Abstract : 1B.2
Numerical air quality modelling along the Brenner South route within the ALPNAP project

Gianluca Antonacci, Massimiliano De Franceschi, Dino Zardi
gianluca.antonacci@cisma.bz.it
CISMA srl

Results of numerical simulations related to a field campaign within the project ALPNAP (funded by the European Regional Development Fund within the Interreg III B initiative) and performed during winter 2006 along the Adige Valley (Northern Italy) are presented. The project aims at simulating the pollutants dispersion along one of the highway routes crossing the Alps. Intensive observations data (de Franceschi et al.) have been used for model calibration, with special emphasis on vertical temperature profiles and turbulence parameters. SODAR data have been used as upper air sounding for an initial-guess wind field. The CALMET model has been used for simulation of the meteorological fields over a domain of 40x95 km, with an horizontal resolution of 250 m. The computed parameters are the 3D field wind and temperature, along with 2D stability parameters. Particular attention was paid on wind direction and temperature profile estimate over complex topography. The second step, the simulation of the dispersion processes, was carried out by using the Eulerian-Lagrangian model CALPUFF which can usefully take advantage of all the meteorological parameters estimated through CALMET, also including a complex-terrain module. The results consist of PM10 and NOx concentration fields within three target areas of about 15x10 km each, which have been identified to highlight to what extent the local meteorology is influenced by the topography and how local features affect the transport of pollutants emitted by the motorway.

Every simulation has been performed by using hourly emission factors, estimated in accordance with measured traffic flows. The comparison between model results and observed data shows, on average, a good agreement with only some underestimations of the concentration during traffic peaks very close to the source. This is a general issue concerning this kind of models and can be partially explained by the model spatial resolution vs. the single point measurements.