Estimates of climate change in Southern Europe derived from dynamical climate model output

Three methods of downscaling are applied to climate change experiments to obtain regional climate information for Spain and the region designated as 'Southern Europe' by the lntergovernmental Panel on Climate Change (IPCC). The first method (direct interpolation of the grid points nearest the region analysed) gives a poor representation of the local climate. Its estimate of climate change simulated by different climate models is inconsistent. The success of the second method (timeslice), which uses a dynamical model to obtain the regional information, strongly depends on the horizontal resolution of the dynamical model. It provides the most reliable assessment of the regional climate, with the highest resolution. However, the computational expense of this high resolution limits the sample size. The third method (statistical downscaling) is an inexpensive tool for obtaining information on a regional scale. The problem is that this approach requires observational data sets to train the model. This limits the application of this method to well-observed quantities and regions. Both the time-slice and the statistical models indicate a lengthening of dry spells over Spain under CO2doubling conditions.