Evaluating CMIP6 models with a simple climatology score

When investigating multi-model ensembles it can be useful to evaluate model performance to make sure that the historical climate of the fields of interest is captured to a satisfactory degree. To this end we define a simple climatology score, based on the root-mean-square error (RMSE) of essential climate variables from the historical experiments of an ensemble of models participating in the sixth Coupled Model Intercomparison Project (CMIP6). 

We consider four key variables: near-surface temperature, precipitation, 850-hPa zonal wind, and 850-hPa air temperature. The focus is on monthly climatologies of global values, but we also explore the sensitivity of the scores to changes in the regions and seasons considered. 

The purpose of the score is to help identify models with relatively large errors in the representation of the variables of interest. This can be useful when considering models to include for storyline analysis or when selecting a subset of models for regional downscaling.