Understanding the response of the Filchner-Ronne Ice Shelf (FRIS), one of Antarctica's largest ice shelves, to climate change is crucial for predicting future sea-level rise. FRIS accounts for over 10% of Antarctic Ice Sheet discharge and plays a significant role in global thermohaline circulation. Unlike ice shelves in the Amundsen Sea, which are already exposed to relatively warm deep ocean water, FRIS's cavity currently contains cold water, resulting in low basal melt rates. Two contrasting hypotheses exist regarding FRIS's future: one suggests reduced melting with climate warming due to decreased sea ice formation and weakened High Salinity Shelf Water (HSSW) flow, the other predicts increased melting due to warm deep water (WDW) intrusion. This study uses a refined coupled ice sheet-ocean model to reconcile these conflicting hypotheses and project FRIS's long-term response to climate change, addressing the significant uncertainty in how easily warm water intrusion might occur and over what timescales.

Previous studies have offered contrasting views on the future of FRIS melting. Nicholls (1997) hypothesized that climate warming would reduce sea ice formation, leading to weaker HSSW flow and decreased melt rates. In contrast, Hellmer et al. (2012) and subsequent studies proposed that increased melting would occur, triggered by WDW intrusion into the cavity via the Filchner Trough. These studies utilized ice-ocean models forced by climate change projections, showing WDW intrusion dramatically increasing basal melt rates. However, these earlier projections relied on older climate model data, raising uncertainty. Recent advancements in Antarctic modeling, with higher resolution and reduced present-day biases, warrant a re-examination of this issue.

This study employs the coupled ice sheet-ocean model ÚaMITgcm, which simulates ocean circulation, sea ice, ice shelf melting, and ice sheet flow. The model was forced with atmospheric and oceanic boundary conditions from the UKESM global climate model, following the CMIP6 protocol. Three idealized scenarios were simulated: piControl (pre-industrial CO2 levels), abrupt-4xCO2 (instantaneous quadrupling of CO2), and 1pctCO2 (1% annual CO2 increase). Each simulation ran for 150 years, followed by a 50-year extension. The ÚaMITgcm model couples the MITgcm ocean model with the Úa ice flow model. MITgcm incorporates ice shelf thermodynamics, sea ice processes, and utilizes a linear free surface scheme for numerical stability. The Úa model uses a finite-element approach to simulate ice sheet flow and geometry, with varying resolution and adaptive time stepping. The model's parameters, such as ice shelf drag coefficient and bathymetry, were refined based on recent observational data and updates. To improve representation of present-day conditions, a coastal wind correction was applied to UKESM output. Specific methods were employed to assess the thresholds for the onset of Stage 1 (reduced melting) and Stage 2 (increased melting), and to calculate the contribution to sea-level rise. The model included a mechanism to account for both iceberg and ice shelf meltwater.

The simulations revealed a two-timescale response of FRIS to climate warming, reconciling previous conflicting hypotheses. Stage 1 is characterized by weakened circulation and reduced basal melting due to freshening of the continental shelf waters, which creates a density barrier impeding HSSW inflow. This freshening is primarily caused by increased local sea ice melting and advection of fresher water from upstream regions. A salt budget analysis confirmed the dominance of surface processes and advection in driving this freshening. During Stage 1, the FRIS cavity water becomes stagnant and cools, leading to lower basal melt rates, despite a longer residence time for source water. Stage 2 is marked by a reversal in density gradients allowing WDW to intrude into the cavity via the Filchner Trough. In the 1pctCO2 scenario, WDW intrusion begins in the final years of the simulation and is repeated in the extension. In the abrupt-4xCO2 scenario, a significant WDW pulse occurred around years 80-90, doubling melt rates, followed by another larger pulse around year 140, eventually resulting in a melt rate 21 times higher than piControl. The changes in basal melt rates influence ice shelf thickness and glacier velocity. During Stage 1, the ice shelf thickens near the grounding lines, and glacier flow slows down. This is reversed in Stage 2, resulting in significant ice shelf thinning and glacier acceleration. The abrupt-4xCO2 simulation led to a total sea-level rise of only 1 cm over 40 years. The timescales of Stage 1 and 2 were found to be dependent on the extent of global warming, with Stage 1 becoming detectable after 3°C (69 years in 1pctCO2) and 5°C (14 years in abrupt-4xCO2) warming, while Stage 2 begins after approximately 7°C of warming.

This study's findings resolve the previous disagreement regarding the response of FRIS to climate change, showing both reduced and increased melting as phases of a single, two-timescale response. The initial phase of reduced melting (Stage 1) was suggested based on observations, but lacked model support. Previous modeling studies directly showed the second phase, overlooking Stage 1, possibly because they were forced by older climate model projections and may have been overly sensitive to WDW inflow. The considerable warming required for Stage 2 (7°C) suggests that increased FRIS melting might not occur in the 21st century, even under high emission scenarios. The results highlight the importance of considering the two-timescale response when predicting future sea-level rise from the Antarctic Ice Sheet. The paradox of reduced melting potentially serving as an early warning sign for later destabilization necessitates continued monitoring of the FRIS region.

This study provides the first comprehensive model-based evidence of a two-timescale response in FRIS melting to climate change, resolving prior conflicting hypotheses. The model predicts that significant increases in melt rates require substantial global warming (7°C), making them unlikely this century unless mitigation efforts fail dramatically. However, the delayed but eventual increase in melting emphasizes the need for long-term monitoring of the FRIS region and further model intercomparison studies to refine our understanding of the system's dynamics and associated sea-level rise contributions.

The study used idealized climate scenarios which may not accurately reflect the complexity of future climate change. The use of a single global climate model (UKESM) and a single ensemble member introduces model uncertainty. The model resolution might limit the accuracy of simulating small-scale processes, such as coastal dynamics. The timescales simulated are limited by the computational expense of coupled ice-ocean models, hence the full response of the ice sheet might take longer than simulated. Internal climate variability may also influence the timing of WDW pulses, adding uncertainty to the exact timescales of Stage 1 and Stage 2.