In this study, statistical analysis and spatial distribution were performed to compare raw data and centred log-ratio (clr) transformed data of three copper (Cu), lead (Pb), and zinc (Zn) potentially toxic elements (PTEs) concentration for 550 surface soil samples in Khuzestan plain. The results of both approaches showed that classical univariate analysis and compositional data analysis are essential to find the real structure of data and clarify its different aspects. Results also indicated that spatial distributions of raw data and clr-transformed data were completely different in three studied metals. Raw data necessarily shows the effects of anthropogenic activities and needs an additional evaluation of human health risk assessment for these three studied elements. Data obtained from clr-coefficient maps also demonstrated the role of geological processes in the distribution pattern of potentially toxic elements (PTEs). To improve the understanding of the implications for PTE pollution and consequences for human health, a RGB colour composite map was produce to identify the potential origin of PTEs from areas with higher than typical baseline concentrations.