Climate models and climate extremes

Climate change will alter expectations about the future to such an extent that past statistics about averages, variability and extremes will no longer be relevant. Scenarios of the future depend upon the rate of greenhouse gas increases (and changes in aerosol distributions) in the atmosphere, and are best produced by coupled ocean-atmosphere climate models. Uncertainties in the scenarios are large, however, not only because future emissions have to be estimated, but also because the climate system is complex and varies considerably from decade to decade naturally. The temporal and spatial scales for which information can be produced generally means that detail can only be developed through downscaling procedures. Until recently, downscaling was generally achieved by statistical means using relationships between spatial scales. Increasingly, downscaling is now being undertaken using regional-scale models embedded within the global model (e.g., for the scale of Europe). Resolutions of 50 km in space at the daily timescale are possible, but just as with the large models there is still a need for model verification, particularly with respect to the daily variability and extremes, the timescale at which much of climate change is perceived by nonatmospheric scientists.