Exposure to air pollution has become an exceedingly inescapable part of urban living. The presence of pollutants in urban air is reported to cause adverse effects on human health, as as damage to structures and materials. Their presence can also be an aesthetic nuisance around urban centres. Modelling of urban air pollution is an important facet of pollution control and abatement. Models explain the occurrence, Intensity, and movement of pollutants. Current techniques used for modelling urban air pollution use dispersion characteristics to predict pollutant levels at locations away from defined sources. They work well in relatively simple situations with few sources in consideration. However, these models cannot handle the mathematics in an airshed with many point and non-point sources of urban air pollutants, This is because many simplifying assumptions have to be made, which limit the accuracy of predictions. Artificial Neural Network (ANN) modelling is a technique that offers many advantages in modelling non-linear type processes Such as those that govern complex urban air pollution situations. This technique uses historical data sets and can recognize the patterns that occur between different pollutants the urban airshed. ANN modelling is a "black-box" method, where equations describing the complex situations are not known or required. The model simply extrapolates patterns, and "learns" based on historical data, much the same,way: the human mind learns. Therefore, ANN may be suitable for air pollution modelling of situations where the governing equations are too complex or expensive to derive. This paper describes the use of ANN to predict behavior of oxides of nitrogen (NOx) within a defined region of the City of Edmonton, Alberta.
