Application of Multivariate Statistical Analysis in Air Pollution Problems

Abstract

This study deals with the application of multivariate statistical methods to the data collected by automated air pollution monitoring stations in the territory of Prague. With respect to the results interpretation, an approach based on maximum daily values of pollutants was chosen. The method of discriminant analysis proved that the measurement of the sum of nitrogen oxides and a fraction of air-borne dust is useful. The application of cluster analysis to different stations pointed out the necessity of modification of the ozone concentration monitoring network and indicated possible release of some pollutants, such as air-borne dust, carbon monoxide and nitrogen oxide. Relations between pollutants were analysed separately for the heating and summer periods.