Most models I know compare observations of mean values, standard deviation, seasonal cycle, frequency distribution and climatic extreme indicators, with results from the models, in order to assess long-term tendency of climate simulations. See for instance: Chou et al. ¨Evaluation of the Eta Simulations Nested in Three Global Climate Models¨. American Journal of Climate Change, 2014, 3, 438-454.
See also: Solman et al. ¨Evaluation of an ensemble of regional climate model simulations over South America driven by the ERA-Interim reanalysis: model performance and uncertainties¨. Clim Dyn DOI 10.1007/s00382-013-1667-2
Climate change is the result of many factors commonly referred to as human activity on earth. On the basis of the trend analysis, you can not predict the future and therefore also reduce the level of CO2. I propose methodology adopted to multidimensional models, in The Food System: a cybernetic approach which is posted on RG please read.