Figure 2.

Diagram that compares traditional model benchmarking (a) and the assumption centered approach championed by the FACE‐MDS project (b) (from Medlyn et al., 2015; Walker et al., 2014). The assumption centered method goes beyond statistical evaluation of models' goodness‐of‐fit and diagnoses the behavior of the models in the context of their underlying process assumptions and hypotheses. The diagnosis describes model behavior in the language of science—mechanistic hypotheses and assumptions—and reconnects model results with experiment scientists. The integration of the assumption centered method into more traditional benchmarking brings models into the scientific method and is the core of the DOE ModEx philosophy (U.S. DOE, 2018)