Results of the first 10-year climate change experiment made with the Rossby Centre regional climate model (RCA) are described. The boundary data for this experiment were derived from two simulations with the .global HadCM2 ocean-atmosphere GCM, a control run anda scenario run with 150% higher equivalent CO2 and 2.6°C higher global mean surface air temperature.

Some of the climate changes (scenario run - control run) simulated by RCA are substantial. The annual mean temperature in the Nordic region increases by roughly 4°C, with largest warming in winter. Annual absolute minimum temperatures increase even more than the winter mean temperature, presumably due to greatly reduced snow and ice cover. Precipitation is also simulated to increase in northern Europe, locally by 40% in the annual mean in Swedish Lappland. The larger time mean precipitation is accompanied by a marked increase in the number of days with heavy precipitation.

The large-scale temperature and precipitation changes simulated by RCA are similar to those in HadCM2. Unlike HadCM2, however, RCA simulates a strong local maximum of wintertime warming over the northern parts of the Baltic Sea. This is caused by radically reduced ice cover, but the crude treatment of the Baltic Sea and its ice even in RCA complicates the interpretation. Large differences between the models occur in the simulated changes of winter mean total cloudiness and near-surface wind speed, demonstrating the sensitivity of these to differences in resolution and/or physical parameterizations.

The significance of the simulated climate changes against interannual variability depends on the parameter considered. Of highest statistical significance are changes in surface air temperature and strongly temperature-related variables such as snow and ice cover. In general, changes in annual means are more commonly significant than those in seasonal means. The impact of the limited averaging period is also studied by comparing the 10-year mean climate changes simulated by the driving HadCM2 mode! with climate changes inferred from much longer HadCM2 integrations.