资源环境科技发展态势分析平台

Climate simulations of future hot extremes exhibit large uncertainties regarding the magnitude of projected warming. We identify two mechanisms that influence how strongly future heat extremes intensify in climate models. First, the magnitude of extreme temperature increases is determined by changes in preceding seasonal precipitation, connected to amplified warming via soil moisture decreases. Second, there are large differences in how models respond to moisture variability; those with a stronger response under current climate simulate larger future increases in hot extremes. We build on this mechanistic understanding of future uncertainty and develop a novel constraint, the observed precipitation-hot temperature relationship, focused on the conditions on the actual hottest day, to identify climate models with realistic land-atmosphere feedbacks on hot extremes. Applying this constraint to the Coupled Model Intercomparison Project Phase 5 ensemble reduces the probability of the largest increases in projected heat extremes, particularly over Europe and North America.

Plain Language Summary Climate model simulations of future hot temperature extremes agree on the regions of strongest warming but disagree on how much temperature will increase. We identify that a large part of the uncertainty is explained by intermodel differences in future precipitation changes, and differences in the model-specific strength of land-atmosphere feedbacks. We constrain the future climate model ensemble simulations by using only those models that simulate land-atmosphere feedbacks more similar to real-world observations over the past 60 years. This reduces the uncertainty of future changes by reducing the probability of the strongest increases in hot extremes particularly over Europe and North America.