Abstract
Keywords: urban air pollution, traffic model, incomplete information, fuzzy neural network.
Pollutant concentrations in urban canyon mainly come from vehicular emissions. The search of a quantitative relationship, linking pollutant emissions from mobile sources to the resulting concentrations, at different receptor sites, has brought us to develop a phenomenological model able to calculate the number of running vehicles once the pollutant ground level concentration of CO is known. Unfortunately, we haven't the possibility to test this model by a direct inspection of the traffic flow. Despite this fact it is our aim to validate the proposed traffic model. Mathematical models are considered useful tools in supporting air quality evaluations. The structure of the models usually embodies the main aspects of the phenomenon and is supposed to provide information about the relationships between pollutant emissions and concentrations. As it is well known, there exist several typologies of models, differentiating each other by levels of complexity and fields of application. In this work an early simulation has been performed by using a commercial available pollution model. First comparisons of the CO concentrations, obtained by using as input of the mathematical model the number of vehicles predicted by the phenomenological model, against the measured data of CO concentrations, show that the proposed traffic model realistically reproduces the characteristic of the traffic flows, at least in the selected town context. The feasibility of the model to be applied in other urban situations that one selected for the validation is also discussed in the paper.