A pioneering climate modeling study has unveiled how regional dynamics critically influence the onset of accelerated basal melting beneath Antarctica’s vast ice shelves. Anchored by the Alfred Wegener Institute Earth System Model (AWI-ESM2), an advanced coupled Earth system model, researchers embarked on a comprehensive simulation project capturing the complex interplay between oceanic, atmospheric, and cryospheric processes. These findings shed new light on the uncertain thresholds that precipitate rapid ice-shelf loss, carrying profound implications for future sea-level rise projections and global climate patterns.
The AWI-ESM2 integrates cutting-edge components: the ocean–sea-ice model FESOM2, atmosphere model ECHAM6, and the land surface model JSBACH, allowing fully coupled feedbacks between each subsystem. This synergy enables unprecedented realism in simulating Antarctic basal melt rates, particularly within cavity regions beneath ice shelves rarely resolved in large-scale climate models. The ocean component’s unstructured CORE-II mesh is enhanced with a carefully designed extension to capture the geometry beneath Antarctica’s floating ice, known as the CORE-ICE mesh, incorporating detailed representations of ice-shelf cavities based on bedrock and bathymetric datasets such as RTopo-2.
Crucial to these simulations is the precise treatment of ice-shelf basal melting, which replaces traditional atmosphere–ocean boundary conditions with an ice-shelf–ocean interface. FESOM2 handles this by applying parameterizations for momentum, heat, and salt fluxes at the ice-shelf base, modulated by velocity-dependent coefficients to capture realistic exchange processes. Unique to these simulations is the assumption of fixed cavity geometry, meaning that changes like ice thickness variation or grounding-line retreat are not included, isolating the oceanic processes driving basal melt dynamics.
Simulations commenced with a millennium-scale spin-up utilizing pre-industrial conditions, employing two principal model configurations: the standard CORE mesh excluding ice shelves, and the enhanced CORE-ICE mesh explicitly resolving cavities under Antarctic ice shelves. From these initial conditions, historical (1851–2014) and future projections under different Shared Socioeconomic Pathway (SSP) scenarios through 2200 were conducted. Notably, freshwater inputs were carefully parameterized: traditional surface runoff was suppressed in ice shelf runs to avoid double-counting, with fresh meltwater introduced directly at the cavity base, dynamically influencing ocean stratification and circulation.
A hallmark of the modeled ocean component is the application of an improved Gent–McWilliams parameterization to represent mesoscale eddy-induced transports despite the mesh resolution’s inability to explicitly resolve these processes globally. This refinement accounts for weakly stratified regions, adjusting eddy diffusion coefficients to prevent exaggerated eddy activity, thus ensuring accurate portrayal of oceanic transport phenomena critical to heat and salt distributions affecting basal melting.
The study also innovates by embedding passive tracer experiments to track Antarctic freshwater contributions within the Southern Ocean. These tracers, released as unit concentration with meltwater or runoff depending on the simulation configuration, provide novel insight into the dispersal and impact of fresh inputs on regional water-mass transformation and global overturning circulation patterns, underscoring the subtle yet far-reaching influence of Antarctic meltwater on ocean dynamics.
One of the most compelling features of this research lies in its detailed density-coordinate diagnostics, allowing for nuanced examination of the meridional overturning circulation (MOC) beyond traditional depth-based frameworks. Such diagnostics reveal shifts in North Atlantic Deep Water (NADW) and Antarctic Bottom Water (AABW) formation under warming scenarios, with ice-shelf meltwater playing a nontrivial role in modulating density structures. Particularly, the Antarctic Bottom Water cell appears denser and stronger in ice-inclusive runs compared to standard configurations, illustrating the meltwater’s paradoxical effect on deep water formation and its climatic relevance.
These simulations uncover that despite massive basal melt rates predicted under high-end warming scenarios, the NADW production does not weaken linearly; rather, it becomes shallower in density space, signaling complex feedbacks between fresh surface inputs and stratification with potential consequences for Atlantic overturning stability. Meanwhile, coastal downslope water formation shifts towards lighter (shallower) density classes as melting intensifies, reflecting a reorganization of Antarctic shelf water processes influenced by the retreat and thinning of sea-ice cover.
