All these models were not developed in isolation, hence each model must share some traits (sub-models, equations, etc.) with other models, thus forming families of models.

Since the models are not independent, the question arises how to select an ensemble of models that will provide unbiased climate estimates? Even if you select all available models for your ensemble (~125 in CDS), you cannot be certain that, e.g. a certain atmosphere chemistry is not over-represented.

Agreed. Models and therefore model simulations can not be considered as independent. Studies evaluating the independency of GCMs in CMIP do exist. I’m not aware of similar studies for RCMs used in CORDEX.

Two very different models giving the same result, that's convincing. Two models that differ mostly by name (since they share many sub-models) giving the same results, that is less convincing. Since I am not an expert, I cannot assess the weight of the evidence provided by different models. To me they just differ by meaningless acronyms.

Well understood. C3S as well as the next IPCC report will provide soon metadata information on the RCMs used in CORDEX, that will allow to better determine the model families. Still, not-expert users may need a simple way to identify « independent » models. Well noticed

Page tree

All these models were not developed in isolation, hence each model must share some traits (sub-models, equations, etc.) with other models, thus forming families of models.