LAND-USE REGRESSION MODEL TO ASSESS SPATIAL VARIATION OF TOPSOIL POLLUTION IN TARKO-SALE

A hybrid model combining land use regression (LUR) and regression kriging (RK) methods is constructed to assess the variation in spatial pollution of urban topsoil by heavy metals. The environmental monitoring data of nickel and manganese content in the topsoil of the Arctic town Tarko-Sale were used. This hybrid method of modelling topsoil pollution is suitable for all pollutants, for different territories and types of human-induced pollution sources. The use of RK improves the LUR model accuracy: the correlation between test and predicted sets increased by 7 and 17% for nickel and manganese, respectively; and the relative root mean squared error (RRMSE) decreased by 10% for both elements. The results of hybrid modeling of LUR with RK showed that the spatial distribution of manganese and nickel content in topsoil of the city does not depend on city vehicles. This points to the natural origin of manganese and nickel in urban soil in the absence of other pollution sources. The sequential inclusion of different pollution sources in the LUR model is a way to assess the contribution of each of the selected sources to pollution by the selected element. The data from technogenic sources used in the regression model did not show relationship with the pattern of manganese and nickel contamination in the topsoil. The spatial distribution of manganese and nickel in the top layer of soil is controlled rather by natural factors and is not associated with anthropogenic activities. The results of modelling LUR with RK allow us to draw conclusions about the origin of heavy metals in the soil. Previous results based on statistical analysis have shown no association between chromium pollution and anthropogenic sources (roads, industrial areas), and nickel and manganese are also not associated with anthropogenic sources. The sequential inclusion of various sources of pollution makes it possible to evaluate the source contribution to the pollution by certain metal.