The Antarctic ice sheet's mass loss has accelerated in recent decades, primarily due to reduced ice shelf buttressing caused by ocean-induced thinning. This is particularly evident in the Amundsen Sea Embayment, where ice shelf cavities are in contact with relatively warm deep water. Ocean modeling studies suggest a potential for near-future thermal regime shifts in the Filchner–Ronne and Ross ice shelves, leading to increased sub-shelf melt rates and ice shelf thinning. This study aims to quantify the impact of such a regime shift on the mass balance and stability of the Antarctic ice sheet, focusing on the implications for future ice loss and grounding line retreat. The significance lies in understanding the potential for irreversible ice loss from these currently relatively stable regions and the consequent contribution to global sea-level rise.

Previous research highlights the potential for a cold to warm ocean regime shift within the sub-shelf cavities of the Filchner–Ronne and Ross ice shelves. Studies have shown the potential for near-future thermal regime shifts where intrusions of warm deep water replace colder shelf waters. These temperature increases would significantly boost sub-shelf melt rates, causing ice shelf thinning and loss of buttressing, ultimately increasing ice flow across grounding lines. While the potential impact is significant, the implications for ice sheet dynamics and future mass loss remain largely unquantified. Existing literature predominantly focuses on the Amundsen Sea Embayment, where this thermal shift has already occurred, but lacks comprehensive analysis of the consequences in currently colder regions like the Filchner–Ronne and Ross basins.

The study employs the ice sheet model Úa, incorporating an ensemble of ocean-circulation model-derived melt rates from two independent modeling groups (FESOM and NEMO). Six perturbed states, representing 10-year averages from 100-year idealized perturbation experiments, were used as forcing for the ice sheet model. To account for uncertainties in melt rates as the geometry evolves, three conceptually different parameterizations were employed: fixed ocean model melt rates (M_o), a modern analogue technique (M_mat), and a quadratic dependency on thermal forcing (M_quad). The ice sheet model was initialized using observed ice geometry, velocities, and a surface mass balance field. Forward-in-time experiments simulated the ice sheet response for 100 years after an instantaneous switch to warm ocean conditions. Reversibility experiments tested the irreversibility of grounding line retreat by removing or reducing the perturbation in ice shelf melt rates after phases of accelerated retreat and continuing the simulations up to 1000 years. The model includes a regularized Coulomb sliding law, Glen’s flow law, and adaptive mesh refinement to maintain high resolution near the grounding line. Calving processes and solid earth feedbacks were not included.

The simulations reveal that an instantaneous switch to warm ocean conditions in the Filchner–Ronne and Ross ice shelves causes them to become significant contributors to global sea-level rise within 100 years. In high-end scenarios, ice volume loss in the Filchner–Ronne region reaches 64–97 mm sea-level equivalent (SLE) after 100 years, with grounding lines in the Robin subglacial basin retreating approximately 160 km inland. The Ross catchment switches from mass gain to mass loss in high-end scenarios, with losses varying from 7 to 39 mm SLE after 100 years. Grounding line retreat shows a dependence on both the magnitude of the perturbation and underlying topography; regions with lightly grounded ice exhibit limited retreat, while those with more heavily grounded ice experience greater retreat. The Amundsen Sea Embayment continues to lose ice at a rate similar to present-day observations, albeit with some variation depending on melt rate parameterization. Crucially, in high-end scenarios, grounding line retreat in the Möller ice stream (Filchner-Ronne) becomes irreversible; even when melt rates are returned to control conditions, the retreat continues. Similarly, at the Bindschadler ice stream (Ross), a region of irreversible retreat occurs. At Thwaites Glacier (Amundsen Sea), retreat becomes irreversible when grounding lines reach approximately 75 km inland. The model also suggests that even with increased snow accumulation in a warmer climate, the irreversibility of grounding line retreat once they migrate far enough inland remains unaffected.

The findings demonstrate that a regime shift in ocean-induced melt rates can trigger substantial and irreversible ice loss from the Filchner–Ronne and Ross regions, regions currently not contributing significantly to sea-level rise. The study highlights the potential for these regions to become major contributors to global sea level rise in the coming centuries, exceeding the current contribution of the Amundsen Sea Embayment. The onset of irreversible retreat, observed across multiple simulations and melt rate parameterizations, underscores the robustness of the findings. While the exact timing of a regime shift remains uncertain, the study demonstrates the potential for rapid and extensive ice loss once such a shift occurs. The observed differences in the reversibility of retreat across different glacier systems may be attributed to differences in underlying topography and ice thickness.

This study demonstrates that a transition to warm ocean conditions in the Filchner–Ronne and Ross ice shelf cavities could trigger substantial and irreversible ice loss, surpassing the present-day contribution of the Amundsen Sea Embayment. Irreversible grounding line retreat occurs across multiple locations, highlighting the potential for significant future sea-level rise contributions. Future research should focus on improving coupled ice-ocean models to better capture the full range of melt-geometry feedbacks, refining model parameterizations of basal sliding and calving, and incorporating the full range of climate change scenarios and their associated uncertainties.

The study uses an ice sheet model and does not incorporate all the complexities of a fully coupled ice-ocean system. Calving processes and solid earth feedbacks were not explicitly included, which could influence the rates of ice loss and grounding line retreat. The instantaneous application of warm ocean conditions in the perturbation experiments does not perfectly represent the gradual nature of a real-world regime shift. The model's reliance on parameterizations for sub-shelf melt rates introduces uncertainties in the precise quantification of ice loss. Finally, the choice of basal sliding law and its uncertainty could affect the transient retreat rates but is shown to be tightly bounded on decadal to centennial timescales.