Determination of heavy metal concentrations in cultivated soils and prediction of pollution risk ındices using the ANN approach

Abstract: Heavy metal pollution in soils may have a significant impact on the environment and human health, disrupting the ecological balance in developing countries. This holds true for both industrial and agricultural endeavors. The study area, located in Vezirköprü district of Samsun province, consisted of 1664.9 hectares, from which 88 soil samples were collected from the surface (0–20 cm) depth. This study analyzed the physicochemical properties and heavy metal pollution indices, including the enrichment factor (EF), geo-accumulation index (I geo), pollution load index (PLI), contamination factor (Cf), potential ecological risk factor (Er), and potential ecological risk index (RI). In addition, the prediction accuracies of these indices were determined using artificial neural networks, considering pH, organic matter (OM), and clay contents, which affect the retention of heavy metals in soil. Based on the analysis results, the average concentration of copper (Cu) was 28.1 mg/kg, which exceeded the upper continental crust (UCC-28.0 mg/kg) and European mean soil value (ESA-17.3 mg/kg), but was below the world mean soil value (WSA) (38.9 mg/kg). The average concentration of nickel (Ni) was 40.3 mg/kg, which was higher than that of WSA (29 mg/kg), but lower than that of ESA (37 mg/kg) and UCC (47 mg/kg). The concentration of cadmium (Cd) exceeded the UCC value of 0.09 mg/kg by 0.19 mg/kg, but remained lower than the ESA and WSA values of 0.28 mg/kg and 0.41 mg/kg, respectively. The levels of other elements were found to be low compared with the UCC, WSA, and ESA results. The correlation values (R) between the actual and predicted values for PLI were higher than those for RI. During the training stage, the correlation values were 0.72 and 0.82 for RI and PLI, respectively. During the testing stage, the correlation values were 0.61 and 0.72, respectively. These results indicate that ANN can be used to predict the pollution status. Graphical abstract: (Figure presented.)