Verification of statistical-dynamical downscaling in the Alpine region

Abstract

A statistical-dynamical downscaling procedure for global climate simulations is verified for the greater Alpine region. This procedure links global and regional model simulations using frequencies of large-scale weather types in order to derive the regional climate corresponding to a given global climate. The results from multi-year global simulations or large-scale analyses are classified into a set of large-scale weather types. Regional model simulations are carried out once for each class. The model output fields are weighted with the frequencies of the corresponding weather type to give the regional climate. As downscaling procedures transmit errors in the global climate simulation to the regional scale, the verification of the statistical-dynamical procedure is based on large-scale observational data, in our case a decade of daily ECMWF (European Centre for Medium Range Weather Forecasts) analyses. The downscaled results are compared with local observations for the same decade. The verification shows that the downscaling procedure is able to reproduce climatological features of the observed regional distributions of wind and temperature. However, the downscaled regional distribution of precipitation exhibits major deficiencies. Possible reasons are discussed and further improvements of the statistical-dynamical downscaling procedure are previewed.