Modelling of the end of last Ice Age in transient framework with a coupled climate model

Abstract

The last deglaciation was characterized by a sequence of abrupt climate events thought to be linked to rapid changes in Atlantic meridional overturning circulation (AMOC). The sequence includes a weakening of the AMOC after the Last Glacial Maximum (LGM) during Heinrich Stadial 1 (HS1), which ends with an abrupt AMOC recovery giving rise to Bølling/Allerød (B/A) warming. This transition occurs under a background with persistent deglacial meltwater fluxes (MWF) that are deemed to play a negative role in North Atlantic Deep Water (NADW) formation. Using a fully coupled Earth system model COSMOS with a range of deglacial boundary conditions and reconstructed deglacial meltwater fluxes, we show that deglacial CO2 rise and ice sheet decline modulate the sensitivity of the AMOC to these fluxes. While declining ice sheets increase the sensitivity, increasing atmospheric CO2 levels tend to counteract this effect. These effects, therefore, might account for the occurrence of abrupt AMOC increase in the presence of meltwater, as an alternative to or in combination with changes in the magnitude and/or distribution of meltwater discharge.

To understand the dynamics of B/A warming, an appropriate Heinrich Stadial ocean state is necessary, which was previously obtained by imposing meltwater flux directly into the key convection sites of the North Atlantic, though without a proper distribution of meltwater flux along continental coasts. Given a more realistic distribution of meltwater flux according to PMIP4 protocol, the HS1-B/A sequence could not be properly reproduced in transient simulations, perhaps due to a reduced freshwater flux during HS1 and significant MPW-1a during B/A. From our transient simulations, the intensity of AMOC is significantly modulated by the amount of freshwater during the HS1-B/A transition and manipulated by the geographic distribution of freshwater injection. Considering only the geographical distribution of freshwater forcing without its temporal variation, our model shows a stadial climate during HS1 and captures the B/A-like AMOC recovery. This B/A-like behavior is characterized by the self-oscillation of the AMOC, and is mainly associated with a gradual change in the orbital parameters. The ice sheet decline after the onset of the abrupt change attenuates the North Atlantic Deep Water (NADW) production, and the accelerated melting of the ice sheet after 14.5 ka (thousand calendar years before present) can develop this process, eventually moving the system out of this self-oscillation regime. The rise in atmospheric CO2 does not offset this trend due to ice sheet retreat.