Basal melting of ice shelves is fundamental to Antarctic Ice Sheet mass loss, yet direct observations are sparse. We present the first melt record (2017 to 2021) from a phase-sensitive radar at Fimbulisen, East Antarctica, one of the fastest flowing ice shelves in Dronning Maud Land. The observed long-term mean ablation below the central part of the ice shelf was 1.0 ±0.4 m yr–1, marked by substantial sub-weekly variability ranging from 0.3 to 3.8 m yr–1. 36-h filtered fluctuations in basal melt exhibit a close alignment with ocean velocity, revealing shear-driven turbulent heat transfer as the predominant driver of melt variability at sub-weekly to monthly timescale. Seasonally, basal melt rates are highest in the austral summer, when ocean temperature is higher. Our observed in-situ melt rates show threefold lower amplitudes and a 3-month delay in seasonality compared to satellite-derived melt rates, however, the long-term multi-year mean is of similar magnitude (1.0 m yr–1 vs 0.8 m yr–1). Our detailed ice–ocean observations provide essential validation data for remote sensing and numerical models aiming to measure and project ice-shelf response to ocean forcing. In-situ measurements and continued monitoring are crucial for accurately assessing and modelling future basal melt rates, as well as understanding the complex dynamics driving ice-shelf stability and sea-level change.

Dome Fuji, inland East Antarctica is one of only few regions where 1.5-Ma old ice can be preserved for investigating the mid-Pleistocene Transition. We used stochastic simulation and various radar datasets to generate a bed topography ensemble with the continuous, realistic roughness necessary to assess basal conditions. Ensemble analysis reveals the magnitude and spatial distribution of topographic uncertainty, facilitating uncertainty-constrained assessments of subglacial drainage and topographic adjustments to geothermal heat flow. We find that topographic variability can lead to widespread local geothermal heat flow variations of ± 20% the background value, which aggregate to raise the regional value and suggest previously underestimated distributions and rates of basal melting. We also find that survey profile spacing has an increasing influence on topographic uncertainty for rougher bed, deriving an empirical relationship that could guide future survey planning based on uncertainty tolerance.