Multivariate Analysis Applied to Air Quality Study

Abstract The air quality study in a city involves different aspects, such as emissions, physical and chemical processes, meteorology and topography, making this study highly nonlinear. In particular, Brazilian cities experience a very different mosaic from world scenario, due to the use of different fuels by our vehicular fleet. In this study, it were used a data set collected by an air quality monitoring station in the city of Rio de Janeiro, at Gávea district, between July and October of 2011, in a total of 2,240 hourly average data set. These data were treated statistically, using a descriptive and multivariate approach in order to elucidate the correlation between the variables involved. Multivariate statistical used the principal component analysis and cluster Euclidean with Ward criteria. The results showed a strong correlation between the primary pollutants nitrogen monoxide and carbon monoxide (0.71), showing that they have the same vehicular origin and nitrogen monoxide and ozone (0.66) and this with solar radiation (0.41), corroborating the photochemical formation of ozone. Other correlation is with temperature and relative humidity (0.64) and that ozone has a contribution of neighboring localities, due to the dependence of this with the wind speed (0.59).