Sustained ocean cooling insufficient to reverse sea level rise from Antarctica

The Antarctic Ice Sheet has contributed several millimeters to global mean sea level rise over recent decades, with losses accelerating as warming oceans and atmosphere increase ice-shelf basal melting and reduce buttressing. The Amundsen Sea Embayment (ASE) is the dominant source of Antarctic mass loss and is susceptible to warm Circumpolar Deep Water reaching sub-ice-shelf cavities via bathymetric troughs, enhancing melt and driving grounding-line retreat. While many projections emphasize upper-end, worst-case outcomes under continued warming, there remain plausible futures—through emissions mitigation, internal climate variability (e.g., SAM/ENSO), or geoengineering—that could reduce ocean-driven melting. This study asks: How do multi-decadal periods of enhanced sub-ice-shelf melting affect ASE mass loss on centennial timescales, and to what extent could subsequent reductions in basal melt mitigate retreat, promote grounding-line re-advance, and offset sea-level contributions?

Prior work shows increasing ASE mass loss linked to enhanced delivery of warm Circumpolar Deep Water (CDW) to ice-shelf cavities, modulated by winds, thermocline depth, and bathymetry. Observations and models indicate decadal variability (e.g., SAM/ENSO) can produce multi-year cool periods coincident with reduced retreat, but long-term projections generally show continued warming and increased melt. Studies highlight marine ice sheet instability mechanisms and the critical role of ice-shelf buttressing. Geoengineering proposals (e.g., submarine walls) have been suggested to block CDW, though with potential side effects for neighboring shelves. Previous modeling emphasizes high-end mass loss scenarios; fewer have explored the impacts of sustained cool-ocean periods or the requirements to halt or reverse retreat. This work builds on ensemble modeling (e.g., ISMIP6, LARMIP-2) and prior ASE sensitivity studies by explicitly testing idealized periods of enhanced melt followed by restored or reduced melt conditions to quantify committed mass loss and the potential for re-advance.

Ice sheet model: The BISICLES ice sheet model (L1L2 stress balance) with adaptive mesh refinement (4 km to 0.25 km) is used. Grounding lines are explicitly resolved at grounded–floating transitions. Domain and inputs: Regional ASE domain; topography/thickness from BedMachine v1. Surface mass balance (SMB) from Arthern et al. is held constant to isolate ocean-forced responses. A 3D internal ice temperature field from Pattyn et al. is prescribed and not time-evolving. Basal traction and viscosity are inferred by inversion to match 2013–2016 observed velocities. Weertman sliding law; fixed calving front; minimum ice thickness 10 m. Melt forcing: A steady-state basal melt field (Cornford et al.) approximates present-day melt rates, tuned to observed thinning and concentrated near grounding lines, updated as grounding lines migrate. Time-varying anomalies are added to this background field, scaled to concentrate near grounding lines as they evolve. Ensemble design: 181 simulations total: 180 forced plus 1 control (constant present-day basal melt) run for 200 years. The 180 forced simulations consist of 6 groups of 30. In each group, melt anomalies increase linearly over fixed durations (20, 40, 60, 80, 100 years) to peak magnitudes (5, 10, 15, 20, 25, 30 m/yr). The first group (R0) then restores melt to present-day (anomaly = 0). The remaining five groups apply constant negative melt anomalies below present-day after the initial forcing, with reductions R1–R5 representing 1–5 m/yr reductions. Simulation codes denote duration (YR), peak magnitude (M), and reduction (R). Sub-ice-shelf melt variability: Reductions up to 5 m/yr are within observed variability inferred from mid-depth temperature fluctuations (∼0.4 °C corresponding to 4–7.6 m/yr melt variability). Filtering against observations: The 180-member ensemble is filtered by consistency with GRACE-based Antarctic mass loss trends over 2007–2022 (assumed to correspond to model years 0–15), converting mass change to SLE using standard densities and ocean area; simulations within ±2 SD of the detrended GRACE trend are retained (120 simulations) for analysis. Diagnostics: Outputs include sea-level equivalent (SLE) contributions, grounded area (GA) change, grounding-line positions by basin (Pine Island Glacier, Thwaites Glacier, and Pope–Smith–Kohler), and relationships with cumulative melt. Comparative control: Constant present-day forcing over 200 years provides a baseline for rates of mass loss and grounding-line evolution.

