Eyring, V., and co-authors, 2007: Multimodel projections of stratospheric ozone in the 21st century. J. Geophys. Res., 112 , D16303. doi:10.1029/2006JD008332. Values (all units) of the first three main components (PC1 to PC3) compared to the individual models of the P-E-C multimodel assemblies for JJA temperature, DJF temperature, JJA precipitation and DJF precipitation. The gray shade (see legend) distinguished the model subsamples, while Inserts showed the variance in each sample normalized by the variance for the set Well co-with the observations in one metric does not guarantee good performance in other variables, but the performance correlations on the variables at least within one component of the climate system are quite high, because many variables by the same processes and settings influence it. Models that represent a few basic variables, such as temperature and precipitation, often perform well in other variables (e.g.B Gleckler et al. 2008). The proportion of the total variance in the forecasts of the average temperature of the Decaden surface air, explained by the three components of the global uncertainty, is shown for the bottom 48 countries (similar results are not shown for Hawaii and Alaska). Orange regions represent human uncertainty or scenarios, blue zones for model uncertainty, and green zones for the internal variability component.

As the size of the region increases, the relative importance of internal variability increases. In interpreting this figure, it is important to remember that it shows the sources of uncertainty breakdown. Total insecurity increases over time. (Image source: Hawkins and Sutton 200998). In summary, this section highlights the difficulty of weighting models on the basis of observations. The regional simulations were transient, covered the entire integration period from 1951 to 2100 and used a common integration zone with a distance of 25 km. In addition, this project defined the precise configuration of the surfaces, with the exception of a few exceptions necessary due to the peculiarities of the map projection used by some regional climate models. Flow changes are projected using a hydrological model provided by 42 CMIP5 GCM.

Christensen JH, Kjellstrom E, Giorgi F, Lenderink G, Rummukainen M (2010) Weight assignment in regional climate models. Clim Res 44:179-194 This article presents future climate and flow projections for the South Asian region in the RCP8.5 scenario with climate change, which is informed by 42 CMIP5-GCMs. The flow is projected for grids at 0.5°, using hydrological models using future climate stimuli obtained through the empirical scale of historical climate series. For the DJF temperature, only the third PC is considered very inhomogeneous in the various modelling projects in Table 2. This model shows a north-south slope (fig. 5 line 2). A sub-quantity of CORDEX for both resolutions shows above-average warming at the northern periphery of the domain, while PRUDENCE has values below the overall mean (Fig. . . .