Exploring the risk of climate-change-induced forest dieback in Amazonia using multi-model ensemble simulations

Some global modeling studies suggested the possibility of a large-scale dieback of the Amazon forest by the end of the twenty-first century resulting from a positive feedback between regional climate (large increase of temperature and decrease of precipitation) and vegetation under enhanced CO₂ emission. However, due to the coarse spatial resolution used, global climate models (GCM) tend to have difficulty in capturing the spatial pattern of precipitation and vegetation distribution, which may influence the sensitivity of the regional climate-vegetation system to CO₂ concentration changes; moreover, results from GCMs are highly model dependent. To tackle this uncertainty, in this study, we take a multi-model ensemble method to evaluate future climate and vegetation changes in Amazonia using a regional climate model RegCM4.3.4-CLM-CN-DV. The initial and boundary conditions for the RCM were derived from four GCMs under the greenhouse gas Representative Concentration Pathway 8.5, and the HadGEM model (that predicted a forest dieback in Amazonia) was among the four. Projected future changes will be evaluate based on the ensemble average and based on individual model as well. The roles of the CO₂ fertilization effects, water stress, and heat stress will be analyzed and compared. Special attention will be paid towards identifying potential tipping point beyond which rainforest would become unsustainable and dieback dramatically.