Sub-shelf melt pattern and ice sheet mass loss governed by meltwater flow below ice shelves

Abstract. Ocean-induced sub-shelf melt is one of the main drivers of Antarctic mass loss. Capturing it in ice sheet models is typically done by using parameterisations that compute sub-shelf melt rates based on local thermal forcing. However, these parameterisations do not resolve the 2D horizontal flow of the meltwater layer, either neglecting it entirely or simplifying its representation. In this study, we present a coupled setup between the ice sheet model IMAU-ICE and the sub-shelf melt model LADDIE. LADDIE resolves 2D horizontal meltwater flow beneath an ice shelf, incorporating both topographic steering and Coriolis deflection of meltwater plumes. We conduct simulations in a framework closely resembling the Marine Ice Sheet Model Intercomparison Project third phase (MISMIP+), which represents an idealised ice sheet-shelf system. Simulations using LADDIE to calculate melt rates reveal key differences compared to simulations using the widely adopted sub-shelf melt parameterisations. These differences primarily emerge as variations in timing, location, and persistence of strong melting, leading to distinct transient volume loss. In the LADDIE experiments, a stepwise increase in ocean temperatures induces an initial steepening of the ice draft near the grounding line. This strengthens the westward flow, which converges into a western boundary channel, leading to persistent strong melting along the western margin. Consequently, western margin thinning results in reduced buttressing and strong volume loss over the first period of the simulations. Over longer timescales, a weakened meltwater flow circulation due to reduced thermal forcing at the grounding line allows the western margin to thicken again, suppressing volume loss. In contrast, the parameterisation's limitations in representing the 2D horizontal meltwater flow prevent these experiments from capturing the influence of ice draft steepening on enhanced margin melt. This results in a different transient volume loss compared to LADDIE. The parameterisations either inherently overestimate the persistence of margin thinning, leading to a sustained strong volume loss, or they underestimate margin thinning, delaying the onset of strong volume loss. Our findings suggest that incorporating the more detailed melt patterns resulting from the 2D horizontal meltwater flow in ice sheet models could significantly alter projections of Antarctic ice sheet evolution compared to melt patterns computed by the currently more common parameterisations.