Deglacial climate changes as forced by different ice sheet reconstructions

Deglacial climate changes as forced by different ice sheet reconstructions

by Nathaelle Bouttes

During the last deglaciation, the climate evolves from a cold state at the Last Glacial Maximum (LGM) at 21 ka (thousand years ago) with large ice sheets to the warm Holocene at 9 ka with reduced ice sheets. The deglacial ice sheet melt can impact the climate through multiple ways: changes of topography and albedo, bathymetry and coastlines, and freshwater fluxes (FWFs). In the PMIP4 (Paleoclimate Modelling Intercomparison Project – Phase 4) protocol for deglacial simulations, these changes can be accounted for or not depending on the modelling group choices. In addition, two ice sheet reconstructions are available (ICE-6G_C and GLAC-1D). In this study, we evaluate all these effects related to ice sheet changes on the climate using the iLOVECLIM model of intermediate complexity. We show that the two reconstructions yield the same warming to a first order but with a different amplitude (global mean temperature of 3.9 C with ICE-6G_C and 3.8 C with GLAC-1D) and evolution. We obtain a stalling of temperature rise during the Antarctic Cold Reversal (ACR, from 14 to 12 ka) similar to proxy data only with the GLAC-1D ice sheet reconstruction. Accounting for changes in bathymetry in the simulations results in a cooling due to a larger sea ice extent and higher surface albedo. Finally, freshwater fluxes result in Atlantic meridional overturning circulation (AMOC) drawdown, but the timing in the simulations disagrees with proxy data of ocean circulation changes. This questions the causal link between reconstructed freshwater fluxes from ice sheet melt and recorded AMOC weakening.

Authors: Nathaelle Bouttes, Fanny Lhardy, Aurélien Quiquet, Didier Paillard, Hugues Goosse, and Didier M. Roche


Temperature evolution (°C) for the simulations with freshwater fluxes (a) of the global mean (b) at NGRIP (Greenland) and (c) at EDC (Antarctica). The simulated results are shown as the running mean over 100 years and compared to proxy data (Shakun et al., 2012; Buizert et al., 2014; Jouzel et al., 2007).