Large-scale atmospheric models as an experimental tool

The utility of large-scale numerical models for exploring the properties of the atmosphere is discussed. Thus the assessment of model simulations of the atmosphere is not of concern here, but rather the application of the models to clarify why the atmosphere actually has its observed characteristics. In addition the possibility of performing null experiments with the models in order to evaluate links in a physical hypothesis is emphasised. The technique is illustrated with examples from experiments involving a sun-weather mechanism, a rotation rate variation, removal of mountains, and modifications to cloud amounts. Considerable potential exists for exploitation of this technique in the future.