Classification of environmental geochemical data using discriminant analysis and neural network in carbonate-sulfide waste dumps of lead and zinc mines

Heavy metal pollution is one of the main environmental problems associated with mineral, industrial and agricultural activities in the world. Due to the long life of these toxic metals and the degree of solubility in acidic and even non acidic conditions, they can have destructive effects on water, soil and humans life in the long time. The lead-zinc mine of Anguran is one of the largest world-class sulfide-carbonate reserves, which, due to mining, has produced a significant amount of mineral wastes, which can be the source of heavy metals to the water and soil of the downstream areas. Therefore, the use of quick and cost-effective methods to classify the risk of contamination of this type of waste can be a useful tool for monitoring, recovery and reconstruction programs in the future. The purpose of this research is to use multivariate statistical techniques, such as discriminant analysis (DA) method and utilize of artificial intelligence technique in order to classify and predict the potential of pollution in mining wastes. To achieve the goals, the samples taken from different surficial parts of the waste dump were analyzed by using the ICP-MS method to determine the concentration of heavy metals. These metals includes: Al, As, Cd, Co, Cr, Cu, Fe, Mn, Mo, Ni, Pb, Sb and Zn. The environmental risk assessment indices (ERAI) and potential load index (PLI) were then modelled. At the end, by using the DA and neural network (NN) methods, potential risk of contamination on the damp surface was classified in three low, medium and high levels with the accuracy of 91.49 and 93.6 percent respectively. The results also showed that these techniques can be used as effective tools for classifying new waste dumps and designing of new constructing dumps based on their contamination level.