Li, Lu1, Aitken, Dr Alan1, Lindsay, Dr Mark1, Jessell, Dr Mark1
1The School of Earth Sciences, The University Of Western Australia, Perth WA, Australia
Antarctica preserves the largest ice sheet in the world, which has a potential contribution to future sea-level rise up to 60m. Understanding subglacial sedimentary basin distribution is essential for studying ice sheet behaviour, as it forms an important basal boundary condition for ice sheet dynamics. It also records the geological history for tectonic evolution and past ice sheet behaviour. However, the subglacial sedimentary basin distribution is poorly known in Antarctica. A map of sedimentary basin distribution is a prerequisite to improving understanding of current and past ice sheet behaviour, aiding to project future ice sheet change and sea-level rise.
In this study, we present a sedimentary basin distribution likelihood map for Antarctica using the supervised machine learning method Random Forest. We apply this to generate a model based on the current understanding of Antarctica bedrock type distribution. We label the sedimentary basin and crystalline basement distribution driven from sparse rock outcrops, seismic imaging, and potential field data interpretation. Evidence layers are chosen from the available continental-scale geophysical datasets. By applying variable importance selection, we remove the unimportant and highly correlated evidence layers. After that, a strata sample process ensured a balanced class of each bedrock type during the training process. A sedimentary basin distribution map is then spatially predicted. The model accuracy is evaluated based on block cross-validation to overcome the underestimate of prediction error in the spatial correlated geophysical data.
Our results confirm the existence of previously documented subglacial sedimentary basins in Antarctica, and in general define the margins and extents of sedimentary basins in more detail. Specifically, in West Antarctica Rift System, the model delimits boundaries between sedimentary basins and volcanic rocks. We find a potential sedimentary basin preserved in Byrd Subglacial Basin, which is highly likely contributing to the currently fast flow in Thwaites Glacier. Further, our model shows more widely distributed subglacial sedimentary basins in East Antarctica than been previously recognized. Properties of geophysical and remote sensing data in Recovery Glaciers suggest a high probability of sedimentary basin preservation.
Lu Li is a PhD student at The University Of Western Australia to study geophysics. His PhD project is using geophysics to understanding lithosphere structure of Antarctica, and its influence on cryosphere/solid-earth interaction.