Sensitivity of the Lambert-Amery glacial system to ice sheet model boundary conditions

The Lambert-Amery Glacial System (LAGS) is a major drainage basin in East Antarctica, and is one of the largest glacial systems on Earth with a total area of ‚Äöv†¬¿ 60,000 km\\(^2\\) . The three largest glaciers within the system, the Lambert, Mellor and Fisher Glaciers, feed the Amery Ice Shelf (AIS). The AIS provides stability to the LAGS and contains some of the deepest ice in contact with the ocean in Antarctica at ‚Äöv†¬¿ 2500 m below sea level. The LAGS drains ice from far inland East Antarctica, and hence incremental changes in either ice shelf geometry or ocean temperature can impact significantly on the whole system. The dynamic response of the AIS to present day boundary perturbations or physical processes is not fully understood due to limited data availability for this difficult to access region. Exploration of the factors that control the dynamics of the AIS, in particular those that are captured by ice sheet models as boundary conditions, is therefore required to progress understanding of the probable future contribution of the LAGS to sea-level rise. The LAGS is a stable system over multi-decadal timescales, and therefore provides an opportunity for modelling studies, compared with observations, for the investigation of ice sheet model boundary conditions. In this thesis, model-based studies are undertaken to examine the impact of the choice of different bathymetries, ice shelf retreat scenarios, and basal melt models on the simulated ice dynamics of the AIS. The Parallel Ice Sheet Model (PISM, Version 0.7.3) was used for these numerical experiments: a three dimensional, thermomechanically-coupled hybrid model that superposes the shallow ice approximation and the shallow shelf approximation. Model resolution and workflows were chosen to carry out the aforementioned studies within the limitations of available computing resources. The geometry of the bedrock beneath a glacier system and its outlet ice shelf is understood to have a significant control on the evolution of that system. However, the depth of the bedrock in most locations under the AIS is not well-known. Tests of the sensitivity to bedrock geometry were carried out using PISM. The model was run using four bathymetry test cases, including one case with the addition of pinning points at the ice front that have previously been inferred from remote sensing data. It was found that the choice of bathymetry substantially impacts the modelled ice velocity across the grounding line, the ice velocity of the ice shelf, and the calving front position. The dynamic response to changes in the geometry of the bathymetry and pinning points that interact with the ice shelf base, even of small-scale, demonstrates the importance of high accuracy bathymetry data. A consequence of these findings is that under-sampled bathymetry can lead to undue emphasis on poorly constrained modelling parameters to reproduce the ice shelf extent. These findings also highlight the complex feedback between ice dynamics and AIS bathymetry, and hence the importance of bathymetry in future modelling of the LAGS. The AIS provides stability to the LAGS due to the effect of buttressing. If the calving front were to retreat from its present day geometry, the reduction in buttressing could cause acceleration of the calving front, and potentially impact ice dynamics at the deep grounding line. To quantify how a retreated ice shelf geometry impacts the stability of the AIS, different scenarios of retreat, including extreme retreat geometries, were imposed. Each scenario was set by following a threshold value in the along-flow strain rate, removing progressively more ice from the ice shelf front. Previously hypothesised criteria for ice shelf stability were explored for each retreat scenario based on the 2nd principal strain rate (the compressive arch‚ÄövÑvp hypothesis), and the angle between the 1st principal stress direction and the ice flow direction. Further experiments that explored calving laws were also carried out. The results of these individual investigations differed in detail, highlighting shortcomings in the current understanding of the dynamics of ice shelves, and indicating that additional exploration of stability criteria and calving laws is needed when identifying tipping points in ice shelf retreat. Taken together, the model results suggest that the AIS is stable until it retreats approximately up to 85 km upstream of the current calving front. Some sections of the ice shelf did not cause a speed-up of the ice flow at the calving front when further ice was removed. This area of passive ice on the AIS is larger than in previous studies, and this may be due to the inclusion of small-scale pinning points near the calving front which were included in the the above bathymetry study, but previously neglected. At some retreat positions, the models of the AIS show significant increase in velocity, which can propagate to the deep grounding line in the case of a calving front in the south of the AIS where the ice shelf embayment narrows. The basal melt rate of an ice sheet is influenced by ocean temperature, salinity and pressure, and since the deepest parts of the LAGS grounding line are up to ‚Äöv†¬¿ 2500 m below sea level, intrusion of Circumpolar Deep Water along with hydrostatic pressure generates large basal melt rates in the southern AIS grounding zone. An increase in melt rates for the AIS, potentially caused by an increase in ocean temperature, could drive retreat of the deep grounded ice and global sea level contribution from the LAGS. To understand the effect of different basal melt rate estimates on the evolution of the AIS, a range of ice-ocean interface model parameter inputs were investigated. The model of the AIS responded differently to these ice-ocean input parameter choices, and from a simple experiment it was clear that the internal basal melt models (PISM V. 0.7.3) could not reproduce the observed spatial pattern of melting and refreezing. Improvement, however, can be made by supplying PISM with a three-dimensional regional ocean model that incorporates spatial variation including the influence on ocean temperature of frazil ice precipitation and refreezing. This resulted in a realistic level of basal melt at the deep grounding line in the southern part of the AIS, and replicated the observed pattern of basal refreezing on the western side of the ice shelf. Basal melt at the grounding line and the pattern of refreezing influence ice dynamics, and cause the calving front to advance significantly compared to other ice-ocean interface model inputs. This result implies that studies that do not incorporate the needed boundary conditions from ocean models could over-estimate the future sea level contribution from the AIS as refreezing supports buttressing and provides stability to the calving front. The experiments carried out throughout this thesis show the importance to ice sheet modelling of comparing multiple input data sets that provide boundary conditions, and also assessment criteria and model choices to identify the possible causes of discrepancies between modelled results and observations. In summary, the findings demonstrate model sensitivity to small-scale bedrock features, and ice-ocean input data, for modelling ice dynamics. In further research involving modelling of the AIS, the LAGS, and other systems over long time-scales, the bathymetry study indicates that a change in ice shelf cavity shape would impact the ice sheet evolution, and therefore points to the need for coupled ice-ocean modelling. Our findings also point to the influence of small-scale bedrock features on the likely calving response of the ice shelf under different retreat scenarios. As a result of the research described in this thesis, it is recommended that efforts continue to improve bathymetry datasets, particularly in the regions upstream of present day calving fronts, and that ocean models are considered as input to studies using PISM, rather than relying on internal parameter selections. Incorporating these improvements should enable future modelling studies to better predict the response of ice sheet systems such as the LAGS to external forcings, and thus improve estimates of future ice mass loss and consequent contributions to sea-level rise.