CR 22:161-173 (2002)  -  doi:10.3354/cr022161

ABSTRACT: Regional risk assessments for the potential effects of climate change rely on plausible small-scale climate change scenario data. To bridge the gap between the coarse scale of general circulation models and the local scale of approximately 11000 sample sites of the Austrian National Forest Inventory (AFI), a seasonally stratified statistical downscaling procedure was applied to a control run and 2 transient experiments of ECHAM4/OPYC3, which are based on the trace gas only or trace gas plus sulphate scenario IPCC IS92a. We fitted multiple linear regression (MLR) models for the micro-scale monthly precipitation and temperature for each AFI point. The meteorological data at the AFI sites were obtained by interpolation of measurements from the dense network of Austrian weather stations for the period 1961-1995. The macro-scale predictors were principal components of monthly NCEP/NCAR reanalysis data (850 and 700 hPa geopotential height, 850 hPa temperature and 700 hPa relative humidity). The results show spatial and temporal heterogeneity for both temperature and precipitation. In case of temperature MLR leads to increases from +1.4 to +4.0ºC (trace gas only integration) and from +1.1 to +2.9ºC (trace gas plus sulphate integration) for a period of about 55 yr relative to the 1961-1995 climatology. The regionalized precipitation changes are both negative and positive. Values range from -44 to +26% (trace gas only integration) and from -29 to +26% (trace gas plus sulphate integration). As expected, the explained variability for temperature was higher than for precipitation and depended on the season. From a validation experiment for model calibration we conclude that MLR shows reliable results for temperature. Even in the case of precipitation the method seems to yield plausible results. Both temperature and precipitation were better reproduced for winter than for summer.

KEY WORDS: Climate change · Downscaling · Principal component analysis · Multiple regression · Forest inventory