Systematic comparison of calving rate parametrisations in an ice-sheet dynamical simulation of the Antarctic Ice Sheet

In recent years, a number of calving parametrisations have been derived and evaluated in simulations on the scale of individual glaciers. However, these new calving parametrisations have not yet been systematically tested in a continental-scale simulation of an ice sheet. We consider four rate-based calving parametrisations of which two match current observations from calving glaciers in Alaska (Mercenier et al. 2018) and the Antarctic Peninsula (Parsons et al. 2025), while the other two are designed as cliff calving parametrisations (Schlemm & Levermann 2019, Crawford et al. 2021) and give non-zero calving rates only for cliff heights larger than currently observed in calving glaciers.
We evaluate these calving parametrisations in an ice-sheet dynamical simulation of the Antarctic ice sheet using the Parallel Ice Sheet Model (PISM). Starting from an ISMIP6 present-day initial state at 8 km resolution, we apply each parametrisation under SSP5-8.5 forcing.
Already at the beginning of the simulation, when forcing is still similar to present day, parametrisations that match observations from current calving glaciers in Alaska and the Antarctic Peninsula produce spurious terminus retreat at locations where no retreat is observed so far. This is due to high initial cliff heights (>100m) along the coast in the ISMIP6 initial state.
In contrast, cliff calving laws, which include a critical cliff height threshold (below which no calving occurs) and produce smaller calving rates, are more conservative and better suited for continental-scale applications, despite not matching observations from current glaciers. However, even these more conservative laws can produce calving in locations where it's not expected. Better results are achieved, when calving of grounded ice is restricted to the Amundsen, Amery, Ross and Ronne-Filchner basins.