Two-timescale response of a large Antarctic ice shelf to climate change

The study investigates how the Filchner-Ronne Ice Shelf (FRIS), a major Antarctic ice shelf crucial to Antarctic mass balance and global thermohaline circulation, will respond to climate warming. Presently, the FRIS cavity contains cold water (~ -2 °C) with relatively low basal melt, unlike Amundsen Sea ice shelves that interact with warmer deep water. Two competing hypotheses exist: (1) warming reduces sea-ice formation, weakens High Salinity Shelf Water (HSSW) inflow, and decreases basal melt (Nicholls, 1997); and (2) warming redirects Warm Deep Water (WDW) into the Filchner Trough, dramatically increasing basal melt (Hellmer et al., 2012, and related work). With advances in global and regional modeling since earlier studies, the paper aims to reconcile these views and quantify the thresholds and timescales by which FRIS may transition from reduced to increased melting under climate change.

Earlier observations linked higher FRIS melt rates with stronger HSSW circulation, leading Nicholls (1997) to hypothesize that climate warming would lessen sea-ice formation, reduce HSSW inflow, and thereby lower melt. In contrast, modeling studies by Hellmer and colleagues (2012, 2013, 2017) and Timmermann & Goeller (2017) suggested that WDW could intrude via the Filchner Trough under warming, greatly increasing melt. Other idealized or sensitivity studies (e.g., Hazel & Stewart, 2020; Daae et al., 2020) showed circulation bistability or responses to altered forcing. However, previous projections of WDW intrusion relied on an older climate model projection and did not account for updated atmospheric biases or more sophisticated regional coupled models. Recent CMIP6 models exhibit improved Antarctic atmospheric circulation, and regional ice-ocean models have advanced in resolution and coupling to ice dynamics, motivating a re-examination of FRIS response.

The authors use the coupled ice sheet–ocean model UaMITgcm for the Weddell Sea/FRIS region, which couples MITgcm (ocean, sea ice, and ice-shelf thermodynamics) and Úa (finite-element ice flow) via an offline annual exchange of basal melt rates and evolving ice-shelf geometry. The ocean domain covers the Weddell Gyre with open boundaries at 30°E and 61°S, horizontal resolution ~0.25° (scaled by latitude), and 120 vertical levels including 25 m resolution in the cavity. Bathymetry is primarily Bedmap2 with updates in the Filchner cavity and inclusion of grounded iceberg A23-A. MITgcm employs a linear free-surface scheme, GM-Redi eddy parameterization, KPP mixing, EVP sea-ice rheology, and an ice-shelf drag coefficient of 0.0025. Úa resolves the FRIS catchment with adaptive mesh (300 m to 50 km), fixed ice front, surface mass balance from RACMO2.1, and basal mass balance from MITgcm. Basal slipperiness and rheology parameters are inverted using MEASURES surface velocities (1996–2016). Coupling proceeds in 1-year segments: MITgcm provides basal melt to Úa, and Úa returns updated ice geometry and floating mask. Newly opened ocean cells are initialized from neighboring properties with transport-preserving velocity adjustments; bathymetry digging is applied as needed to avoid spurious subglacial lakes.

