CR 19:1-13 (2001)  -  doi:10.3354/cr019001

ABSTRACT: A method is introduced by which the results of a regional climate model (RCM) that is temporarily nested in a general circulation model (GCM) are extrapolated in time. The procedure is based on a weather-type classification scheme using cluster analysis. It relies on statistical relationships between GCM and RCM output which are deduced from the overlapping nesting period. For a validation of the extrapolation scheme, a 30 yr GCM simulation of the Hadley Centre with a continuously nested RCM is used. The predictor is the large-scale 500 hPa geopotential height, and the predictands are the regional-scale surface temperature and precipitation. A comparison of the results of the extrapolation scheme with the direct RCM output for the winter temperature and precipitation in Central Europe demonstrates the potential of the method developed. Provided that results of a long-term GCM run are available and the statistics can be transferred from the overlapping period to the extrapolation period, the new method allows a reduction in the computational effort needed for long-term RCM simulations by shortening the integration time considerably.

KEY WORDS: Climate change · Regional climate models · Statistical-dynamical extrapolation · Downscaling