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.

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 model 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 indicate that more than half of the iceberg melting is discharged over the Antarctic continental shelf, reaching an average of 631 Gt yr-1 if the interaction of icebergs with bathymetry is neglected, and 802 Gt yr-1 if a simple grounding scheme is implemented. Such additional freshwater injected into regions prone to sea ice production has important consequences for the ocean properties around Antarctica. This freshwater enhances the ocean stratification, limiting deep convection and favoring the intrusion of relatively warm Circumpolar Deep Water onto some parts of the continental shelf. Moreover, the simulated distribution of grounded icebergs blocks drifting sea ice, thereby promoting areas 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.