An ensemble of global climate scenarios from different coupled atmosphere–ocean general circulation models are evaluated against re-analysed observations in terms of temperature trends over the period 1958–1998. The spatial warming rate patterns for the “past” derived from the individual models exhibit large differences, but the median values of the ensemble appears to capture some of the spatial structure observed. A probabilistic approach describing the past warming rates is evaluated, and the combination of large inter-model differences and the good evaluation results justifies a probabilistic view on the future climate.
Regional temperature scenarios are presented for northern Europe in the form of probability distributions, based on spatially interpolated empirically downscaled trends, derived using a multi-model ensemble as well as various downscaling options. The estimation of probabilities is based on the assumption that there is no common systematic error in the various model scenarios, the IS92a scenario is valid for the future and their scatter portrays the actual uncertainty. The results point to a model consensus of stronger warming over land during winter, with maximum warming over Finland. The same analysis for July suggests an overall weaker summertime warming with a smaller difference between the coasts and the interior.