Mapping high-resolution basal topography of West Antarctica from radar data using non-stationary multiple-point geostatistics (MPS-BedMappingV1)

Abstract. The subglacial bed topography is critical for modelling the evolution of Thwaites Glacier in Amundsen Sea Embayment (ASE), where rapid ice loss threatens stability West Antarctic Ice Sheet. However, mapping subject to uncertainties up hundreds metres, primarily due large gaps tens kilometres airborne ice-penetrating radar flight lines. Deterministic interpolation approaches do not reflect such spatial uncertainty. While traditional geostatistical simulations can model uncertainty, they become difficult apply because significant non-stationary variation over surface area. In this study, we develop a multiple-point (MPS) approach interpolate areas with irregular geophysical data and it uncertainty entire ASE basal topography. We collect 166 high-quality topographic training images (TIs) resolution 500 m train gap-filling gaps, thereby simulating realistic maps. TIs are extensively sampled from deglaciated regions Arctic as well Antarctica. To address non-stationarity modelling, introduce Bayesian framework that models posterior distribution assigned local line data. Sampling then provides candidate constrained Compared MPS consider uncertain TI sampling, our results improvement quality efficiency simulation algorithm. Finally, simulate multiple realizations high-resolution use investigate impact on hydrological flow patterns.