A Bayesian assessment of climate change using multimodel ensembles. Part II: regional and seasonal mean surface temperatures

Abstract

A Bayesian approach is applied to the observed regional and seasonal surface air temperature ( SAT) changes using single- model ensembles ( SMEs) with the ECHO- G model and multimodel ensembles ( MMEs) of the Intergovernmental Panel on Climate Change ( IPCC) Fourth Assessment Report ( AR4) simulations. Bayesian decision classifies observations into the most probable scenario out of six available scenarios: control ( CTL), natural forcing ( N), anthropogenic forcing ( ANTHRO), greenhouse gas ( G), sulfate aerosols ( S), and natural plus anthropogenic forcing ( ALL). Space - time vectors of the detection variable are constructed for six continental regions ( North America, South America, Asia, Africa, Australia, and Europe) by combining temporal components of SATs ( expressed as Legendre coefficients) from two or three subregions of each continental region. Bayesian decision results show that over most of the regions observed SATs are classified into ALL or ANTHRO scenarios for the whole twentieth century and its second half. Natural forcing and ALL scenarios are decided during the first half of the twentieth century, but only in the low- latitude region ( Africa and South America), which might be related to response patterns to solar forcing. Overall seasonal decisions follow annual results, but there are notable seasonal dependences that differ between regions. A comparison of SME and MME results demonstrates that the Bayesian decisions for regional- scale SATs are largely robust to intermodel uncertainties as well as prior probability and temporal scales, as found in the global results.