Séminaire ISPA Pr Belinda E. Medlyn, University of McQuarie, Sydney - salle de réunion ISPA ex EPHYSE, La Ferrade

Process-based models of forest ecosystems are widely used to predict the likely future impacts of environmental change. However, models frequently differ in their predictions, leading non-modellers to ask which model is “best”. I argue that we should consider models not as entities, but as collections of assumptions about individual processes, and we should instead be asking which assumptions are best. Model predictions can and should be understood by relating them back to their underlying assumptions. Understanding the effect of individual assumptions allows us identify major uncertainties in the models and work towards improving them. I will show how this assumption-centred approach to modelling is being used to improve our understanding of the effects of rising CO₂, temperature and water availability on forest function.