drives an unrealistic mass loss signal in the Amundsen Sea sector in the ISMIP6 projections over 2015-2050. We then focus on the data assimilation models and explore their projection of the dynamic contribution to sea levels by 2050. We complement these with data-driven estimates based on linear or linear-plus-quadratic models fit to gridded satellite altimeter data while also considering natural climate variability that dominates decadal-scale surface mass balance variations. Historical trends (i.e., pre-2015) are not necessarily captured in the ISMIP6 2100 models, partly due to a lack of observational constraints before the satellite record. Hence, we use gridded empirically-derived surface lowering trends combined with the ISMIP6 projected trends, and compare them to the linear and quadratic linear and quadratic empirical extrapolations to 2050. Finally, we explore the differences and sensitivities in sea level fingerprints deriving from these estimates and their potential implications for decision-making processes.