The Last Interglacial (LIG), spanning 129,000 to 116,000 years before present, presents a valuable analog for understanding future climate change. Northern high-latitude summer temperatures during the LIG were significantly higher (4–5 °C) than pre-industrial levels, primarily attributed to orbital forcing. However, greenhouse gas concentrations were slightly lower than pre-industrial levels. Antarctica's contribution to the LIG's high sea level is estimated to be substantial (1–7 m sea level equivalent (SLE)), significantly exceeding Greenland's contribution (0.4–4.4 m SLE). Paleo-proxy records suggest warmer Southern Ocean temperatures during the LIG, but existing climate models struggle to reproduce this warming, potentially due to the absence of Antarctic ice sheet changes in these simulations. Coupling dynamic ice sheet models with fully coupled climate models remains a significant technical challenge, leading researchers to use prescribed ice sheet changes as boundary conditions in climate models to study these impacts. Previous studies explored different mechanisms for LIG sea level rise, but the interaction between ice sheet melt, ocean warming, and atmospheric changes remains poorly understood. This study aims to investigate the climate impacts of a partial removal of the Antarctic ice sheet, considering both the topographic changes and meltwater input as separate and combined forcings in a coupled climate model simulation.

Studies on the LIG's climate highlight discrepancies between proxy data and model simulations. Paleo-proxy records indicate higher Southern Ocean summer sea surface temperatures (SSTs) (1.8 ± 0.8 °C) and annual mean SSTs (up to 3 °C) at 127 ka compared to pre-industrial levels. However, the Paleoclimate Model Intercomparison Project 4 (PMIP4) experiments tend to underestimate this warming anomaly, suggesting a lack of key factors in the model simulations. Ice core records show an Antarctic surface air temperature (SAT) anomaly of approximately -2.2 °C compared to pre-industrial levels, but model simulations show a smaller change (-0.5 °C). This discrepancy may stem from the omission of Antarctic ice sheet changes in PMIP4 experiments. While some progress has been made in coupling dynamic ice sheet models with climate models, this approach is still under development, especially in the context of paleoclimate simulations. Previous work suggests that subsurface warming of the Southern Ocean is crucial for AIS retreat, potentially leading to West Antarctic Ice Sheet (WAIS) collapse. However, the relative roles of WAIS retreat, sea ice loss, and meltwater input remain unclear. Moreover, while meltwater input is known to increase Southern Ocean stratification, leading to subsurface warming and potentially enhanced ice sheet disintegration, more research is needed to fully understand this complex system.

This study utilized the coupled climate model GFDL CM2.1, which includes components for ocean, sea ice, atmosphere, and land surface processes. The model was run with pre-industrial (PI) and LIG boundary conditions, following the PMIP4 protocol for greenhouse gas concentrations, orbital parameters, and solar constant. The researchers performed six experiments: two ice-sheet removal experiments (SL4.1 and SL7.1, representing 4.1 m and 7.1 m SLE ice loss, respectively), derived from Golledge et al. (2015), two meltwater forcing experiments (FW4.1 and FW7.1, equivalent to 4.1 m and 7.1 m SLE freshwater input), and two combined experiments (COMB4.1 and COMB7.1) with both ice sheet removal and meltwater forcing. The ice sheet removal experiments involved modifying the ice sheet topography and bathymetry in the model, replacing the ice with ocean grid cells where appropriate. The freshwater forcing experiments introduced a constant freshwater flux over 500 years adjacent to the Antarctic coast, distributed based on the modeled ice loss in each longitudinal sector. The combined experiments applied both perturbations simultaneously. The model's response, including SST, SAT, sea ice extent, ocean circulation (meridional overturning circulation and Antarctic Bottom Water formation), and neutral density surfaces, were analyzed to assess the climate impacts of each perturbation.

The study revealed contrasting climate responses to ice sheet removal and meltwater forcing. Ice sheet removal (SL4.1 and SL7.1) led to a 0.5–1.5 °C annual mean SST increase in the Weddell and Ross Seas, predominantly during summer, accompanied by summer sea ice retreat. East Antarctic coastal SSTs increased by 0.5–1 °C, with a stronger signal in SL7.1. However, there was a 1–2 °C subsurface cooling in the Amundsen and Bellingshausen Seas, potentially due to a northward shift of the Antarctic Circumpolar Current. Importantly, ice sheet removal caused a 2–4 °C increase in East Antarctic SAT in SL7.1, consistent with some proxy data. Meltwater forcing (FW4.1 and FW7.1) produced a 1–3 °C SST decrease in certain regions due to increased surface stratification and reduced deep water formation. This was associated with sea ice expansion and a subsurface warming of up to 2 °C in the Ross Sea. Antarctic SAT decreased with meltwater forcing, contradicting some proxy data. The combined experiments (COMB4.1 and COMB7.1) showed a complex interaction: surface cooling dominated due to meltwater, but subsurface warming was enhanced, especially in the Ross Sea, indicating a nonlinear interaction between the two forcings. Deep ocean warming was also significantly stronger in the combined experiments than in the meltwater-only experiments. The changes in Antarctic Bottom Water (AABW) formation were substantial across experiments, reflecting the impact of both ice loss and meltwater on deep water circulation. There were noticeable changes in wind patterns across all simulations reflecting interactions with surface temperature changes and topographic changes due to ice loss.

The findings highlight the complex interplay between ice sheet changes, ocean circulation, and atmospheric processes in shaping Antarctic climate. Previous studies often considered meltwater forcing or ice loss in isolation. This study demonstrates that these processes have contrasting and sometimes counter-intuitive effects on the Southern Ocean. The significant East Antarctic warming observed in the ice-loss-only experiments is consistent with some proxy data and highlights the importance of considering topographic changes and the associated changes in wind patterns. The meltwater-induced surface cooling is, however, at odds with some existing data which is potentially indicative of the timing and phasing of meltwater input vs. warmer SST anomalies. The non-linear interaction of ice loss and meltwater forcing on subsurface warming is a crucial finding, implying that climate models may underestimate future warming in the Southern Ocean if they do not adequately consider these coupled feedbacks. The subsurface warming could further destabilize the AIS through marine ice sheet instability, especially in vulnerable regions with reverse-sloping bedrock, even in East Antarctica.

This study demonstrates that the combined effects of Antarctic ice sheet loss and enhanced meltwater input during the LIG caused significant warming in East Antarctica and complex changes in Southern Ocean temperatures and circulation. The non-linear interaction between these forcings warrants further investigation. Future climate model projections should include fully coupled ice sheet-climate models to capture these complex feedbacks and provide a more accurate assessment of future Antarctic climate change. Further work is needed to clarify the timing and phasing of meltwater pulses and their role in shaping the observed temperature patterns during the LIG.

The study uses prescribed ice sheet changes as boundary conditions rather than a fully coupled ice sheet-climate model, limiting the model's ability to capture the full dynamic interaction between ice and climate. The atmospheric resolution (~2°) may limit the accuracy of resolving katabatic wind effects. The model results show some discrepancies with certain paleo-proxy SST data, potentially indicating issues with model representation of surface processes in the Southern Ocean. The specific spatial distribution of freshwater forcing and its rate may impact the results, and further work may be needed to investigate alternative parameterizations of meltwater input.