The ‘Mystery Interval’ (17.5–14.5 ka BP) is characterized by a large decline in atmospheric Δ14C synchronous with an increase in atmospheric CO2. The most widely accepted hypothesis to explain these observed shifts involves the existence of an isolated ‘old’ ocean carbon reservoir that was subsequently ventilated. Here we use the UVic Earth System Climate Model to locate a potential carbon rich and Δ14C depleted water mass under 17.5 ka BP boundary conditions. We then investigate two mechanisms for the potential ventilation of such a reservoir, namely the weakening of the North Atlantic Meridional Overturning due to iceberg calving and latitudinal shifts in Southern Hemisphere Westerlies (SHW) due to southern hemispheric warming. We find that simulations derived from an equilibrium state forced with present‐day SHW and moderate North Atlantic Deep Water (NADW) formation are in better agreement with atmospheric and ocean Δ14C reconstructions than simulations derived from an equilibrium state forced with a northward shifted SHW belt resulting in a shut‐down of the Atlantic Meridional Overturning and formation of North Pacific Deep Water. For simulations with present‐day SHW, the oldest water masses are found in the North Pacific, although the Southern Ocean cannot be ruled out as a potential ‘Mystery Reservoir’. According to our simulations, the strength of Atlantic overturning is the dominant mechanism in increasing the ocean‐atmosphere carbon flux, while shifting SHW results in a rearrangement of deep ocean carbon largely between the Atlantic and Pacific basins. In our ‘best case’ scenario, the model can account for 58% of the atmospheric CO2 increase and 48% of the atmospheric Δ14C decline. While the rate of ventilation and the age of ventilated water masses are comparable with observations, the ventilation in the model could not be sustained long enough to account for the full excursion seen in paleodata.

Abstract. The past and future evolution of the Antarctic Ice Sheet is largely controlled by interactions between the ocean and floating ice shelves. To investigate these interactions, coupled ocean and ice sheet model configurations are required. Previous modelling studies have mostly relied on high resolution configurations, limiting these studies to individual glaciers or regions over short time scales of decades to a few centuries. We present a framework to couple the dynamic ice sheet model PISM with the global ocean general circulation model MOM5 via the ice-shelf cavity module PICO. Since ice-shelf cavities are not resolved by MOM5, but parameterized with the box model PICO, the framework allows the ice sheet and ocean model to be run at resolution of 16 km and 3 degree, respectively. This approach makes the coupled configuration a useful tool for the analysis of interactions between the entire Antarctic Ice Sheet and the Earth system over time spans on the order of centuries to millennia. In this study we describe the technical implementation of this coupling framework: sub-shelf melting in the ice sheet model is calculated by PICO from modeled ocean temperatures and salinities at the depth of the continental shelf and, vice versa, the resulting mass and energy fluxes from the melting at the ice-ocean interface are transferred to the ocean model. Mass and energy fluxes are shown to be conserved to machine precision across the considered model domains. The implementation is computationally efficient as it introduces only minimal overhead. The framework deals with heterogeneous spatial grid geometries, varying grid resolutions and time scales between the ice and ocean model in a generic way, and can thus be adopted to a wide range of model setups.

No abstract

Abstract. The past and future evolution of the Antarctic Ice Sheet is largely controlled by interactions between the ocean and floating ice shelves. To investigate these interactions, coupled ocean and ice sheet model configurations are required. Previous modelling studies have mostly relied on high-resolution configurations, limiting these studies to individual glaciers or regions over short timescales of decades to a few centuries. We present a framework to couple the dynamic ice sheet model PISM (Parallel Ice Sheet Model) with the global ocean general circulation model MOM5 (Modular Ocean Model) via the ice shelf cavity model PICO (Potsdam Ice-shelf Cavity mOdel). As ice shelf cavities are not resolved by MOM5 but are parameterized with the PICO box model, the framework allows the ice sheet and ocean components to be run at resolutions of 16 km and 3∘ respectively. This approach makes the coupled configuration a useful tool for the analysis of interactions between the Antarctic Ice Sheet and the global ocean over time spans of the order of centuries to millennia. In this study, we describe the technical implementation of this coupling framework: sub-shelf melting in the ice sheet component is calculated by PICO from modelled ocean temperatures and salinities at the depth of the continental shelf, and, vice versa, the resulting mass and energy fluxes from melting at the ice–ocean interface are transferred to the ocean component. Mass and energy fluxes are shown to be conserved to machine precision across the considered component domains. The implementation is computationally efficient as it introduces only minimal overhead. Furthermore, the coupled model is evaluated in a 4000 year simulation under constant present-day climate forcing and is found to be stable with respect to the ocean and ice sheet spin-up states. The framework deals with heterogeneous spatial grid geometries, varying grid resolutions, and timescales between the ice and ocean component in a generic way; thus, it can be adopted to a wide range of model set-ups.