MULTIVARIATE STATISTICAL-ANALYSIS OF AEROSOL IMMISSION DATA - AN EXEMPLARY INVESTIGATION OF THE NORTH FRISIAN ISLAND OF PELLWORM

Element concentrations of 23 elements in different particle size fractions of aerosol samples from the island of Pellworm in the German Bight were analyzed by the help of total reflection X-ray fluorescence analysis. These immission data were investigated with multivariate statistical methods. Multivariate correlation analysis, as a newer chemometric method, demonstrates the autocorrelation of the immission state for different particle size fractions of dustlike aerosols. The immission of smaller particles is more strongly autocorrelated than that of larger particles. The factor analysis of each particle size fraction allows the extraction of two pollution factors for each fraction. The weights of these factors, anthropogenic and sea spray, change with size. The anthropogenic factor is more highly weighted for smaller particles, the sea spray factor is stronger in larger particles. The dependence of factor scores for smaller particles on wind direction indicates the sources of the extracted factors.