- Two-stage response to warming: (Stage 1) weakened circulation and reduced basal melting; (Stage 2) WDW intrusion and increased melting. - Shelf freshening drives both stages. Average continental shelf salinity in front of FRIS decreases by ~0.7 psu (1pctCO2) and ~1 psu (abrupt-4xCO2) relative to piControl, largely due to increased local sea-ice melt, with contributions from advection of fresher water (from upstream increased ice-shelf melt and enhanced Southern Ocean precipitation) and, later, reduced sea-ice formation; partially offset by enhanced diffusion. Freshening in abrupt-4xCO2 plateaus after ~80 years. - Stage 1 (decreased melt): Reversal of the density gradient between continental shelf HSSW sources and the cavity reduces HSSW inflow, slows circulation, cools cavity waters, and lowers basal melting. Basal mass loss is 20% lower (1pctCO2) and 27% lower (abrupt-4xCO2) versus piControl averages during Stage 1. - Stage 2 (increased melt): With continued freshening, the density contrast between ISW and offshore WDW reverses in the Filchner Trough, enabling WDW pulses to enter the cavity. In 1pctCO2, initial WDW intrusion begins near year 145; in abrupt-4xCO2, a first major pulse occurs around years 80–90, warming the cavity by ~0.2 °C and doubling melt relative to piControl. A later, larger pulse from ~year 140 leads, by the end of the 50-year extension, to melt rates ~21× piControl and cavity temperatures ~2.7 °C warmer. - Timescales and thresholds: Stage 1 becomes detectable after ~69 years in 1pctCO2 (~3 °C global warming) and ~14 years in abrupt-4xCO2 (~5 °C). Stage 2 begins at ~79 years (abrupt-4xCO2) and ~147 years (1pctCO2), corresponding to ~7 °C of global mean near-surface warming above pre-industrial. - Ice-sheet impacts: During Stage 1, thicker ice shelves near grounding lines and slower upstream glaciers (e.g., Slessor, Foundation) due to increased buttressing. During Stage 2, widespread ice-shelf thinning, acceleration of glaciers (notably Institute Ice Stream and Slessor Glacier), and grounding-line retreat (largest near Institute). Simulated sea-level contribution reaches ~1 cm over the modeled period, occurring entirely in the last ~40 years and accelerating, implying longer-term impacts beyond simulation length.

The results reconcile prior conflicting hypotheses by showing a two-timescale evolution: an initial prolonged phase of reduced FRIS melt due to weakened HSSW inflow and cavity cooling, followed eventually by WDW intrusion and enhanced melt once sufficient freshening reverses key density gradients. This framework clarifies that observed or modeled reductions in melt do not necessarily imply long-term stability. The thresholds identified (Stage 1 detectability around 3–5 °C and Stage 2 onset near 7 °C of global warming in UKESM) suggest that increased FRIS melt is unlikely within the 21st century under most scenarios, although extreme forcing could trigger WDW inflow earlier. The findings emphasize sensitivity to continental-shelf freshening mechanisms (sea-ice processes, precipitation-driven freshwater, upstream melt) and point to the role of internal variability in timing warm-water pulses. The study highlights the need for coordinated model intercomparisons and sustained observations (especially in Filchner Trough) to constrain timescales and mechanisms, and to assess generality across models and forcings.

The study demonstrates that FRIS exhibits a two-stage response to climate warming: a long period of weakened circulation and reduced basal melting followed, under stronger warming, by warm deep-water intrusion and substantial melt increase. The Stage 2 threshold is reached only after ~7 °C global warming in the UKESM-forced experiments, implying limited likelihood of increased FRIS melt this century under most scenarios. However, persistent warming eventually overcomes initial stabilization, so reduced melt should not be interpreted as unconditional stability. The work reconciles prior hypotheses, quantifies thresholds and timescales, and underscores the importance of shelf freshening processes. Future research should include multi-model and multi-ensemble intercomparisons, higher-resolution atmosphere–ocean–ice coupling to capture feedbacks, and long-term observational programs to detect Stage 1 signatures and better constrain model projections.

- Forcing from a single climate model (UKESM1-0-LL) with relatively high climate sensitivity; thresholds and timelines may differ across CMIP6 models. - Single ensemble member per scenario; internal variability may affect timing of warm-water pulses and detectability. - Idealized forcing experiments (piControl, 1pctCO2, abrupt-4xCO2) rather than full-length realistic scenarios; extensions repeat the final decade of forcing. - UKESM low coastal resolution and known biases in Antarctic coastal winds; a static coastal wind correction is applied, but uncertainties remain in coastal dynamics and the relative role of advection in the salt budget. - Regional, offline-coupled ice–ocean framework lacks fully coupled atmosphere–ocean–ice feedbacks; iceberg and meltwater freshwater flux treatments differ between UKESM and regional model, adding structural uncertainties. - Computational constraints limit simulation length and ensemble size; the modeled sea-level contribution (~1 cm) likely underestimates long-term response because full ice-sheet adjustment occurs on longer timescales.