Experimental and artificial neural network approach for forecasting of traffic air pollution in urban areas: The case of subotica

Abstract

In the recent years, artificial neural networks have been used to predict the concentrations of various gaseous pollutants in ambient air, mainly to forecast mean daily particle concentrations. The data on traffic air pollution, irrespective of whether they are obtained by measuring or modeling, represent an important starting point for planning effective measures to improve air quality in urban areas. The aim of this study was to develop a mathematical model for predicting daily concentrations of air pollution caused by the traffic in urban areas. For the model development, experimental data have been collected for 10 months, covering all four seasons. The data about hourly concentration levels of suspended particles with aerodynamic diameter less than 10 μm (PM10) and meteorological data (temperature, air humidity, speed and direction of wind), measured at the measuring station in the town of Subotica from June 2008 to March 2009, served as the basis for developing an artificial neural networks based model for forecasting mean daily concentrations of PM10.