Application of artificial neural networks to modeling and prediction of ambient ozone concentrations

The deterministic modeling of ambient O-3 concentrations is difficult becau se of the complexity of the atmospheric system in terms of the number of ch emical species; the availability of accurate, time-resolved emissions data; and the required rate constants. However, other complex systems hare been successfully approximated using artificial neural networks (ANNs). In this paper, ANNs are used to model and predict ambient O-3 concentrations based on a limited number of measured hydrocarbon species, NOx compounds, tempera ture, and radiant energy. In order to examine the utility of these approach es, data from the Coastal Oxidant Assessment for Southeast Texas (COAST) pr ogram in Houston, TX, have been used. In this study, 53 hydrocarbon compoun ds, along with O-3, nitrogen oxides, and meteorological data were continuou sly measured during summer 1993. Steady-state ANN models were developed to examine the ability of these models to predict current O-3 concentrations f rom measured VOC and NOx concentrations. To predict the future concentratio ns of O-3, dynamic models were also explored and were used for extraction o f chemical information such as reactivity estimations for the VOC species. The steady-state model produced an approximation of O-3 data and demonstrat ed the functional relationship between O-3 and VOC-NOx concentrations. The dynamic models were able to the adequately predict the O-3 concentration an d behavior of VOC-NOx-O-x system a number of hourly intervals into the futu re. For 3 hr into the future, O-3 concentration could be predicted with a r oot-mean squared error (RMSE) of 8.21 ppb. Extending the models further in time led to an RMSE of 11.46 ppb for 5-hr-ahead values. This prediction cap ability could be useful in determining when control actions are needed to m aintain measured concentrations within acceptable value ranges.