A statistical model is constructed to determine the change in wintertime mo
nthly mean Alpine precipitation due to a tripling of the CO2 concentration.
The model is based on optimally correlated pairs of predictand (Alpine pre
cipitation) and predictor (500 hPa geopotential) patterns which were determ
ined from observational data by a canonical correlation analysis. The stati
stical model is applied to the change in large scale climate simulated by a
general circulation model (GCM for a 3 x CO2 concentration. The statistica
l model predicts a change of Alpine precipitation of up to 6% of the observ
ed monthly rainfall for a 3 x CO2 climate. A maximal increase of wintertime
precipitation is obtained for the stations in the south of the Alps and on
ly slight changes are predicted for the stations in the north of the Alps.
it is found to be important to remove the area averaged change of the geopo
tential simulated by the GCM and to apply the statistical model only to the
spatial variation of the geopotential. The regional climate change estimat
ed with the statistical model depends only slightly on the exact set-up of
the model. Thus the error of the statistical model seems to be small in com
parison with the uncertainty in the large scale climate predicted by the GC
M for 3 x CO2 conditions.