How forest models are connected to reality: evaluation criteria for their use in decision support

Abstract

Choice of a model for exploring forest management options depends on the decision space defined by the actions, indicators, ecosystem scope, and cybernetic context of the decisions. To be useful in a particular decision context, candidate models must include all relevant hypotheses of effects of the actions on the indicators in a spatial and temporal structure appropriate for the particular decision. The architecture of a suitable model is implied or constrained by these components of the decision space. A set of attributes for assessing a model's suitability for decision support is proposed. In addition to a firm foundation in science, decision support models should provide predictions with quantified bias and precision, and without artifacts that influence choice of management alternatives. Descriptions of information flow across levels of integration within and between models and between models and field observations should be included in model descriptions. Schematic diagrams of these flows illustrate several broad classes of how modelling systems may be linked to reality to improve their utility.