Circular classification of air quality regimes from imperfect wind-direction measurements

Air pollution is a very critical issue that directly impacts our quality of life. It occurs when air pollutants are present in the atmosphere at high concentrations, surpassing normal levels and significantly degrading our living conditions. Considering the impact of air pollution on human and environmental health, it is crucial to investigate about the levels of air pollutants and their dispersion in the different sites. This phenomenon is strictly related to wind direction because wind is responsible of the dispersion, concentration, and transportation of pollutants in the atmosphere. In this context we pursue a classification task of circular data observed with measurement error and apply a nonparametric statistical discrimination rule to a real dataset involving wind direction and two specific air pollutants. We distinguish two levels of air quality according to the World Health Organization’s Air quality guidelines and take into account the amount of particulate matter and nitrogen dioxide pollutants.