Clustering Methods for Spatial Assessment of Soil Contamination in Airport Areas

The article presents data on soil contamination with moving forms of Cu, Pb, Zn within a radius of 20 kilometers from Moscow Domodedovo Airport. 69 soil samples were analyzed using a flame atomic absorption spectrophotometer in ammonium citate buffer. Taking into account the morphometric characteristics of the relevant terrain, clustering was carried out using various methods; the most suitable of which turned out to be K-means, Mean Shift, DBSCAN. Various clustering algorithms were analyzed using the Davis-Baldwin Index, Dunn Index, and S_Dbw Index metrics. It has been established that the DBSCAN method combines all analyzed samples into one cluster, and therefore is not informative for the purposes of the study. The K-means and Mean Shift methods show similar results when estimating the spatial distribution of Zn. This element enters the soil not only from transport emissions, but also when applying fertilizers to fields. For Cd and Pb, the K-means method divides the territory into 2 clusters: the “polluted” one, confined to roads and the airport territory, and the “background” one. The Mean Shift method allows to identify many clusters for these pollutants; pollution is observed in the southeastern part of the airport and is explained by the presence of lateral geochemical flows.