Explaining the chain of events of the last deglaciation through the convection-advection oscillator mechanism

The last glacial period, spanning between 115 and 12 thousand years before present, exhibited pronounced millennial-scale climate variability. This includes abrupt and semi-periodic transitions between cold stadial and warm interstadial climates, known as Dansgaard-Oeschger (D-O) events, and the chain of events of the last deglaciation. Abrupt climate changes have been linked to switches in regimes of the Atlantic Overturning Meridional Circulation (AMOC), but the exact mechanisms behind abrupt climate changes and AMOC regimes switches remain poorly understood.  

This work introduces the convection-advection oscillator mechanism to explain the millennial-scale climate oscillations observed in a set of HadCM3 general circulation model simulations forced with snapshots of deglacial meltwater history. The oscillator can be separated into two components acting on different time scales. The fast convection component responds to changes in vertical stratification in the North Atlantic by activating or deactivating its deep water formation sites. The slow advection component regulates the accumulation and depletion of salinity in the North Atlantic. This mechanism is triggered only when the right balance of magnitude and location of the freshwater forcing and boundary conditions are obtained.  

The chain of events of the last deglaciation may have been caused by the triggering of millennial-scale variability through this mechanism. We tested this hypothesis with HadCM3 simulations forced with two different ice sheet reconstructions and their associated meltwater forcing, and showed that under fixed LGM boundary conditions, only one of the ice sheet reconstructions can produce abrupt climate changes. Modifying the boundary conditions modify the location of the window of opportunity and, therefore, the forcing necessary to activate the convection-advection mechanism.  

The uncertainties around the ice sheet reconstructions and lack of climate-ice feedback still prevent the exact reproduction of the chain of events of the last deglaciation, but this study provides a new way to understand the window of parameters where millennial-scale variability can occur in simulations of the last deglaciation.