Integrating Mechanistic Modelling and Statistical Learning Theoretic-Methods: initial steps toward a strategy for model evaluation and structure selection

The traditional model validation procedure has been criticized in many areas of environmental engineering. In this paper, it is suggested to replace it by a more reasonable procedure: “model evaluation”. A framework for model evaluation and structure selection (FMESS) is presented. Based on statistical learning theory, this framework considers the model identification step as a learning problem. The model evaluation process is based on one single criterion: model performance. The latter is measured by a mathematical deviation between the model prediction and reality. Although it is an exact measure of model performance, this deviation cannot be computed, but can be related to the empirical measure that system modellers have traditionally used in the steps of model identification and validation. The relationship between the exact and empirical measures is called an uncertainty model. Two uncertainty models are presented in this paper. For these models to be valid, a set of conditions with regard to the system uncertainty needs to be satisfied. Although quite weak, these conditions may not always hold true in the case of complex environmental systems. Mechanistic information on the system’s dynamic processes would be required to ensure the fulfilment of these conditions.