Integration of Geochemical Analysis and K-Means Clustering for Sustainable Management of Various Dam Sediments

Abstract

Optimizing the use of existing dams can reduce the need for new construction and support sustainable dam management. In this study, key sediment parameters including humic acid (HA), fulvic acid (FA), %C, %H, %N, total organic matter (TOM), pH, conductivity, and shrink/swell capacity were analyzed. Heavy metal concentrations ranged from 1.62 to 7.74 mg/kg (As), 1.40-2.91 mg/kg (Cd), 6.79-18.44 mg/kg (Co), 19.46-85.61 mg/ kg (Cr), 21.12-63.60 mg/kg (Cu), 8000-46,500 mg/kg (Fe), 260-1120 mg/kg (Mn), 27.12-180 mg/kg (Ni), 2.52-10.22 mg/kg (Pb), and 30.50-88.10 mg/kg (Zn). Organic material contents were 0.050-0.88 % for HA and 0.01-1.21 % for FA. Measured pH values ranged from 6.99 to 7.92, conductivity from 0.26 to 4.49 mS/cm, and shrink/swell capacity from 34.37 to 54.11 %. The dataset was normalized using Min-Max scaling to ensure consistency and reduce bias. K-means clustering was applied to identify sediment profiles, yielding insights into pollution levels, soil fertility, and retention capacity. The integration of geochemical analysis with artificial intelligence (AI)-based clustering demonstrated the effectiveness of machine learning (ML) methods in classifying sediments based on heavy metal concentrations. Additionally, SEM analysis revealed distinct layered surface properties with nanoglobular structures ranging from 100 nm to less than 10 nm, offering further insights into the sediment characteristics and potential agricultural applications. This study underscores the importance of integrating AI techniques with traditional analyses to enhance sediment characterization and promote sustainable environmental management.