- Restored-melt (R0) experiments show substantial committed mass loss from finite periods of enhanced basal melting: after 100 years, ASE contributes 44–130 mm SLE; by 200 years, 82–165 mm SLE, even though melt is restored to present-day after the forcing period. - When anomalies are removed (restored to present-day), the rate of mass loss immediately drops below the control, demonstrating strong, direct sensitivity of discharge to basal melt rates. High-end (larger total melt) simulations show the greatest subsequent rate reductions. - Grounding-line response is basin-dependent. After 100 years, retreat spans a 14,494 km² difference between lowest and highest forced cases. By the second century, in the absence of further enhanced melt, grounding lines tend toward topographically controlled positions; Pine Island and PSK can re-advance (e.g., PIG advances ∼1,839 km²; PSK ∼640 km²), while Thwaites remains retreated and accounts for the greatest SLE (up to ∼70 mm by year 200 in R0). - Reduced-melt (R1–R5) scenarios consistently lower 200-year SLE relative to R0 and promote grounding-line re-advance toward initial positions. Increasing the reduction magnitude narrows ensemble spread and increases the number of simulations with lower SLE than control by year 200. - Each 1 m/yr reduction in basal melt anomalies decreases the ensemble maximum SLE by approximately 7–14 mm (figure description), and elsewhere the study reports 14–20 mm reductions in upper and lower bounds per 1 m/yr reduction. - Applying up to 5 m/yr reduction below present-day lowers the upper-end 200-year SLE by about 60 mm (from ∼165 to ∼105 mm). Grounding lines can re-advance to near present-day positions under R3–R5, yet a total mass deficit persists. - For the most strongly forced case (YR100-M30), an R2 reduction (2 m/yr below present-day) can reduce the ongoing SLE contribution rate to approximately 0 mm/yr by year 200. - Despite re-advance under strong cooling (R5), total SLE over 200 years remains ∼42–105 mm, indicating that grounding-line position and total mass balance are decoupled; widespread catchment thinning is not fully reversed. - Constant prescribed accumulation offsets about 0.7 mm/yr SLE; to fully neutralize SLE in a low-end, strongly cooled case (e.g., YR20-M5-R0), a ∼60% increase in accumulation sustained for at least 176 years would be required, highlighting that cooling alone is insufficient to reverse total mass loss. - The rate and magnitude of ongoing mass loss after cooling depend on prior melt history. High cumulative-melt simulations, upon reduction, exhibit stronger dynamic thickening and buttressing, yielding greater mitigation of discharge rates, especially for PIG and PSK; Thwaites remains the largest contributor and most complex responder.

The experiments demonstrate that even short, multi-decadal intervals of enhanced sub-ice-shelf melting commit the ASE to centuries of elevated sea-level contribution. Restoring melt to present-day can immediately slow discharge and facilitate grounding-line re-advance in some basins, indicating transient stabilization. However, positive mass-loss rates persist under present-day melting, meaning continued sea-level rise. Only reductions below present-day melting produce substantial mitigation, lower SLE, and widespread re-advance; yet the total mass loss accrued cannot be eliminated by cooling alone over two centuries because dynamic thickening near grounding lines does not compensate for prior, basin-wide thinning. The dependence on cumulative melt (duration × magnitude) underscores that total melt governs total mass loss: short, intense forcing can equal long, weaker forcing in sea-level impact. Thwaites Glacier dominates committed mass loss and exhibits complex, geometry-controlled sensitivity, consistent with marine ice sheet instability tendencies. Overall, the findings address the research question by showing that while sustained ocean cooling can slow or halt further increases in mass loss rates and promote re-advance, it is insufficient by itself to reverse Antarctica’s net contribution to sea-level rise without substantial increases in accumulation.

This study quantifies ASE sensitivity to idealized periods of enhanced basal melting followed by restoration or reduction of melt rates. Key contributions include: (1) demonstrating that total melt (magnitude × duration) determines committed centennial SLE, with short intense forcing rivaling long weaker forcing; (2) showing immediate mitigation of discharge when melting is reduced, with greater effects for higher-end cases; (3) establishing that strong, sustained cooling (up to 5 m/yr reduction) can drive grounding-line re-advance to near-present positions but cannot by itself erase the cumulative mass deficit; and (4) identifying that substantial, prolonged increases in accumulation (∼60% for ≳176 years in a low-end case) would be required to offset SLE entirely. Future research should: couple ice-sheet and ocean–atmosphere models to assess realistic pathways and variability of ocean cooling; evaluate combined strategies of emissions mitigation, potential geoengineering, and enhanced snowfall; extend analysis beyond 200 years and across AIS basins; and reduce uncertainties via improved bed topography, basal conditions, and ice-shelf buttressing representation.

- Forcing is idealized: stepwise linear increases and fixed reductions in sub-ice-shelf melt anomalies, not fully coupled to time-evolving ocean/atmosphere dynamics. - SMB is held constant; required increases in accumulation are estimated rather than simulated under interactive climate. - Internal ice temperature field is prescribed and non-evolving; calving front is fixed and minimum ice thickness imposed, potentially affecting dynamic responses. - Regional ASE setup excludes potential teleconnections and feedbacks with neighboring sectors; Thwaites’ complex buttressing evolution is sensitive to geometry and bed uncertainties. - Ensemble filtering relies on GRACE-derived Antarctic mass loss trends over 2007–2022 and an assumed alignment of model year 0 with 2007, introducing observational and alignment uncertainties. - Simulations span 200 years; longer-term hysteresis or delayed instabilities may not be fully captured. - Geoengineering scenarios are represented abstractly as melt reductions; practical feasibility, side effects, and circulation impacts are not assessed.