Multivariate probabilistic projections using imperfect climate models. Part II: robustness of methodological choices and consequences for climate sensitivity

Abstract

A method for providing probabilistic climate projections, which applies a Bayesian framework to information from a perturbed physics ensemble, a multimodel ensemble and observations, was demonstrated in an accompanying paper. This information allows us to account for the combined effects of more sources of uncertainty than in any previous study of the equilibrium response to doubled CO2 concentrations, namely parametric and structural modelling uncertainty, internal variability, and observational uncertainty. Such probabilistic projections are dependent on the climate models and observations used but also contain an element of expert judgement. Two expert choices in the methodology involve the amount of information used to (a) specify the effects of structural modelling uncertainty and (b) represent the observational metrics that constrain the probabilistic climate projections. These choices, effected by selecting how many multivariate eigenvectors of a large set of climate variables to retain in our analysis, are investigated in more detail. We also show sensitivity tests that explore a range of key expert choices. For changes in annual global mean temperature and regional changes over England and Wales and Northern Europe, the variations in the projections across the sensitivity studies are small compared to the overall uncertainty, demonstrating that the projections are robust to reasonable variations in key assumptions. We are therefore confident that, despite sampling sources of uncertainty more comprehensively than previously, the improved multivariate treatment of observational metrics has narrowed the probability distribution of climate sensitivity consistent with evidence currently available. Our 5th, 50th, and 95th percentiles are in the range 2.2–2.4, 3.2–3.3, and 4.1–4.5K, respectively. The main caveat is that the handling of structural uncertainty does not account for systematic errors common to the current set of climate models and finding methods to assess the impact of this provides a major challenge.