Abstract
Due to imperfections inherent to numerical climate simulations, projections of climate change need to be accompanied with an uncertainty estimate, typically derived from multi-model ensembles. Here, an analysis of regional climate model outputs spread and its temporal evolution is presented, for the European air temperatures simulated over the 21st century.
To quantify the overall uncertainty, standard deviation (sd) is used and compared to median absolute deviation and inter-quartile range. The outcomes are shown for annual and seasonal means of air temperature at 2m, 850 hPa, 700 hPa, 500 hPa and 300 hPa levels, simulated by ENSEMBLES regional climate models for eight geographical areas across Europe.
Our results indicate that the estimated uncertainty of simulated air temperature changes over Europe generally increases during the 21st century, and typically decreases with height of the studied vertical level. The influence of driving global climate model on the regional simulation generally increases with height of the studied vertical level.
In lower tropospheric levels, the clustering of regional climate models according to driving global climate model is stronger in winter than in summer.