Multi-model estimate of Antarctic ice-shelf basal mass budget and ocean drivers

Societal adaptation to rising sea levels requires robust projections of the Antarctic Ice Sheet's retreat, particularly due to ocean-driven basal melting of its fringing ice shelves. Recent advances in ocean models that simulate ice-shelf melting offer an opportunity to reduce uncertainties in ice–ocean interactions. Here, we compare several community-contributed, circum-Antarctic ocean simulations to highlight inter-model differences, evaluate agreement with satellite-derived melt rates, and examine underlying physical processes. All but one simulation use a melting formulation depending on both thermal driving (T^⋆) and friction velocity (u^⋆), which together represent the thermal and ocean current forcings at the ice–ocean interface. Simulated melt rates range from 650 to 1277 Gt yr⁻¹ ($m=\mathrm{0.45}-\mathrm{0.91}$ m yr⁻¹), driven by variations in model resolution, parameterisations, and sub-ice shelf circulation. Freeze-to-melt ratios span 0.30 % to 30.12 %, indicating large differences in how refreezing is represented. The multi-model mean (MMM), produces an averaged melt rate of 0.64 m yr⁻¹ from a net mass loss of 843 Gt yr⁻¹ (876 Gt yr⁻¹ melting and 33 Gt yr⁻¹ refreezing), yielding a freeze-to-melt ratio of 3.92 %. We define a thermo-kinematic melt sensitivity, $\mathit{\zeta }=m/\left(T^{\star }{u}^{\star }\right)=\mathrm{4.82}\times {\mathrm{10}}^{-\mathrm{5}}$ °C⁻¹ for the MMM, with individual models spanning $\mathrm{2.85}\times {\mathrm{10}}^{-\mathrm{5}}$ to $\mathrm{19.4}\times {\mathrm{10}}^{-\mathrm{5}}$ °C⁻¹. Higher melt rates typically occur near grounding zones where both T^⋆ and u^⋆ exert roughly equal influence. Because friction velocity is critical for turbulent heat exchange, ice-shelf melting must be characterised by both ocean energetics and thermal forcing. Further work to standardise model setups and evaluation of results against in situ observations and satellite data will be essential for increasing model accuracy, reducing uncertainties, to improve our understanding of ice-shelf–ocean interactions and refine sea-level rise predictions.