Glacier dynamics, ice mass unloading and bedrock response in the southern Antarctic Peninsula

Rapid regional warming in the Antarctic Peninsula led to significant retreat and eventual collapse of several major ice shelves since the 1970s, followed by the subsequent acceleration and thinning of their feeding glaciers. The Wordie Ice Shelf, lying off the west coast of the southern Antarctic Peninsula, has undergone long-term disintegration since the 1960s with a substantial calving event occurring around 1989, followed by continuous steady retreat and its almost-complete disappearance by 2008. The dynamic response of the upstream glaciers to the ice shelf collapse and the response of the solid Earth to the associated mass loss are not fully understood. To quantify the mass loss from the catchment region of the Wordie Ice Shelf, a digital elevation model (DEM) was generated using airborne vertical and oblique imagery from 1966 and compared to a DEM derived from 2008 SPOT satellite data. This analysis reveals lowering over that time of approximately 60 m at the front of Fleming Glacier, the major glacier feeding the former ice shelf. Using IceBridge and ICESat-1/GLAS data spanning 2002-2014, the mean elevation change rate was estimated. The rates post-2008 (-6.25¬±0.20 m yr\\(^{-1}\\)) were more than twice those of 2002-2008 (-2.77¬±0.89 m yr\\(^{-1}\\)) near the ice front. These data quantify the change in mass load that is subsequently used as a basis for the simulation of viscoelastic solid Earth deformation. To infer properties of Earth rheology, modelled elastic deformation rates, and a suite of modelled viscous rates, were subtracted from GPS-derived three-dimensional bedrock velocities at GPS sites to the south of Fleming Glacier. Assuming the pre-breakup bedrock uplift was positive due to post-Last Glacial Maximum (LGM) ice retreat, the derived viscoelastic-corrected GPS uplift rates suggest upper mantle viscosities are > 2√v=10\\(^{19}\\) Pa s and likely >1√v=10\\(^{20}\\) Pa s in this region, 1-2 orders of magnitude greater than previously found ~500 km further north in the Antarctic Peninsula. After the application of elastic and plate-tectonic corrections, horizontal bedrock velocities at the GPS site nearest the Fleming Glacier, point away from Marguerite Bay rather than away from the present glacier front. This suggests that horizontal bedrock motion in the region reflects the earlier retreat of the glacier system following the LGM, compatible with a relatively strong mantle in this region. These findings highlight the need for improved understanding of ice load changes in this region through the late Holocene in order to accurately model present-day glacial isostatic adjustment. The observed thinning of Fleming Glacier over five decades represents an opportunity to further understand glacier responses to ice shelf disintegration. Understanding of the dynamics of fast-flowing glaciers such as the Fleming Glacier, and their potential future behavior, can be improved through ice sheet modelling studies. Here, the Stokes model Elmer/Ice was used to simulate the Wordie Ice Shelf-Fleming Glacier system. Inverse methods are commonly used in ice sheet models to infer the spatial distribution of a basal friction coefficient, which has a large effect on the basal velocity and ice deformation. Using an inverse method in Elmer/Ice, the basal friction coefficients were inferred from the surface velocities observed in 2008. A multi-cycle spin-up scheme was developed to reduce the influence of initial temperature assumptions on the final inversion. This is particularly important for glaciers like the Fleming Glacier, which have areas of strongly temperature-dependent, deformational flow in the fast-flowing regions. Sensitivity tests using various bed elevation datasets, ice front positions and boundary conditions demonstrate the importance of high-accuracy ice thickness/bed geometry data and precise location of the ice front boundary. Recent observational studies have suggested the 2008-2015 velocity change and the dynamic thinning of the Fleming system was due to the ungrounding of the Fleming Glacier front. It is important to know whether the substantial additional speed-up and surface draw-down of the glacier since 2008 is a direct response to increasing ocean forcing or driven by the feedback within the grounded glacier system or a combination of both. To explore the mechanism underlying the changes, the Elmer/Ice model was used to simulate the basal shear stress of the Fleming system in 2008 and 2015. High-resolution modelling reveals that the ungrounding process of the Fleming Glacier might not have started in Jan 2008, which is consistent with a height above buoyancy calculation for 2008. Comparison of the inversions for basal shear stresses for 2008 and 2015 suggests upstream migration of the grounding line by ~9 km by 2015, while the 2008 ice front/grounding line positions virtually coincided with the 1996 grounding line position. This shift is consistent with the change in floating area deduced from the height above buoyancy in 2015. The retrograde bed underneath the Fleming Glacier has likely promoted the significant migration of its grounding line. The increase in basal sliding and grounding line retreat might be caused by increased subglacial water volume and/or pressure through greater frictional heating at the bed further upstream in the fast-flowing region as a result of acceleration. Improved knowledge of bed topography near the grounding line and further transient simulations with oceanic forcing are required to predict accurately the future movement of the Fleming system grounding line and better understand its ice dynamics and contribution of freshwater flux into the ocean. The Fleming system now represents one of the best observed examples of multi-decadal glacier change following ice shelf disintegration; combination of various observational datasets (including those presented here) with the new Stokes model provides a much-improved capability to understand the future of this glacier system.