Deglacial evolution of regional Antarctic climate and Southern Ocean conditions in transient climate simulations

Abstract

Constraining Antarctica's climate evolution since the end of the Last Glacial Maximum (∼18 ka) remains a key challenge, but is important for accurately projecting future changes in Antarctic ice sheet mass balance. Here we perform a spatial and temporal analysis of two transient deglacial climate simulations, one using a fully coupled GCM (TraCE-21ka) and one using an intermediate complexity model (LOVECLIM DGns), to determine regional differences in deglacial climate evolution and identify the main strengths and limitations of the models in terms of climate variables that impact ice sheet mass balance. The greatest continental surface warming is observed over the continental margins in both models, with strong correlations between surface albedo, sea ice coverage, and surface air temperature along the coasts, as well as regions with the greatest decrease in ice surface elevation in TraCE-21ka. Accumulation–temperature scaling relationships are fairly linear and constant in the continental interior, but exhibit higher variability in the early to mid-Holocene over coastal regions. Circum-Antarctic coastal ocean temperatures at grounding line depths are highly sensitive to the meltwater forcings prescribed in each simulation, which are applied in different ways due to limited paleo-constraints. Meltwater forcing associated with the Meltwater Pulse 1A (MWP1A) event results in subsurface warming that is most pronounced in the Amundsen and Bellingshausen Sea sector in both models. Although modelled centennial-scale rates of temperature and accumulation change are reasonable, clear model–proxy mismatches are observed with regard to the timing and duration of the Antarctic Cold Reversal (ACR) and Younger Dryas–early Holocene warming, which may suggest model bias in large-scale ocean circulation, biases in temperature reconstructions from proxy records, or that the MWP1A and 1B events are inadequately represented in these simulations. The incorporation of dynamic ice sheet models in future transient climate simulations could aid in improving meltwater forcing representation, and thus model–proxy agreement, through this time interval.