Future Evolution of the Amundsen Sea Embayment, West Antarctica: Exploring the Modelled Ice Stream Sensitivity to Numerical Representation

The Amundsen Sea Embayment (ASE) is among the most dynamic catchments in Antarctica. With a recent history of increasing mass loss, and the potential to raise global sea levels by 1.2 m, the evolution of ice streams in the ASE in response to a changing climate must be understood. Sub-ice shelf (IS) basal melting of ASE ice streams is projected to increase in the future in response to climate change, which could initiate unstable retreat in the region. Ice sheet models are used to numerically represent ice flow and simulate future evolution of ice sheets in response to climate. Representation of processes, boundary conditions, model initialisation and the input datasets are at the discretion of model users. However, numerical representation impacts the modelled behaviour of ice streams and can lead to uncertainty in future projections. Understanding the interplay between uncertainty in external forcing and internal ice sheet model configuration is important for constraining future estimates of ASE sea level contribution. This thesis explores the future evolution of ASE ice streams with a focus on the sensitivity of ice streams to IS melting and the role of numerical representation in altering the sensitivity of ice streams to future forcing. Here, century scale simulations are performed on a regional ASE domain with the high resolution adaptive mesh refinement model, BISICLES. BISICLES captures both shear-dominated flow of grounded ice and longitudinal stress-dominated flow of floating ice, while resolving the grounding line to 250 m, meaning the fast evolving ice streams in the ASE are well represented. Experiments use perturbations of IS melting to force the ice sheet over 200 years. Different representations of subglacial rheology (sliding law), calving front position, geometry product, choice of ice sheet model and parameterization of IS melting are explored for simulations with varying future IS melt rates. Following a multi-decadal period of elevated IS melting, a sustained reduction in melting to below present-day rates can limit the total mass loss from the ASE and drive regrowth of ice streams through grounding line advance. Pine Island Glacier (PIG) is more sensitive to the choice of sliding law than to the migration of a calving front, however, sliding law and ice front are most important during application of positive forcing. Thwaites Glacier is less sensitive to the sliding law than PIG, but is more sensitive to the bed geometry product used, particularly in inducing acceleration consistent with observations. Differences in the parameterization of IS melt over the 21 st century leads to a doubling in projected mass loss for a single future climate scenario. The results of a model comparison demonstrate the choice of ice flow model is less important than the model initialisation, which amplifies model differences in response to perturbed IS melting. The exploration presented here into the role of numerical representation provides context for the interpretation of existing ASE simulations and also helps clarify which model configuration(s) are best able to reduce sea-level projection uncertainty.