Abstract
Two statistical downscaling methods have been tested in terms of their ability to construct indices of extremes of daily precipitation and temperatures from large‐scale atmospheric variables with the aim of developing a tool for the construction of future scenarios of the extremes. One of the methods implements an approach for constructing seasonal indices of extremes of precipitation and temperature from seasonal measures of large‐scale variables, while the other method implements a stochastic model for generating daily series of precipitation and temperature whose parameters are conditioned on large‐scale circulation patterns. While both models generally tend to perform fairly well in reproducing indices of precipitation in winter, their performance for the summer season is not attractive. For indices of temperature, the performance of both models is better than the corresponding performance for indices of precipitation and the seasonal variation in performance is less prominent. The models were applied to construct scenarios of the extremes for the end of the 21st century using predictor sets simulated by the Hadley Centre GCM (HadAM3P) forced by two of the special report on emission scenarios (SRES) emission scenarios. Both models project an increase in both the mean daily minimum and mean daily maximum temperatures for future climate change scenarios in all seasons. The summer increase is accompanied by an increase in the inter‐annual variability of the temperatures. On the other hand, they show consistency in the direction of the projected changes in indices of precipitation only in winter, where they projected an increase in both the magnitude and frequency of extremes as well as the mean precipitation. The disparity in the changes simulated by the two models revealed the existence of considerable inter‐model uncertainty in predicting changes for future climate. Copyright © 2007 Royal Meteorological Society