Pahs urban concentrations maps using support vector machine

The studies about health effects are often based on data inferred by monitoring stations. For this purpose, pollutants exposure maps are crucial for evaluating health effects. Studying the Polycyclic Aromatic Hydrocarbons (PAHs) and the Benzo(a)Pyrene (BaP) exposure in urban areas is the major goal of the EXPAH LIFE+ Project. An integrated approach, based on measurements and modeling techniques, was applied to simulate PAHs and BaP levels in the Rome metropolitan area. Field campaigns of PM2.5 with PAHs content were performed for the period June 2011 - May 2012, and a chemical transport model (FARM) was applied to reconstruct PAHs urban concentrations. In this work, Machine Learning methods have been applied to forecast atmospheric pollution, trying also to improve the results achieved by FARM. In particular, Support Vector Machines (SVMs) have been used. They represent one of the most common approaches among Machine Learning methods. Starting from the experimental data, SVM methods have been applied to build models able to forecast PAHs and BaP exposure. The SVM models seem to show excellent results in the reproduction of experimental data and in generalization, improving those achieved by FARM. Finally, the SVM models have produced very congruent PAHs and BaP exposure maps.