Ocean-driven basal melting of Antarctic ice shelves is a major contributor to sea level rise, yet it remains difficult to observe directly due to the inaccessibility of sub-shelf cavities. As a result, basal mass balance (BMB) is typically inferred indirectly from sparse remote sensing observations, but existing methods are limited by coarse fixed grids, numerical approximations, and uncertain ice thickness estimates.

We present FluxNet, a physics-informed deep learning framework that reconstructs continuous fields of BMB from sparse ice shelf thickness and satellite velocity observations. FluxNet learns ice flux with a neural network architecture designed to represent flow fields through a Helmholtz decomposition, separating components that do and do not contribute to flux divergence. This enables direct estimation of BMB from the learned flux field using automatic differentiation, avoiding numerical differencing and resolution constraints.

We apply FluxNet to the Ross Ice Shelf, leveraging 3.2 million Bedmap3 thickness observations. In reconstruction experiments, FluxNet outperforms physics-agnostic neural networks and conventional geospatial interpolation methods, reducing RMSE by 11.7% and 31.4% respectively. An ensemble of FluxNet models resolves enhanced basal melting near grounding lines and ice shelf margins, successfully mitigating satellite striping artefacts while quantifying predictive uncertainty. Overall, FluxNet offers a fast, physics-grounded, and data-driven complement to existing approaches for basal melt estimation.

Subglacial hydrological discharge is a potentially important but poorly constrained source of freshwater to the circum-Antarctic ocean system. Idealised and regional studies have demonstrated that subglacial discharge can strongly impact ice shelf-ocean interactions, including substantial enhancement of basal melting. However, its integrated impact on Southern Ocean circulation, shelf water mass transformation, and ice shelf basal melting at the continental scale remains unknown.

We present simulations forcing a whole-Antarctic ice shelf-ocean model configuration (WAOM; Whole Antarctic Ocean Model) with a continental-scale subglacial hydrology model (GLADS; Glacier Drainage System model). This enables, for the first time, an investigation of the circum-Antarctic impacts of subglacial freshwater discharge. Initial results demonstrate the first-order sensitivity of circum-Antarctic melt rates and shelf water mass properties to realistic subglacial hydrological forcing. The grounding line proximity of the maximum perturbation caused by outflow on melt rates is of particular interest, as marine ice sheets are far more sensitive to melt increases close to the grounding line than further downstream along the shelf. These results provide new insight into the role of subglacial freshwater in shaping ice-ocean interactions and motivate improved freshwater parameterisations in large-scale climate and ice-ocean models, with implications for projections of Antarctic contribution to sea level rise.

Ice shelf basal melting is characterised by turbulent processes that couple the broader ocean circulation to the ice-ocean interface. While the impact of subglacial discharge on circulation and basal melting has been increasingly investigated, recent work has demonstrated the influence of sediment-laden subglacial discharge, sediment transport, and evolving seabed morphology on the turbulent ice-ocean exchange. However, the role of sediment-driven morphodynamics in realistic ice-shelf cavities remains poorly constrained. In particular, the relative contributions of cavity-scale sediment redistribution and localised sediment-laden subglacial discharge to circulation, melting, and seabed evolution remain unquantified. Here, we investigate coupled ice shelf-ocean-sediment processes beneath the Amery Ice Shelf using a high-fidelity numerical model that resolves sediment erosion, transport, deposition, and bathymetric evolution. Sensitivity experiments are used to isolate the effects of sediment and freshwater forcing on basal melt across spatial scales. Preliminary results suggest that cavity-scale currents dominate sediment redistribution and long-term morphodynamic changes, while sediment-laden subglacial discharge primarily affects circulation and melt rates near the grounding zone. By comparing sensitivity experiments, we assess the relative importance of sediment dynamics and freshwater forcing in controlling basal melting, cavity circulation, and seabed evolution beneath the Amery Ice Shelf. These results provide new insight into how sediment-driven morphodynamics influence ice-ocean interactions and investigate the importance of representing sediment processes in ice shelf-ocean models.

Basal melting of Antarctica's ice shelves plays a key role in modulating the mass balance of the Antarctic Ice Sheet. While satellite observations provide a broad view of average melt rates, ground-base phase-sensitive radar (ApRES) observations provide increased accuracy and temporal resolution at a local scale. By quantifying temporal variability to sub-daily time scales these observations can provide insights into the oceanographic drivers of melting.

Here we describe lessons learned and selected results from a network of 12 ApRES instruments deployed on the north-western Ross Ice Shelf in 2022. The observations show strong seasonal and spatial melt rate variability, especially near the ice front where warm water enters the cavity in summer. While melting is the dominant process, the records indicate brief periods of freezing at most sites during winter. Concurrent oceanographic records from near the ice front are used to examine drivers of the most rapid melting observed.

To date the network has been serviced on a 2-yearly interval, and we briefly describe the practical issues associated with maintaining the instruments through multi-year deployments.

Accurately predicting basal melt rates remains challenging due to limited understanding of the ice shelf–ocean boundary layer (ISOBL), a meters-thick turbulent layer that exchanges heat and salt between the ocean and the ice. In large-scale ice–ocean models, basal melt rates are estimated using parameterizations that generally assume a well-mixed, shear-controlled regime. However, recent studies have shown that such parameterizations can substantially overestimate observed melt rates in buoyancy-controlled regimes. In these regimes, stratification due to meltwater buoyancy and double-diffusive processes play a dominant role in governing boundary-layer dynamics.
Despite some recent progress, accurate parameterizations for different melting regimes, as well as thresholds for the transitions between shear- and buoyancy-dominated melting, remain poorly constrained. To address this gap, we will present high-resolution three-dimensional large-eddy simulations (LES) of the ISOBL beneath a horizontal ice shelf using Oceananigans. Basal melting will be quantified across a range of ocean velocities and temperatures, and compared with existing observations and parameterizations. Regime transitions will be identified using nondimensional parameters, including viscous Obukhov scale, which characterizes the effects of buoyancy or viscous forces felt by ISOBL turbulence. We will also present a regime diagram indicating expected melting responses around Antarctica and provide improved parameterizations.