Uncertainty in the projected Antarctic contribution to sea level due to internal climate variability

Abstract. Identifying and quantifying irreducible and reducible uncertainties in the Antarctic Ice Sheet (AIS) response to future climate change is essential for guiding mitigation and adaptation policy decision. However, the impact of the irreducible internal climate variability, resulting from processes intrinsic to the climate system, remains poorly understood and quantified. Here, we characterise both the atmospheric and oceanic internal climate variability in a selection of three Coupled Model Intercomparison Project Phase 6 (CMIP6) models (UKESM1-0-LL, IPSL-CM6A-LR, and MPI-ESM1.2-HR) and estimate their impact on the Antarctic contribution to sea-level change over the 21st century under the SSP2-4.5 scenario. To achieve this, we use a standalone ice-sheet model driven by the ocean through parameterised basal melting and by the atmosphere through emulated surface mass balance estimates. The atmospheric component of internal climate variability in Antarctica has a similar amplitude in the three CMIP6 models. In contrast, the amplitude of the oceanic component strongly depends on the climate model and its representation of convective mixing in the ocean. A low bias in sea-ice production and an overly stratified ocean lead to a lack of deep convective mixing which results in weak ocean variability near the entrance of ice-shelf cavities. Internal climate variability affects the Antarctic contribution to sea-level change until 2100 by 45 % to 93 % depending on the CMIP6 model. This may be a low estimate, as the internal climate variability in the CMIP models is likely underestimated. The effect of atmospheric internal climate variability on the surface mass balance overwhelms the effect of oceanic internal climate variability on the dynamical ice-sheet mass loss by a factor of 2 to 5, except in the Dronning Maud area and the Amundsen, Getz, and Aurora basins, where both contributions may be similar depending on the CMIP model. Based on these results, we recommend that ice-sheet model projections consider (i) several climate models and several members of a single climate model to account for the impact of internal climate variability and (ii) a longer temporal period when correcting historical climate forcing to match present-day observations.