hide
Free keywords:
-
Abstract:
In the EU-Project ENSEMBLES, a major objective was the development of a weighting system for regional climate models (RCMs) based on the ability of the RCMs to reproduce observed characteristics of important atmospheric variables. Within this context, a suite of state-of-the-art RCMs was employed for the creation of probabilistic future regional climate change scenarios for Europe. Among others, one measure to determine the individual RCM-weights is their ability to reproduce the observed trends in 2 m temperature in the reanalysis (ERA40)-driven experiments for the period 1960-2000. As the reference, we used the new ENSEMBLES observational gridded dataset for Europe (E-OBS). As additional datasets for comparisons, we also used the near-surface temperature datasets from the CRU observations and from the ERA40 and NCEP/NCAR reanalysis. The analysis was performed for the land fraction of 8 different European regions, the so-called PRUDENCE regions defined within the PRUDENCE pr
oject (http://prudence.dmi.dk). Annual and seasonal linear trends in near-surface temperature were computed for each ENSEMBLES RCM, the E-OBS dataset, and for the additional datasets mentioned above. In all regions, the computed linear temperature trends based on annual mean temperatures showed smaller values for the RCMs and the NCEP/NCAR reanalysis than for both observational datasets, and in most regions also smaller values than for the ERA40 reanalysis dataset. Depending on the magnitude of the difference in linear trends between the individual RCMs and the E-OBS dataset, skill scores were assigned to each RCM. The resulting skill scores were of similar magnitude (0.7-0.9) for the different models and regions (except for Scandinavia, which had lower skill scores around 0.6-0.8). Spatially aggregated for all of Europe, and combined from annual and seasonal into one value for each RCM, these skill scores were included in the general ENSEMBLES RCM weighting system
