Model evaluation should include qualitative as well as quantitative examinations of the model. The qualitative parts should comprise a critical appraisal of model logic as well as theoretical and biological realism of the model. The quantitative parts should comprise statistical tests and comparisons of predictions with observations independent of those used to fit the model. Comprehensive model evaluation requires several alternative approaches and criteria. Model evaluation is not one simple procedure, but consists of a number of interrelated steps that should not be separated from each other or from model construction. It is stressed that models can only be evaluated in relative terms, and their predictive value is always open to question. Thus, model evaluation is an ongoing process.
A case study with the PBRAVO growth model for maritime pine (Pinus pinaster Ait.) in the Leiria forest, Portugal, illustrates the utility of selected criteria and graphical techniques. Based on theoretical examinations and tests with data from continuous forest inventories, we conclude that the Leiria version of the PBRAVO model does not adequately represent reality and that forecasts lack sufficient accuracy for forest management purposes.