674 J SCI IND RES VOL 67 SEPTEMBER 2008 Journal of Scientific & Industrial Research Vol. 67, September 2008, pp. 674-679 *Author for correspondence E-mail: goksel.demir@bahcesehir.edu.tr Prediction and evaluation of tropospheric ozone concentration in Istanbul using artificial neural network modeling according to time parameter Goksel Demir 1 *, Gokmen Altay 2 , C.Okan Sakar 3 , Sefika Albayrak 1 , Huseyin Ozdemir 1 and Senay Yalcin 4 1 Department of Environmental Engineering, Bahcesehir University, Besiktas 34349, Istanbul, Turkey 2 Department of Electrical & Electronics Engineering, Bahcesehir University, Besiktas 34349, Istanbul, Turkey 3 Department of Software Engineering, Bahcesehir University, Besiktas 34349, Istanbul, Turkey 4 Department of Energy Systems Engineering, Bahcesehir University, Besiktas 34349, Istanbul, Turkey Received 10 July 2007; revised 10 June 2008; accepted 13 June 2008 In this paper, lower tropospheric ozone concentration was modeled using artificial neural networks (ANNs) according to 1 day, 3 days and 7 days time periods to determine best prediction period. In model formation, data that was taken from ozone measuring stations and Government Meteorology Works Office was daily averages of last 6 months of 2003 and first 6 months of 2004. Air pollutant parameters (6) and meteorological parameters (8) were used in ANN architecture for Anatolian and European sides of Istanbul separately. Correlation factor was determined to examine model effectiveness for each time period. Weekly average prediction model has been observed with highest correlation factor and three day’s correlation factor was higher than daily’s. Keywords: Air pollution, Istanbul, Lower tropospheric ozone, Multilayer perceptron, Time series prediction Introduction Increase of ozone (O 3 ) concentration in lower troposphere causes green house effect, resulting in global warming. High concentrations of ozone also damage plants and trees, leading breathing difficulties for human beings. Availability of accurate information of atmospheric O 3 is essential for monitoring O 3 layer 1 . Lower tropospheric O 3 pollutants are not emitted directly into air. It is secondary pollutant that results from complex photochemical reactions between primary pollutants nitrogen oxides (NO x ) and non-methane hydrocarbons (NMHC) under sunlight in the atmosphere. Therefore, primary pollutants (NO x and NMHC) are referred as O 3 precursors 2-5 . Ozone regulates oxidizing capacity of atmosphere via production of OH radical that acts as principal cleaning agent in atmosphere 6 . In lower atmosphere, elevated O 3 may cause eye irritation, cough, reduced athletic performance and possible chromosome damage, etc 3,7,8 . Emission of some pollutant gases (NOx & NO 2 ) and volatile organic compounds (VOCs) into lower troposphere presents a health risk. Chemical interactions between these contaminants in presence of sunlight lead to appearance of O 3 in medium-high concentrations in lower troposphere. Complex photochemical formation of secondary pollutant is regulated by both natural and anthropogenic emissions and also by meteorological conditions 2 . Source of big amount of O 3 is found in lower troposphere due to industrial emissions of toxic gases. NO which is emitted from these sources turns to NO 2 in the atmosphere. With the effect of sunlight, NO 2 combines with free oxygen and causes O 3 formation. Afterwards, formed O 3 increases NO 2 amount by reacting with NO. However, specific amount of O 3 is produced by natural ways in stratosphere as the result of oxygen photolysis 3,9 . By the way a part of stratospheric O 3 is transferred to troposphere 3,6,10 (5-15 ppb). Tools predicting O 3 concentrations has been developed world over 11,12 . O 3 is a complex non-linear process. ANN has been successfully applied for atmospheric pollution forecasting of pollutants 2 (SO 2 , O 3 , and benzopyrene). In lower troposphere, therefore, ANN is a well-suited model for simulating non-linear relationships between variables 13,14 , without having explicit by their unique