The medium and long-term effects of anthropogenic emissions are currently studied using global climate model projections of the future climate under specific greenhouse gas emission scenarios. These simulations provide valuable information on large-scale patterns of change for important climate variables such as temperature and rainfall, but their coarse resolution does not allow scientists to assess potential changes due to small scale or very fast weather events, even for areas as large as the UK. These phenomena could have a bigger impact on society than the changes in spatially and temporally averaged variables estimated by global models. For this reason, a proper assessment of the impacts of climate change should include the study of the impacts at the local scale.
Downscaling tools such as regional climate models can be used to add finer scale information to the global climate scenario. The example shown in the figure is a comparison between projected winter rainfall changes (between the period 2070-2100 and 1960-1990 for the A2 SRES emissions scenario) for the regional climate model (right-hand side) and the driving global model (left-hand side). The large-scale patterns are similar, but we can easily distinguish the lower increases over the Pyrenees and the Alps in the regional model, not present at the global model scale. Since these mountain chains are badly resolved at the global model scale, the only way to obtain specific information about areas with complex orography is by downscaling.
For a given emissions scenario, the existing global models give a range of possible, plausible future climate. The most rigorous way to assess climate change for a small region is by including the largest possible number of global model projections in the study. The uncertainty due to the global model is not the only one: for a given global model simulation, slightly different results can be obtained by applying a different downscaling strategy. Climate projections at a local scale can be constructed by combining all the information obtained from a suitable range of downscaling tools applied to a set of global climate model simulations. The ENSEMBLES project has generated these projections for the European continent (see ENSEMBLES final report, pages 60-61 and 144-148).