Subglacial topography and basal conditions form critical controls on ice-sheet dynamics and ice-flow pathways. These controls modulate basal shear stress, affect grounding-line stability and allow the potential for marine ice-sheet instability. Deep troughs and reverse slopes facilitate rapid retreat driven by ocean warming, while topographic ridges and bumps can anchor ice margins. However, substantial data gaps in ice-sheet bed measurements limit the accuracy of sea-level rise projections from numerical ice-sheet models that require such information as inputs. Interpolation techniques often smooth over key features, creating digital elevation models (DEMs) that do not replicate 'real' glacierized systems. This causes uncertainty in simulations of ice-sheet evolution. Advances in physics-informed methods, which use surface velocity and mass conservation to infer bed elevation, have improved reconstructions of bed topography. Nonetheless, important details that characterize glacierized surfaces remain to be resolved in the DEMs and bed topography grids that models rely on. Future priorities to improve these data products involve dense, targeted surveys in key areas such as grounding zones. Machine learning (ML) offers promising tools for optimizing interpolation, prioritizing survey targets and planning future surveys. As ice-sheet model projections extend to 2300 CE and beyond, survey strategies must anticipate migrating grounding lines. Automation and repeat observations, including swath radar and unmanned aerial vehicle (UAV)-based campaigns, will be vital for maintaining up-to-date, high-resolution bed datasets. Ultimately, significant advancements in subglacial mapping are possible within the next 10-20 years, which could greatly improve model accuracy and better inform sea-level rise mitigation and adaptation strategies. This article is part of the Theo Murphy meeting issue 'Next generation ice-sheet bed measurements'.
Millman et al. (Thu,) studied this question.