Prediction of Ground-Level Air Pollution Using Artificial Neural Network in Tehran

Abstract

Novel technologies and subsequent pollutions are serious threats to the environment and public health. The environmental pollutions, especially air pollution, are currently leading environmental concerns in developing countries, including Iran. In the present study, the air quality and meteorological data were employed to achieve potent models based on an Artificial Neural Network (ANN) for the prediction of air pollution in Tehran, Iran. The developed models manage to predict daily concentrations of various air pollutants such as O3, PM10, NO2, CO, and PM2.5. The required data were collected daily through the Air Quality Organization from all air quality stations of Tehran within a four-year period (from 2012 to 2015). Training the models was on the basis of Multi-Layer Perceptron (MLP) with the Back Propagation (BP) algorithm using MATLAB program. The results indicated appropriate agreement between the observed and predicted concentrations, as the values of the coefficient of multiple determinations (R2) for all models were more than 0.83. In conclusion, the studied meteorological parameters are effective on all pollutants concentrations.