Integrating self-organising maps and Monte Carlo probabilistic model for groundwater pollution source apportionment and risk analysis at Kokompe in southwestern Ghana

Abstract Groundwater is a key water source for many Ghanaian communities, yet it remains highly vulnerable to pollution from human-induced activities, particularly in urban areas such as Kokompe, where domestic and industrial wastes are indiscriminately disposed of. Addressing such a menace requires a robust framework for a comprehensive assessment of the resource. This study collected thirty (30) groundwater samples and analysed them for Ca2+, Mg2+, Na+, K+, Cl−, SO4 2−, HCO3 −, NO3 −, Fe, Mn, Zn, Cr, and Pb. Weighted average water quality index (WAWQI) and heavy metal evaluation index (HEI) were employed to assess groundwater quality and the extent of pollution; self-organising maps (SOMs) and Monte Carlo Probabilistic model (MCPM) were utilised to identify pollution sources and quantify related human health risks. HEI and WAWQI revealed that the groundwater in the area is highly polluted and unsuitable for human use. Combining principal component analysis and SOMs revealed three clusters: Cluster 0 (Zn), Cluster 1 (Ca2+, Mg2+, Na+, K+, Cl−, SO4 2− ,and NO3 −), and Cluster 2 (Fe, Pb, Mn, Cr, and HCO3 −). Cluster 0 suggests metal pollution from man-induced activities, whereas clusters 1 and 2 showed the influence of combined geogenic processes and man-induced activities. MCPM revealed that approximately 4.87% of the adult and 7.71% of the children populations are at non-carcinogenic risk, with Pb contributing largely (55.2%) to the non-carcinogenic risk, followed by Mn (31.5%) and then Cr (9.0%). This study demonstrates that integrating SOM and MCPM offers a robust framework for appraising groundwater pollution in convoluted hydrogeochemical terranes. The findings offer invaluable insights for regulatory bodies to implement targeted pollution control measures and safeguard public health at Kokompe.