An illustration of the use of an approach for treating model uncertainties in risk assessment

This paper discusses an approach for treating model uncertainties in relation to quantitative risk assessments. The analysis is based on a conceptual framework where a distinction is made between model error—the difference between the model prediction and the true future quantity—and model output uncertainty—the (epistemic) uncertainty about the magnitude of this error. The aim of the paper is to provide further clarifications and explanations of important issues related to the understanding and implementation of the approach, using a detailed study of a Poisson model case as an illustration. Special focus is on the way the uncertainties are assessed.