Artificial neural networks for modeling and forecasting ground ozone concentration

We present the ability of the Artificial Neural Network in modeling and forecasting ground ozone concentration in a high traffic area located in Bergamo. These models seem to be able to capture any kind of non linear dynamics between ozone and its previous values, its precursor, other pollutants and meteorolog- ical variables. We propose some ad hoc models to forecast the ground level ozone for medium and long term. We compare the performance of the Artificial Neural Network in modeling and forecasting ground ozone concentration with some non paramet- ric linear models recently introduced.