Séminaire
Many studies have sought to constrain climate projections and climate sensitivity based on recent observations. Until recently, these constraints had limited impact, and projected warming ranges were driven primarily by model outputs. Here, we use the newest climate model ensemble (CMIP6), improved observations, and a new statistical method to narrow uncertainty on estimates of past and future human-induced warming. Our new statistical method allows us to simultaneously attribute historical changes to specific forcings (attribution) and constrain projections. It provides a consistent picture of on-going changes, through merging model simulations and observations in a Bayesian fashion. Cross-validation suggests that our method produces robust results and is not overconfident. We derive consistent observationally constrained estimates of attributable warming to date and warming rate, the response to a range of future scenarios, and metrics of climate sensitivity. We find that historical observations narrow uncertainty on projected future warming by about 50%. Our results suggest that using an unconstrained multimodel ensemble is no longer the best choice for global mean temperature projections. In a somewhat paradoxical way, they also suggest that (i) the upper end of CMIP6 models projected warming cannot be reconciled with observations, and that (ii) the lower end of previous estimates of 21st century warming can now be excluded.
Connexion :
https://www.gotomeet.me/SeminairesGeosciencesENS
helene.rouby@ens.fr