Introducing a New Method for Evaluating Performance of Global Climate Models and Ranking them

Over the past decades, more than 40 global climate models have been developed in scientific centers around the world to simulate and predict the world climate change. These models are different depending on the initial and boundary conditions, the variables used in the climate, and the structure. Therefore, to use these models to predict future climate in each region, systematically evaluation performance of these models in simulating the time series of observed climatic parameters such as daily, monthly and annually temperature and precipitation will be required. Hence, in this study, an innovative systematic approach for evaluating the performance of 36 global climate models from the Fifth Intergovernmental Panel on Climate Change report on the southern slopes of Alborz Mountains is done. Based on the results, applying the innovative method will have incomparable accuracy in the right choice of general circulation model in each region. The results also indicate the inappropriate accuracy of most models in simulating temperature and especially historical precipitation in the study area (as in other valid studies in the world) and even in the top models of the study area the historical monthly precipitation simulation was not appropriate. Finally based on five statistical properties of Nash–Sutcliffe, bias, correlation, root mean square error and mean absolute error, ACCESS1.0 and GFDL-CM3 models with higher priority and CNRM-CM5 and GFDL-ESM2G models are proposed in the next priority for climate change studies and future temperature and precipitation values prediction.