Analysing the climate sensitivity of the HadSM3 climate model using ensembles from different but related experiments

Projections of future climate are determined by a wide array of detailed and imperfectly understood physical processes. While ensembles of global climate model (GCM) simulations can be run to sample the resulting uncertainties, statistical tools are needed to supplement these simulations, and hence provide a more comprehensive picture of how interactions between different processes influence the response of climate. We illustrate this by combining ensembles from two different experiments to study the response of climate sensitivity in the HadSM3 climate model to 31 model-parameters. We use a Bayesian statistical framework based around linked emulators. Expert judgements are required to quantify the relationship between the two emulators, and these are validated by detailed diagnostics. Using our emulator, we identify the entrainment rate coefficient of the convection scheme as the most important single parameter; find that this interacts strongly with three of the large scale cloud parameters, Ct, Cw (land), and Rhcrit; and represent these interactions visually.