Atmospheric deposition; Biomonitoring; Cluster analysis; Moss

The article introduces the possibility of identification of pollution origin using cluster analysis provided by self-organizing (Kohonen) map. The pollution was detected in naturally growing moss to monitor the atmospheric deposition in the Czech-Polish borderlands (Moravian and Silesian regions). The method was applied on the dataset of 44 samples collected from 41 sites where the content of 38 elements was determined using Neutron Activation Analysis. The method was used for the classification without any data transformation necessary and any information loss, so determined clusters hold the information about chemical content needed for the identification of the pollution origin. The method classified biomonitoring data into four clusters with specific chemical composition. Now it is possible to correlate them with emission to determine the origin pollution. The applied method is recommended for the employment after further testing.