Transient and equilibrium responses of the Atlantic meridional overturning circulation to warming in coupled climate models

The long-term response of the Atlantic meridional overturning circulation (AMOC) to anthropogenic climate change remains poorly understood in part, due to the computational expenses associated with running fully-coupled climate models to equilibrium. Here, we use a collection of millennial-length simulations from multiple state-of-the-art climate models to examine the transient and equilibrium responses of the AMOC to an abrupt quadrupling of atmospheric carbon-dioxide. All climate models exhibit a weakening of the AMOC on centennial timescales, but they disagree on the recovery of the AMOC over next millennia, despite the same greenhouse-gas forcing. In some models, the AMOC recovers after approximately 200 years, while in others the AMOC does not fully recover even after approximately 1000 years. To explain the behavior of the AMOC we relate the overturning circulation in the North Atlantic to the meridional density difference between the basin interior and the region of deep-water formation. This scaling both reproduces the initial decline and gradual recovery of the AMOC, and explains the inter-model spread of the AMOC responses. The initial shoaling and weakening occurs on centennial timescales and is attributed to the warming of the northern convection region. We argue that the AMOC weakens on a timescale linked to a combination of its initial depth and the global surface heat flux sensitivity. The recovery of the AMOC results from a pile-up of salinity in the Atlantic basin, when the AMOC is weakened, that propagates northward and reinvigorates convection. A weaker AMOC recovery is associated with a smaller salinity anomaly. We further show through surface water mass transformation that Southern Ocean processes may impact the salinity anomaly in the Atlantic basin. These results highlight the importance of considering the evolution of the AMOC and ocean heat transport beyond the 21st century as short-term changes are not indicative of long-term changes.