CR 29:255-268 (2005)  -  doi:10.3354/cr029255

ABSTRACT: Studies from recent years involving development and application of statistical downscaling models for Scandinavia (mainly Norway and Sweden) are reviewed. In most of the studies linear techniques were applied. Local temperature and/or precipitation were predictands in a majority of the studies. Large-scale temperature fields, either from 2 m or 850 hPa, were found to be the best predictors for local temperature, while a combination of atmospheric circulation indices and tropospheric humidity information were the best predictors for local precipitation. Statistically downscaled temperature scenarios for Scandinavia differ depending on climate model, emission scenario and downscaling strategy. There are nevertheless several common features in the temperature scenarios. The warming rates during the 21st century are projected to increase with distance from the coast and with latitude. In most of Scandinavia higher warming rates are projected in winter than in summer. For precipitation, the spread between different scenarios is larger than for temperature. A substantial part of the projected precipitation change is connected to projected changes in atmospheric circulation, which differ considerably from one model integration to another. A tendency for increased large-scale humidity over Scandinavia still implies that projections for the 21st century typically indicate increased annual precipitation. This tendency is most significant during winter. In northern Scandinavia the projections tend to show increased precipitation also during summer, but several scenarios show reduced summer precipitation in parts of southern Scandinavia. Comparisons with results from global and regional climate models indicate that both regional modeling and statistical downscaling add value to the results from the global models.

KEY WORDS: Local climate scenarios · Temperature · Precipitation · Scandinavia