The physics of cloud and water vapour feedbacks in perturbed-physics ensembles

This project applies two powerful and complementary tools, the very large perturbed-physics ensembles that can be performed by distributed computing and detailed physically-oriented observational validation, to a state-of-the-art climate model to provide a more quantitative understanding of the atmospheric feedbacks that determine the climate sensitivity, or warming response to increasing greenhouse gases. Using the experience of previous perturbed-physics ensembles performed by the Hadley Centre and climateprediction.net (cpdn), we will identify a range of physical perturbations to the latest Hadley Centre AGCM (atmospheric general circulation model), HadGAM, and a compact set of diagnostics aimed at testing a wide range of aspects of the simulation, with emphasis on the physical processes involved in the cloud and water vapour feedbacks. Unlike earlier experiments, perturbations will encompass both parameter variations and structural modifications. We will port the AGCM, with these perturbations and diagnostics, to the BOINC (Berkeley Open Infrastructure for Network Computing) public-domain distributed computing framework. Developments in personal computing processor technology mean that all computations will be performed in the model's native (64-bit) precision, simplifying comparison with supercomputing results. Thousands of short perturbed-physics simulations driven with observed sea surface temperatures (SST) for recent (satellite-era) periods will be performed by volunteers from the general public on home computers, and output uploaded and archived by cpdn. Software embedded into the distributed computing package will allow accurate comparison with satellite datasets without the requirement of recovering prohibitively large four-dimensional datasets. In conjunction with non-NERC-funded workers at the Hadley Centre and NASA Langley, we will validate these models against a wide range of satellite and other diagnostics, and analyze the effect of different parametrization choices and their interactions on the quality of the simulation, particularly of cloud, water vapour and radiation. The relatively short duration of these runs means that we will be able to use successive runs to optimise, for example, the atmospheric energy budget in promising model versions. The response of clear- and cloudy-sky top-of-atmosphere fluxes to interannual SST variations will be used to identify a much smaller set of models that validate well across the board and are likely to display a wide range of water vapour and cloud feedbacks. The Hadley Centre will then run idealized climate change simulations with these models to establish the actual strength of these feedbacks. This will provide an unprecedentedly detailed and quantitative understanding of the range of climate sensitivities that are consistent with observations interpreted with a state-of-the-art AGCM. It will also provide for the first time an ensemble of possible versions of the atmospheric component of a state-of-the-art Earth System Model rather than the traditional single best-guess version. This will significantly improve the objectivity of, and treatment of model error in, future Earth System research.