Particulate matter with a diameter smaller than 2.5 μm (PM2.5 ) is a key urban air pollutant. Modeling of ambient PM2.5 is considered important for understanding its contribution sources. However, conventional models employing mathematical formulae are difficult to use for modeling ambient PM2.5 because of complexities involved in its formation and behaviour in the atmosphere.

Artificial neural network (ANN) is more suited to solving complex non-linear problems associated with formation and behaviour of PM2.5 in the atmosphere. ANN was used to model ambient PM2.5 concentrations at Fort McKay, Alberta in order to examine the potential influence of operations at the Syncrude Canada Ltd. Mildred Lake oilsand facility.

Results of ANN modeling were able to demonstrate that changes in ambient PM2.5 concentrations at Fort McKay, Alberta were, in part, due to changes in ambient PM2.5 concentrations observed at the Syncrude facility. ANN showed promise as a tool for predicting ambient PM2.5 concentrations.