The most important guide for designing a model is the question addressed. This question, or problem, allows us to decide, in an experimental way, which element of the real system to represent in the model and at which resolution. With POM, in addition we are asking: what patterns can we observe? If we agree that a certain pattern is typical, or even essential for describing the system's identity, we should chose a model structure that in principle allows the same pattern to emerge in the model. This means in particular that we have to include the state variables of the pattern. If, for example, the pattern is spatial, we need to include spatial variables in the model; if the pattern is in size distributions, we need to include size as an individual's state variable; if the pattern is a response to a disturbance event, for example, a drought, we need to include soil moisture as an environmental state variable, etc. Patterns thus make the choice of the model structure and complexity less arbitrary and more directly linked to the internal organization of the real system. The task of the modeler then is to check whether the model is able to reproduce the observed patterns. If not, key processes might still be missing in the model.
Was this article helpful?