École Nationale Supérieure de
Techniques Avancées

Résumé

Ongoing greenhouse gas emissions put several sub-systems of the climate at risk of crossing critical thresholds (so-called tipping points), leading to abrupt irreversible climate change. These include Arctic sea ice and permafrost, the Amazon rainforest, the polar ice sheets, as well as the Atlantic ocean circulation. The crossing of a tipping point in one climate sub-systems might increase the likelihood of tipping in another, potentially leading to a domino-effect with a cascade of subsequent tipping points. This phenomenon might also occur locally within a climate sub-system, which has not been studied in much detail yet. In tropical rainforests, where full forest cover and savanna compete locally, local-scale tipping interacts with a regional-scale moisture feedback to determine the stability of the rainforest. Perturbations, such as forest fires or deforestation, may cause a cascading tipping of the ecosystem. In this study, a model describing these local and regional effects was developed with the formalism of a cellular automata, in order to evaluate the impact of deforestation on the moisture recycling network of tropical rainforests and to find novel early-warning signals. It models two main features of tropical rainforests : the forest-tree cover feedback and a moisture recycling network. We find that all scenarios of deforestation induce water scarcity over the rainforest by interfering with the moisture recycling network. Hence deforestation generates forest dieback in climate configurations that should prevent forest collapse, suggesting that deforestation could lead the rainforest to tip to a savanna-type state before climate change makes forest dieback relentless. In addition, special pattern formation resulting from the nature of the moisture recycling network was found and could be used as an early-warning signal, but this last result is very uncertain given the idealized nature of the model.