2. An Application of Artificial Neural Networks for the Prediction of Surface Ozone Concentrations in Malaysia

In this study, an artificial neural network (ANN) was used to extract the complex relationships among divergent parameters that have the capabilities to predict O₃ concentrations which serve as an input to meteorological conditions and precursor concentrations. The ANN was trained using samples of daily maximum data provided by the Malaysian Department of the Environment (DOE) over a period of 9 year (2003–2011) in the towns of Gombak and Shah Alam in Malaysia. Furthermore, surface O₃ concentrations from the two locations (Gombak and Shah Alam) were estimated using surface meteorological variable as predictors for the ANN. Based on the results, it can be deduced that the relationship between the parameters and the O₃ concentrations are highly complex and non-linear. Analysis of the regression based model results between Gombak and Shah Alam were evaluated using the ANN. Based on the sample results it was confirmed that Shah Alam has the highest regression result of R = 0.64 in comparison with Gombak station. The inference drawn from this study shows that neural network model consistently gives superior predictions.