ABSTRACT
Background: Air pollution and concerns about health impacts have been raised in metropolitan cities like Tehran. Trend and prediction of air pollutants can show the effectiveness of strategies for the management and control of air pollution. Artificial neural network [ANN] technique is widely used as a reliable method for modeling of air pollutants in urban areas. Therefore, the aim of current study was to evaluate the trend of sulfur dioxide [SO2] air quality index [AQI] in Tehran using ANN
Methods: The dataset of SO[2] concentration and AQI in Tehran between 2007 and 2013 for 2550 days were obtained from air quality monitoring fix stations belonging to the Department of Environment [DOE]. These data were used as input for the ANN and nonlinear autoregressive [NAR] model using Matlab [R2014a] software
Results: Daily and annual mean concentration of SO[2]except 2008 [0.037 ppm] was less than the EPA standard [0.14 and 0.03 ppm, respectively]. Trend of SO[2] AQI showed the variation of SO[2]during different days, but the study declined overtime and the predicted trend is higher than the actual trend
Conclusion: The trend of SO[2] AQI in this study, despite daily fluctuations in ambient air of Tehran over the period of the study have decreased and the difference between the predicted and actual trends can be related to various factors, such as change in management and control of SO[2] emissions strategy and lack of effective parameters in SO[2] emissions in predicting model