Large-scale atmospheric circulation biases and changes in global climate model simulations and their importance for climate change in Central Europe
[摘要] The quality of global sea level pressure patterns has been assessed forsimulations by 23 coupled climate models. Most models showed high patterncorrelations. With respect to the explained spatial variance, many modelsshowed serious large-scale deficiencies, especially at mid-latitudes. Fivemodels performed well at all latitudes and for each month of the year. Threemodels had a reasonable skill.
We selected the five models with the best pressure patterns for a moredetailed assessment of their simulations of the climate in Central Europe. Weanalysed observations and simulations of monthly mean geostrophic flowindices and of monthly mean temperature and precipitation. We used threegeostrophic flow indices: the west component and south component of thegeostrophic wind at the surface and the geostrophic vorticity. We found thatcirculation biases were important, and affected precipitation in particular.Apart from these circulation biases, the models showed other biases intemperature and precipitation, which were for some models larger than thecirculation induced biases.For the 21st century the five models simulated quite different changes incirculation, precipitation and temperature. Precipitation changes appear tobe primarily caused by circulation changes. Since the models show widelydifferent circulation changes, especially in late summer, precipitationchanges vary widely between the models as well. Some models simulate severedrying in late summer, while one model simulates significant precipitationincreases in late summer. With respect to the mean temperature thecirculation changes were important, but not dominant. However, changes in thedistribution of monthly mean temperatures, do show large indirect influencesof circulation changes. Especially in late summer, two models simulate verystrong warming of warm months, which can be attributed to severe summerdrying in the simulations by these models. The models differ alsosignificantly in the simulated warming of cold winter months. Finally, themodels simulate rather different changes in North Atlantic sea surfacetemperature, which is likely to impact on changes in temperature andprecipitation. These results imply that several important aspects of climatechange in Central Europe are highly uncertain. Other aspects of the simulatedclimate change appear to be more robust. All models simulate significantwarming all year round and an increase in precipitation in the winterhalf-year.
