Intercomparison, sensitivity and uncertainty analysis between different urban dispersion models applied to an Air Quality Action Plan in Tuscany, Italy

The Tuscan Regional Administration funded project MoDiVaSET-2 (MOdellistica DIffusionale per la VAlutazione di Scenari Emissivi in Toscana 2) was established in order to develop a decision support modelling system for implementing the Air Quality Action Plan for the metropolitan area of Florence, Prato and Pistoia. The objective of the work is to build an integrated meteorological and dispersion models for simulating and evaluating different future emission scenarios of PM10, NOx, and No2 in the study area. With this purpose, the project included several 1-year long dispersion modelling applications and a detailed evaluation study, including sensitivity, validation and uncertainty analysis. Several dispersion models (ADMS-Urban, CALPUFF, CALINE4, SAFE AIR II and CALGRID) were applied and evaluated against monitoring data; the intercomparison between different models is crucial in order to develop reliable modelling techniques. The obtained results point out the importance of including the following critical factors: smaller scale effects (monitoring stations are often located in complex environments; this implies a decrease in the effectiveness of validation studies) and secondary pollution (primary PM10 levels are only a small part of the total PM10 concentrations; much of the urban PM10 is actually produced by chemical transformations and other physical mechanisms, for example, resuspension). In order to understand the weight of these issues, further modelling options (full chemistry and street canyon simulations) were investigated by using CAMx and smaller scale nested models. All the factors listed above affected the evaluation work. However, this does not alter the validity of the scenario analysis, because it is based on the differences between calculated primary pollutants concentrations.