Pierre Mathiot

Numerical ocean models remain essential for investigating ice–shelf–ocean interaction to estimate the fate of the Antarctic Ice Sheet in a warming climate. A persistent challenge in z-coordinate ocean models is the artificial stepwise representation of cavity geometries, which induces excessive diapycnal mixing and numerical noise beneath the ice draft. Here, we assess the potential of the Brinkman Volume Penalization (BVP) method for ice–ocean modeling, in which the ice–ocean interface is replaced by a porous region that is partially ocean and partially ice. The benefits of penalization are evaluated in a z-coordinate ocean model using an idealized ISOMIP+ configuration (Asay-Davis et al. 2016) under both warm and cold scenarios. First, smoothing the cavity with the BVP acts as an effective local increase in vertical resolution, improving ocean circulation near the grounding line and allowing strong basal melt at the ice-shelf base. Second, we show that the distribution of heat and meltwater within the penalized region reduces spurious mixing, leading to colder and fresher ice-shelf waters and a strengthened overturning circulation. Hence, the BVP approach provides a promising alternative to a Losch (2008) melt-layer formulation, for which similar noise reduction requires prescribing a melt-layer depth that is too large and leads to excessive mixing. Finally, we show that at least four vertical grid cells are required to resolve grounding-zone overturning and to obtain strong and consistent basal melt patterns. In summary, these results provide practical guidance for improving the numerical treatment of ice-shelf cavities in OGCMs.

The importance of Antarctic iceberg meltwater for the Southern Ocean has been a strong incentive to include Lagrangian icebergs in ocean and climate models. However, the modeled iceberg thickness has previously been constrained to 250 m, which appears to be underestimated given the frequent observations of icebergs grounded on substantially deeper bathymetric ridges. In this study, we define the initial iceberg thickness based on the thickness of the ice shelf from which they calve and rationalize the way to define the iceberg size classes. Ocean–sea‐ice–iceberg simulations reveal that more than half of the iceberg melting is discharged over the Antarctic continental shelf, reaching an average of 631 Gt when the interaction of icebergs with bathymetry is neglected, and 802 Gt when a simple grounding scheme is implemented. Such additional freshwater injected over the continental shelf prone to sea ice production has important consequences for the ocean properties around Antarctica. It enhances ocean stratification, inhibits deep convection and isolates dense deep waters from the colder and fresher surface layers. This isolation allows the deep waters to warm and become saltier, thereby intensifying subsequent ice‐shelf melting. Moreover, the simulated distribution of grounded icebergs blocks drifting sea ice, promoting the formation of thick sea ice and polynyas when the fast ice parameterization is included. This work emphasizes the need for further research into the physical representation of iceberg grounding and iceberg–sea‐ice interaction.

The Antarctic coastal marine region is a unique and highly complex environment, of which landfast ice and polynyas are key features, especially in the context of dense water formation. Current large‐scale ocean‐sea ice models used in climate studies simulate hardly any Antarctic landfast ice, which has presumably negative implications on sea ice and polynya dynamics. Here we develop, implement, and evaluate an empirical circumpolar Antarctic landfast‐ice representation for large‐scale ocean‐sea ice models. This representation is based on the restoring of sea ice velocity to zero where and when landfast ice is observed, according to a recently released circum‐Antarctic landfast ice database. Using 2001–2017 hindcast simulations with the NEMO‐ model, we demonstrate that prescribing landfast ice not only ensures accurate landfast ice coverage, as expected, but also largely improves the simulated landfast ice thickness and polynya dynamics. This includes more realistic polynya coverage, individual polynya shape, frequency, and ice production rates. Additionally, the model low bias in summer ice extent is reduced, as prescribing landfast ice locks thicker ice near the coast, taking longer to melt. Our simulations also give the first estimate of landfast ice volume, representing 10.6% of the pan‐Antarctic total, compared to 3.8% of the total Antarctic sea ice extent. We argue that velocity restoring is appropriate for some investigations of the Antarctic landfast ice over the recent past, but not for the remote past or future projections, for which a physical representation of landfast ice drivers, particularly iceberg‐sea ice interactions, is necessary.

Antarctica’s contribution to global sea-level rise is deeply uncertain, with subglacial water suspected to play a critical role, yet its impact remains unclear. We demonstrate that water at the base of ice sheets influences sliding behaviour and that its exclusion from models can underestimate sea-level rise projections and delay the predicted onset of tipping points. Here we use an Antarctic Ice Sheet model (Elmer/Ice) to explore how different assumptions about water pressure at the ice base affect sea-level rise projections from 2015 to 2300. Our results indicate that incorporating subglacial water can amplify ice discharge across the Antarctic Ice Sheet by up to threefold above the standard approach, potentially contributing an additional 2.2 metres to sea-level rise by 2300. Notably, a smoothly decreasing basal drag near the grounding line more than doubles grounding line flux by 2300 relative to scenarios where effective pressure is simplified into a spatially constant coefficient. Basin-specific responses vary significantly, with some scenarios advancing tipping points by up to 40 years. These findings underscore the critical need to integrate evolving subglacial hydrology into ice sheet models.

