This work carried out a geochemical study of three watersheds located in the upper course of the Paracatu River (one of the most important sub-basins of the upper-middle São Francisco River). It aimed to establish a geochemical model that can assist in identifying the sources and distribution of physico-chemical components and metals in water and sediments, assessing how the local geology and human activities influence the quality of the aquatic systems. The predominant geology of the region includes the Vazante Group, basically formed of a sequence of slate and dolomitic slate where zinc and lead mineralization occurs, and the Canastra Group (Paracatu Formation) composed of a sequence of phyllite and carbonaceous phyllite with intercalations of quartzites, where gold occurs in iron and arsenic sulfides. The cities of Paracatu, Guarda-Mor, Vazante and Lagamar, located in the study area, have, respectively, 75,216, 6,656, 18,928 and 7,710 inhabitants, with a predominantly urban population. They do not have catchment and treatment systems of sewage and garbage that serve the entire population. Water samples were analyzed using ICP/OES, AES, spectrophotometery (UV/Visible) techniques and volumetric methods. Sediment samples solubilization (fraction < 63 μm) used lithium metaborate fusion for major elements and acid attack for trace elements. The determination of sediment elements also used ICP/OES and AES techniques and the determination of minerals used XRD. The results were interpreted using multivariate statistical analysis, including principal components, multiple regression and hierarchical clustering, as well as the geoacumulation index, sediment toxicity criteria and T Test. The proposed model determined that Fe, Mn, SO 42- , and color come primarily from sulfide oxidation, HCO 3- , Ca, Mg, Zn, and Ba, from dolomitic limestone dissolution, and PO 43- , NO 3- , NH 3 and turbidity from the input of urban sewage. Indices were created for each group of parameters representing the intensity of these processes and classifying the sampling sites according to their predominant chemical composition (water with low mineralization, high carbonate, or sulphate mineralization and pollution by urban sewage). The Principal Component Analysis, interpreted together with the mineralogical analysis and geo-accumulation index, distinguished, in a multi-elemental approach, the different influences caused by mineral deposits in the sediments metal accumulation. The metals that presented higher toxicity potential were Fe, Mn, Pb and, particularly, Zn. The lower Escuro and Santa Catarina River courses showed high toxicity potential for Zn and Pb, and the upper course of the Rico River displayed high toxicity for Fe and Mn. Hg showed low toxicity, however it displayed moderate to strong accumulation showing the impact of alluvial gold mining in the upper course of the Rico River. The proposed model can be used in regional water quality monitoring, and as a reference for other locations.
