Modelling the Hydro-fracture driven collapse of the Larsen B ice shelf

Ice shelves play a key role in buttressing upstream ice - modulating the flow of grounded ice into the ocean and in turn affecting ice sheet contribution to sea level. Iceberg calving, which has approximately equalled thinning in terms of mass loss from Antarctic ice shelves since 2007, is an important ablation process for balancing accumulation. However, rapid ice shelf disintegration, driven by surface melting, can dramatically impact grounded ice dynamics on relatively short time scales. In 2002, the Larsen B ice shelf lost ~60% of its area, following extensive surface melt ponding, observed over months.  This event has been associated with a ‘hydro-fracture’ mode of ice shelf collapse, where surface melt ponds enhance surface crevasse penetration, causing the ice shelf to disintegrate. Following the collapse of Larsen B, retreat of its largest tributary glacier increased by >50% over two years. Whilst surface melting on the scale that preceded Larsen B collapse has historically been limited to the North of the Antarctic Peninsula, under future anthropogenic warming, more of Antarctica’s ice shelf area could become vulnerable to hydro-fracture.

To quantify the role of ice shelf hydro-fracture in ice sheet response to warming, the PSU ice sheet model (PSUISM) incorporates a simple parameterisation of this process, as well as ice cliff failure following loss of buttressing. With its computationally tractable hydro-fracture parameterisation, PSUISM has been used to reproduce large long term Antarctic mass loss under periods of past warmth. It has also simulated high Antarctic contribution to future sea level. However, the break-up of Larsen B, which provides the observational basis for its hydro-fracture scheme, has been less well explored in PSUISM.

We present a suite of high-resolution simulations of the Larsen B ice shelf and its tributary glaciers. We explore the role of hydro-fracture parameters and a range of climate boundary conditions in driving ice shelf collapse. We also compare modelled ice shelf retreat, and grounded ice response, to available observational data. Finally, we explore modifications to the simple hydro-fracture scheme that can better capture Larsen B shelf collapse.

Ice shelf processes remain a key challenge in predicting future Antarctic ice sheet retreat. Despite advances in ice sheet modelling, capturing hydro-fracture in models capable of long integration times, at high resolution, remains a challenge. Our work explores how well the current approach in PSUISM captures the best observed ‘hydro-fracture’ driven ice shelf collapse, and how that impacts our understanding of existing projections.