Differential SAR interferometry-based, Machine Learning-supported delineation of grounding zones around Antarctica.

Abstract:

The grounding line of a glacier is a transition area where ice motion shifts from basal-drag controlled to drag-free, side shear controlled and also represents the ideal location to estimate ice fluxes into the ocean. The location of the grounding line is not static, it retreats or advances depending on the state of mass balance of the glacier which, in combination with bed slope, is a crucial parameter for glacier evolution. Furthermore, recent research shows that this boundary is also impacted by changes in oceanic tides. The minima and maxima of the short-term tide-based modulation defines the "grounding zone", i.e. an area over which seawater accesses glacier ice and may significantly affect basal resistance to flow. Differential SAR interferometry is a most accurate method for grounding line measurements and the basis for the MEaSUREs grounding line product in Antarctica. The availability in regular acquisitions in coastal Antarctica using Sentinel-1a/b allows the delineation of grounding lines multiple times per year versus one or two per decade in the past. The huge increase in data availability creates difficulties in digitizing the grounding lines manually. We use a machine learning approach to handle this mapping. We establish a workflow with minimal human interaction to do this continent wide following an evaluation of the approach on one large basin. We present grounding zone information for the years 2018 and 2019. We compare the grounding zone width with that expected from interactions on a hard bed or a visco-elastic bed and highlights sectors where the grounding zone is particularly wide and wider than expected. Grounding zones product will be derived from this analysis and made available as MEaSUREs products to assist mass balance assessment and ice sheet modeling activities.