Multi-GCM climate projections for Europe: GCMs performance and uncertainties

Future climate projections rely mostly on Global Climate Models (GCMs). As the number of GCMs increases, the problem how to effectively deal with the multi-GCM information arises. Typically, one chooses several GCMs to develop a set of climate change scenarios to be used in the climate change impacts analysis, the choice of GCMs being based on the ability of the GCMs to reproduce the present climate. Alternatively, one can employ a climate scenario emulator (generator), which estimates joint probability density function of climatic characteristics based on the multi GCM information and then generates an arbitrarily large set of climate scenarios to be used in the probabilistic assessment of a response of the weather-dependent system to the climate change. In the both above cases, the quality of the GCMs plays an important role: as a criterion for choosing a subset of GCMs in the first case, and as the basis for defining the GCM-specific weights in the scenario emulator in the latter case.