Antarctica's contribution to global sea-level rise (SLR) is deeply uncertain, with subglacial water suspected to play a critical role, yet its impact remains unclear. We demonstrate that water at the base of ice sheets influences sliding behavior and that its exclusion from models can underestimate SLR projections and delay the predicted onset of tipping points. Here we use the Antarctic Ice Sheet model (Elmer/Ice) to explore how different pressure determination methods at the ice base, with subglacial water present, affect SLR projections from 2015 to 2300. Our results indicate that incorporating subglacial water can amplify ice discharge across the Antarctic Ice Sheet by up to threefold above the standard approach, potentially contributing an additional $~$2.2 metres to SLR by 2300. Basin-specific responses vary significantly, with some scenarios advancing tipping points by up to 40 years. These findings underscore the critical need for integrating evolving subglacial hydrology into ice sheet models.

The glaciers of North Greenland are hosting enough ice to raise sea level by 2.1 m, and have long considered to be stable. This part of Greenland is buttressed by the last remaining ice shelves of the ice sheet. Here, we show that since 1978, ice shelves in North Greenland have lost more than 35% of their total volume, three of them collapsing completely. For the floating ice shelves that remain we observe a widespread increase in ice shelf mass losses, that are dominated by enhanced basal melting rates. Between 2000 and 2020, there was a widespread increase in basal melt rates that closely follows a rise in the ocean temperature. These glaciers are showing a direct dynamical response to ice shelf changes with retreating grounding lines and increased ice discharge. These results suggest that, under future projections of ocean thermal forcing, basal melting rates will continue to rise or remain at high level, which may have dramatic consequences for the stability of Greenlandic glaciers.

A generalised methodology to deploy different types of vertical coordinate systems in arbitrarily defined time-invariant local areas of quasi-Eulerian numerical ocean models is presented. After detailing its characteristics, we show how the novel localisation method can be used to improve the representation of the Nordic Seas overflows in the UK Met Office NEMO-based eddy-permitting global ocean configuration. Three z*-levels with partial steps models localising different types of terrain-following vertical coordinates in the proximity of the Greenland-Scotland ridge are developed and compared against a control. Experiments include a series of idealised and realistic numerical simulations where the skill of the models in computing pressure forces, reducing spurious diapycnal mixing and reproducing observed properties of the Nordic Seas overflows are assessed. Numerical results prove that the localisation approach proposed here can be successfully used to embed terrain-following levels in a global geopotential levels-based configuration, provided that the localised vertical coordinate chosen is flexible enough to allow a smooth transition between the two. In addition, our experiments show that deploying localised terrain-following coordinates via the multi-envelope method allows the crucial reduction of spurious cross-isopycnal mixing when modelling bottom intensified buoyancy driven currents, significantly improving the realism of the Nordic Seas overflows simulations in comparison to the other models. Important hydrographic biases are found to similarly affect all the realistic experiments and a discussion on how their interaction with the type of localised vertical coordinate affects the accuracy of the simulated overflows is provided.

Melt rates at the base of Antarctic ice shelves are needed to drive projections of the Antarctic ice sheet mass loss. Current basal melt parameterisations struggle to link open ocean properties to ice-shelf basal melt rates for the range of current sub-shelf cavity geometries around Antarctica. We present a novel parameterisation based on deep learning. With a simple feedforward neural network, or multilayer perceptron, acting on each grid cell separately, we emulate the behavior of circum-Antarctic cavity-resolving ocean simulations. We explore different neural network sizes and find that, in all cases containing at least one hidden layer, this kind of emulator produces reasonable basal melt rates for our training ensemble, closer to the reference simulation than traditional parameterisations. For testing, we use an independent ensemble of simulations that was produced with the same ocean model but with different model parameters, different cavity geometries and different forcing. In this challenging test, traditional and neural network parameterisations yield similar results on present conditions. In much warmer conditions than the training ensemble, both traditional parameterisations and neural networks struggle, but the neural networks tend to produce basal melt rates closer to the reference than a majority of traditional parameterisations. These neural networks are therefore suitable for century-scale Antarctic ice-sheet projections.

Antarctic Ice Sheet projections show the highest sensitivity to increased basal melting in the Amundsen Sea. However, little is known about the processes that control future increase in melt rates. We build an ensemble of three ocean–sea‐ice–ice‐shelf simulations for both the recent decades and the late 21st century, constrained by regional atmosphere simulations and the multi‐model mean climate change of the fifth Climate Model Intercomparison Project under the RCP8.5 scenario. The ice‐shelf melt rates are typically multiplied by 1.4–2.2 from present day to future, for a total basal mass loss increased by 346 Gt yr⁻¹ on average. This is equally explained by advection of warmer water from remote locations and regional changes in Ekman downwelling and in the ice‐shelf melt‐induced circulation, while increased iceberg melt plays no significant role. Our simulations suggest that high‐end melt projections previously used to constrain recent sea level projections may have been significantly overestimated.