CO dispersion models for signalized road intersections: Gaussian vs. empirical approach

A comparison has been made between a Gaussian and an empirical approach to model carbon monoxide (CO) atmospheric dispersion at a signalized road intersection. Starting from the Webster queue and the Kunselman modal emissions algorithms both implemented in the APRAC/3 model, the MODEM emission model has been applied for estimating dynamic emission rates along each link connecting to the intersection. As regards the Gaussian dispersion approach, both the algorithm implemented in the CALINE/4 model as well as the one in the APRAC/3 model have been applied. Their performances have been compared with an empirical approach developed by the authors which was based on a simplified box model applied to the intersection area. Experimental data measured in Firenze (Italy) have been used for calibrating each dispersion model to the study case. The calibrated models have then been applied over a different time period in order to test their performances. As a final result the empirical box model proved to be more accurate than the Gaussian models.