Human-induced global warming is causing a rise in global mean sea level, with the Antarctic Ice Sheet (AIS) contribution being the most uncertain component. Projections from the Ice Sheet Model Intercomparison Project 6 (ISMIP6) show a wide range of AIS contributions under different emissions scenarios, partly due to uncertainties in climate response and ice sheet model parameterizations. Key uncertainties include sub-ice shelf basal melting, surface mass balance, and parameters related to bedrock conditions, basal friction, and ice rheology. These uncertainties obscure any clear emissions scenario dependence in 21st-century sea-level projections. This research aims to quantify the impact of ice flow-related uncertainty on AIS sea-level contribution projections under different Representative Concentration Pathways (RCPs) and relate these findings to global warming levels (GWLs) to provide long-term projections.

Previous studies have highlighted uncertainties in ice sheet projections, particularly concerning the parameterization of sub-ice shelf basal melting and surface mass balance. The influence of poorly constrained model parameters (bedrock conditions, basal friction, ice rheology) on ice sheet sensitivity to climate forcing across different timescales has also been emphasized. Statistical techniques have been employed to determine probabilistic sea-level contributions, but the temporal evolution of these contributions under varying emissions scenarios remains unclear. Existing studies often underpredict observed ice mass loss, pointing to limitations in our understanding of ice sheet dynamics.

The study uses ensembles of ice sheet simulations with the Parallel Ice Sheet Model (PISM) under RCP2.6 and RCP8.5 scenarios, extending simulations to 2300. Climate forcing data is derived from the NorESM1-M CMIP5 model. A full-factorial parameter sampling of four unconstrained model parameters related to ice flow is conducted (shallow ice and shelf approximations enhancement factors, sliding law exponent, minimum till friction angle). These simulations are used to train a statistical emulator using Gaussian Process regression (GPR), which incorporates the direct, cumulative, and committed effects of global warming. The emulator's accuracy is validated through hindcasting, comparing emulated ice mass changes to historical estimates. Parameter combinations producing unrealistic hindcasts are excluded, resulting in a historically constrained parameter space. The emulator is then used to explore the sensitivity of Antarctica’s sea-level contribution to model parameters under various scenarios (including RCP4.5 and RCP6.0) and to analyze spatial patterns of ice sheet change. Sensitivity to ocean temperatures is explored by incorporating climate forcings from other CMIP5 models. Threshold exceedance is examined by comparing ice thickness changes under different scenarios to a control ensemble with constant climate conditions. Finally, projections are analyzed in relation to Global Warming Levels (GWLs) to understand the impact of warming magnitude and rate on AIS sea-level contribution.

Historically constrained projections show a narrower range of sea-level contributions (0.12–0.44 m for RCP2.6 and 0.21–0.56 m for RCP8.5 at 2100). Median sea-level contributions diverge after 2050, but substantial overlap between RCP2.6 and RCP8.5 persists for over a century. The time of emergence, where the difference between the two scenarios exceeds the 95% confidence interval of RCP2.6, is 2189. Spatial patterns show similar thinning and grounding-line retreat in the Amundsen Sea Embayment (ASE) under both scenarios, but contrasting behavior in the Ross and Ronne-Filchner ice shelves. Ocean temperature variations among different CMIP5 models lead to differing amounts of ice shelf thinning, underscoring uncertainty in basal melt rates. The ASE is identified as the region most sensitive to ice flow parameters. The rate of sea-level contribution diverges around 2070, with the RCP8.5 showing a faster increase due to enhanced ice shelf thinning. The long-term (285-year) response indicates continued mass loss, but the primary difference between high and low emission scenarios lies in the magnitude of ice loss in Wilkes Land and the wider East Antarctic Ice Sheet (EAIS). Analysis of different scenarios (RCP4.5, RCP6.0, SSP126, SSP585) and GWLs (1°C to 5°C) revealed considerable overlap in 21st-century sea-level contributions, with significant divergence only emerging in subsequent centuries primarily due to ice shelf disintegration. Changing the rate at which 2°C warming is achieved had minimal impact on long-term sea-level commitment, but exceeding 2°C significantly increased projected sea-level rise.

The findings suggest that the impact of emissions on AIS sea-level contribution will not be clearly distinguishable for over 100 years. Partial collapse of the Ross and Ronne-Filchner ice shelves, driven by decades of accumulated ocean warming, would be an early warning sign. The study's projections are markedly higher than those of ISMIP6, primarily because the study includes the committed ice sheet response and applies historical constraints that limit the range of plausible projections. The results align with other studies showing a divergence in the rate of ice loss between different emissions scenarios after 2060–2080. Ocean thermal forcing is the main driver of AIS mass loss, with uncertainties in Southern Ocean heat uptake and redistribution being crucial. Meltwater feedback, not included in the models, could further amplify ice loss. Regional variations in ice sheet response are highlighted: MISI-prone regions like the ASE and Wilkes Land are highly sensitive to ice flow parameters, whereas ice shelf thinning in the Ross and Ronne-Filchner sectors is strongly climate scenario-dependent.

This study demonstrates that total 21st-century warming, rather than the rate of warming, is the primary control on long-term AIS sea-level commitment. While uncertainties remain in ice flow and sliding parameters, particularly in MISI-prone regions, the disintegration of large ice shelves under high GWLs (>2°C) leads to irreversible, multi-meter sea-level rise. Further research focusing on improving parameterizations of ice flow, basal melt, and meltwater feedback is essential for reducing uncertainties in future AIS projections.

The study's reliance on a single AOGCM for climate forcing introduces limitations. Other AOGCMs display different magnitudes of Southern Ocean warming, which could affect the timing and magnitude of projected sea-level rise. The study does not include meltwater feedback mechanisms, which might amplify ice loss. Model resolution could also affect the results. Finally, the assumption of constant climate conditions after 2100 provides a minimum estimate for the sea-level commitment.