Validation against observationally inferred basal melt rates reveals a strong overall agreement, although regional biases remain. FESOM2’s ocean-only setup aligns well with observed melt rates in most sectors, notably capturing the total melt budget accurately. However, disparities are apparent, such as underestimation of melt in the Amundsen Sea and overestimation in the Ross Ice Shelf and Weddell regions. The fully coupled AWI-ESM2 simulations exhibit larger hydrographic biases, particularly over certain sectors like the Amery Ice Shelf and Bellingshausen Sea, underscoring challenges in coupled model configurations where atmospheric variability induces greater systemic uncertainty.
The ensemble approach, incorporating nine members branched from different initial conditions, adds robustness to these findings by sampling internal variability, a critical feature given the chaotic nature of climate-ocean systems. Through this lens, individual simulations reveal considerable spread in basal melt responses and oceanic variables, contextualizing the confidence ranges of projections and emphasizing the necessity of ensemble modeling for anticipatory climate science.
Underlying the ocean model, the unstructured CORE-II-based meshes represent regional resolution challenges intrinsic to modeling Antarctica. While the standard mesh offers coarse resolution (~1°) over much of the ocean, it achieves finer detail (~15 km) in coastal and polar domains, enabling better representation of narrow ice-shelf cavities and continental shelves. The mesh extension technique preserves the integrity of original domains while permitting high-resolution enhancements focused exclusively on Antarctica, thus mitigating artificial impacts on global circulation patterns.
Fundamental physical processes such as brine rejection during sea-ice formation and surface cooling-driven convection are captured in these simulations, revealing their interplay in dense water formation and overall meridional overturning. For instance, brine rejection emerges as a notable contributor to deep water formation beneath the ice shelves in pre-industrial climates, while warming reduces its role, offset by increases in surface cooling contributions due to retreating sea-ice insulation effects. Such delicate balances demonstrate the importance of resolving coupled atmosphere-ice-ocean processes for accurate predictions.
The research’s time-stepping strategy, with different time steps for atmosphere, land, and ocean components and coupling occurring hourly, ensures numerical stability while facilitating frequent exchange of physical information. This temporal resolution, combined with the model’s vertical coordinate system that approximates a linear free surface in ice-shelf cavities, effectively maintains constant cavity volumes, essential for realistically simulating basal meltwater fluxes without volumetric inconsistencies.
These sophisticated modeling efforts come amid pressing concerns over Antarctic ice-sheet stability and its implications for global sea level. The study’s findings suggest that regional ocean conditions, modulated by bathymetry and ice geometry, dictate when and where accelerated basal melting surpasses critical thresholds. Such spatial heterogeneity challenges simplistic uniform melting assumptions and calls for refined regional analyses when assessing the Antarctic contribution to future climate scenarios.
Despite its advancements, the study acknowledges limitations, notably the absence of dynamic ice-shelf geometry changes such as calving, grounding-line migration, and iceberg melting. These omissions are intentional to isolate ocean-induced melt mechanisms but highlight frontiers for future model development to capture the full spectrum of ice-ocean feedbacks necessary for robust long-term predictions.
Moreover, the research confronts the challenges posed by the freely evolving atmospheric component in coupled models, which, while more physically consistent, introduces greater uncertainties relative to prescribed forcing ocean-only setups. This atmospheric freedom leads to enhanced hydrographic discrepancies, a trade-off inherent in coupling complexity that requires continuous model tuning and observational benchmarking.
In conclusion, this groundbreaking study embodies a significant leap forward in Antarctic ice shelf melt simulation, leveraging fully coupled Earth system modeling with dedicated high-resolution ocean meshes and innovative parameterizations. Its insights into the spatial variability of basal melt thresholds, freshwater impacts on overturning circulation, and ocean-ice interactions provide an indispensable foundation for refining climate projections and informing international mitigation and adaptation strategies.
Subject of Research: Antarctic ice-shelf basal melt rates and their sensitivity to regional oceanic and atmospheric conditions modeled through the Alfred Wegener Institute Earth System Model (AWI-ESM2).
Article Title: Regional conditions determine thresholds of accelerated Antarctic basal melt in climate projection.
Article References:
Song, P., Scholz, P., Knorr, G. et al. Regional conditions determine thresholds of accelerated Antarctic basal melt in climate projection. Nat. Clim. Chang. (2025). https://doi.org/10.1038/s41558-025-02306-0
Image Credits: AI Generated
