The use of observations, or in some cases a reference model, during model development to ensure that the model output compares favourably with the properties of the system being modelled.