Comparison of dispersion models for vehicular air pollutants in urban areas

The temporal increasing trend of the vehicle fleet has made traffic the major air pollution source mainly in urban areas. The assessment of air pollution is realised by air quality monitoring networks and in parallel by air pollutant dispersion models. There are many models used for simulating the air pollutants dispersion. These models are either large or small scale models. An attempt was made to evaluate nine street canyon models (small scale), that simulate the air quality at six specific roadside sites. The models and the application sites are the Operational Street Pollution Model (OSPM) at Helsinki, Finland, the Indic Airviro, system at Singapore, the Local Scale Dispersion Model (VADIS) at Lisbon, Portugal, the RNG model at Fung-Shan, Taiwan, the models APRAC, GZE, CALINE4 and PWILG at Guangzhou, China and the Atmospheric Dispersion Modelling System (UK-ADMS Urban) and Indic Airviro at Leicester, United Kingdom. The evaluation took part with the aid of several means such as a review of the models' set up and statistical analysis. The air quality models like the ones reviewed within this work that have been designed to model point, line and area sources, are not able to adequately model emissions from congested urban areas, since the statistical analysis review showed large differences between modelled and monitored data. Nevertheless, between the models that are reviewed in this study, the ones that seem to model air pollution best are the OSPM, the RNG and the Indic Airviro. On the other hand, ADMS seems to be the less adequate in NO2 modelling, but is on the average for CO modelling. VADIS, APRAC, GZE, CALINE4 and PWILG are models that cannot describe pollution dispersion as adequately as the first three. The study concentrated on three of the main pollutants: carbon monoxide (CO), nitrogen oxides (NOx) and nitrogen dioxide (NO2) and used the street canyon model for calculation of the air pollutant dispersion in the urban environment. CO is generally better modelled than NO2 at least from most of the models that were used in this study.