Answers are grounded in this paper and cite their sources.
The environmental modeller faces a dilemma. Science often demands that more and more process representations are incorporated into models (particularly to avoid the possibility of making missing process errors in predicting future response). Testing the causal representations in environmental models (as multiple working hypotheses about the functioning of environmental systems) then depends on specifying boundary conditions and model parameters adequately. This will always be difficult in applications to a real system because of the heterogeneities, non-stationarities, complexities and epistemic uncertainties inherent in environmental prediction. Thus, it can be difficult to define the information content of a data set used in model evaluation and any consequent measures of belief or verisimilitude. A limit of acceptability approach to model evaluation is suggested as a way of testing models, implying that thought is required to define critical experiments that will allow models as hypotheses to be adequately differentiated.
No takes yet. Share an insight, caveat, or question.
Keith Beven (2012) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: