Exploration of the spatially varying relationships between lead and aluminium concentrations in the topsoil of northern half of Ireland using Geographically Weighted Pearson Correlation Coefficient

The investigation of spatially varying relationships provides a novel and effective way to identify the influencing factors and potential sources of potentially toxic elements (PTEs). The varying relationships for PTEs at different sampling locations can be associated with the different controlling factors from geological units or pollution sources, which benefits environmental studies and management for local government. Generally, the concentrations of lead (Pb) and aluminium (Al) in the soil are reported as being positively correlated under most natural conditions, due to Pb and Al being enriched in clay minerals. However, the original positive relationship can be masked by external influences at the local level, thus the resulting varying relationships can be observed and associated with the potential pollution sources. In this study, Geographically Weighted Pearson Correlation Coefficient (GWPCC) was applied to explore the spatial relationships between Pb and Al in soils. Both positive and negative correlation coefficients were observed from the results of GWPCC at different scales with bandwidths ranging between 43 and 300 nearest neighbours, suggesting the relationships between Pb and Al are spatially varying at the regional scale. Negative correlations were observed in more than 35% of the study area, mainly clustered in the western and north-eastern of Ireland. The positive correlations were found in the central-western and midlands, while comparatively mixed correlations occurred in the eastern coastal areas. The spatial patterns for the majority of negative correlations showed a clear association with blanket peat, which likely can be attributed to long-distance transportation of Pb through atmospheric deposition. Moreover, anthropogenic activities weakened the relationships in the eastern coastal areas. These results demonstrated the efficiency of GWPCC in exploring the spatially varying relationships between environmental variables and identifying associations with related influencing factors, which could be hardly achieved by traditional techniques.