The Antarctic Ice Sheet (AIS) is the largest potential source of future sea-level rise (SLR), but predictions of its behavior vary widely and are sensitive to poorly constrained parameters, forcing, and choices of model physics. We present a sensitivity study of the AIS under various climate forcings to 2300 using the higher-order, three-dimensional MPAS-Albany Land Ice model on a 4–20 km resolution mesh. We initialize our model in the year 2000 using an adjoint optimization that solves for basal friction and ice stiffness fields that minimize the misfit with observed velocities. Our baseline ensemble of simulations follows the ISMIP6 AIS 2300 protocol. The baseline ensemble uses a semi-plastic Weertman-style sliding law with an exponent of 1 ⁄ 5, a non-local quadratic sub-shelf melt parameterization, and thermomechanical coupling. Ice advected beyond the initial ice extent is calved away, but no other calving is applied.

Our sensitivity experiments use CCSM4 and HadGEM2 RCP8.5 climate forcing to 2300. In our baseline ensemble, these resulted in 0.8 m and 2.9 m SLR by 2300, respectively. We examine:

1. Sub-shelf melt, with low and high sensitivity to ocean thermal forcing
2. Bed rheology, with exponents of 1 (linear viscous) and 1/10 (effectively plastic)
3. Thermal coupling, using a static temperature field.

For the weaker CCSM4 climate forcing, our low and high basal melt simulations predict -40% and +35% SLR relative to the baseline simulation, respectively; for the stronger HadGEM2 forcing, they predict ±13%. The linear viscous bed reduces SLR by 72% and 49%, while the plastic bed increases SLR by 9% and 14%. However, the onset of rapid grounding-line retreat in the Amundsen Sea is delayed by ~70–180 years by the more-plastic beds, relative to the linear viscous bed. Finally, a simulation under the HadGEM2 climate forcing with a fixed ice temperature field yields 15% more SLR than its counterpart with thermomechanical coupling, indicating that advection of cold, stiff ice from the ice sheet interior may strongly regulate the rate of marine ice sheet retreat.

These results illustrate the profound impacts that common assumptions regarding ice-shelf melt, basal friction, and thermal coupling have on predicted SLR.