The role and rationale for modelling

Models increasingly permeate every aspect of Earth-system science. In part, this reflects the need to acquire a better understanding of likely changes in the future; but even without this imperative, models would have an essential role to play. The complex, interactive and non-linear nature of the Earth system limits the extent to which Earth-system science can progress by traditional methods in which variables are isolated experimentally, or simple cause–effect linkages defined. Linking together observations to the point where they provide a coherent view of the functioning of the complex system frequently requires the development of models that satisfy at least the following criteria:

• Internal consistency.

• Compatibility, within uncertainties, with all applicable biophysical laws.

• Compatibility with the constraints imposed by secure and relevant data, bearing in mind the uncertainties attached both to the model and to the data.

• Robust performance under a range of credible boundary conditions.

Models that satisfy these requirements can play a role in many ways similar to that played by hypotheses in the classic Popperian formulation of deductive science. In Popper's scheme of reasoning (Popper, 1963), hypotheses are developed such that they are potentially refutable through crucial tests in the form of observations or experiments. Refutation leads to the rejection or modification of the hypothesis, after which the process of hypothesis building and testing can enter a further stage using the newly gained insights. In the case of model development and testing, refutation is, in practice, usually a less rigorous and narrowly defined concept.