Hotspot analysis for discriminating geochemical anomalies in the soil of an intensely anthropized volcanic region in Italy.

Soils result from physical, chemical, and biological processes that affect rocks and their weathered products. In historical times, natural processes have also been widely influenced by human activity (such as industrial production, motor vehicle mobility, waste disposal, and agricultural practices). Consequently, soils represent a reservoir of chemical elements and compounds with extreme spatial variability across Earth's surface.

Defining the distribution of chemical elements and their anomalies and understanding the nature of factors controlling their spatial variability is essential for those committed to environmental issues management, especially when effects on ecosystems and living beings must be addressed, targeting the development of remedial actions.

In recent years, with the rapid data volume growth, effective methods are required for data analytics for large geochemical datasets. Spatial machine learning technologies have been proven to have the potential to reveal hidden patterns based on geochemical information. In this study, a spatial clustering technique of Getis-Ord Gi* statistic was performed on 21 characterizing elements using more than 7000 topsoil samples (~ 7300) proceeding from the Campania region territory in southern Italy.

The analysis found spatial clusters of significantly high (hot spots) and low values (cold spots) for the selected elements, showing a strong correlation with the geological features of the study area, particularly volcanic and siliciclastic units.

Volcanic units were associated with high concentrations of elements such as As, Ba, Be, Bi, Cu, Sr, Th, Tl, U, and V, while siliciclastic units were associated with high values of Co, Cr, Ni, and Mn. Additionally, the high concentration of Cd, Hg, Pb, Sb, Sn and Zn showed a clear association with the region's main urban and industrial centres.

The results highlight the power of spatial clustering techniques in discriminating geogenic from anthropogenic processes and identifying hidden spatial patterns, thus offering valuable insights for environmental studies and management.