Setting Acceptance Criteria for Air Quality Models

Abstract

Is an air quality model acceptable for use (fit for purpose)? We now have several state-of-the-art statistical methodologies in place for calculating model performance measures and for determining the statistical significance of the results. But a more difficult question is whether, based on the calculated performance measures, the model is acceptable or not. This question is being addressed in several of the authors’ recent studies, and an example carried out for the Joint Effects Model (JEM) is the focus of the current paper. The rationale for selecting acceptance criteria for air quality models is described, and the results of applications of the method to four urban field experiments are presented. The values of the acceptance criteria are based on the authors’ experiences with a wide variety of air quality models and observations. For example, an acceptable FAC2 is>0.3 for urban scenarios. Furthermore, since a model is unlikely to fulfill all acceptance criteria at every field experiment site, we require that the model meet the individual criteria for at least 50% of the performance measures and field experiments and input options used in the study. It is shown here that the JEM model meets the 50% criterion.