Resilience of the Amazon rainforest to an AMOC collapse

The Amazon rainforest and the Atlantic Meridional Overturning Circulation (AMOC) are two crucial components of the Earth system that may be on tipping course. As the world’s largest tropical rainforest, the Amazon is a major carbon sink and biodiversity reservoir. Therefore, its transition into a savanna under the pressure of climate change may have large-scale consequences. The AMOC plays an important role in the global meridional heat transport and may collapse due to greenhouse gas emissions. This would, among other consequences, alter the ocean-induced moisture inflows over the Amazon rainforest, where precipitation over certain areas may increase and decrease over other areas.

Resilience of the Amazon is here defined as the ability of the Amazon rainforest to remain in a rainforest regime or to return to this state, against perturbations that may bring it towards a savanna-like state. Our objective is to analyse the resilience of the Amazon rainforest with respect to a collapse of the AMOC. More precisely, we study the interaction between the Amazon rainforest and AMOC in a conceptual coupled. We have estimated parameter values (such as mean annual precipitation) of the Amazon rainforest dynamics from a AMOC collapse experiment carried out in a CMIP5 global model (CESM1).

The notion of resilience we consider here is solely based on footprints extracted from an ensemble of AMOC tipping trajectories. It is difficult to simulate many instances of an AMOC collapse because of the rarity of such event. We overcome this problem using a rare-events algorithm, Transition-Adaptive Multilevel Splitting (TAMS), that iteratively pushes trajectories towards a tipping, however unlikely it is, until obtaining an ensemble of tipped trajectories and the corresponding probability of tipping. This algorithm allows us to obtain an ensemble of simulations where the Amazon rainforest tips (i.e. loses resilience) under the influence of the AMOC, at a much lower computational cost than with Monte-Carlo simulations. From this ensemble, we can then extract from this ensemble, footprints characterizing the behaviour of the system as it is losing resilience. With this approach, we can precisely quantify the resilience of the Amazon rainforests to changes induced by an AMOC collapse.