Spatial modelling of hazardous elements at waste dumps using geostatistical approach: a case study Sarcheshmeh copper mine, Iran

• Published in: Environmental earth sciences 2017-08, Vol.76 (15), p.1, Article 532
• Main Authors: Hasani, Shabnam, Asghari, Omid, Doulati Ardejani, Faramarz, Yousefi, Saeed
• Format: Article
• Language: English
• Subjects: Agglomeration; Aggregation; Biogeosciences; Cadmium; Case studies; Chromium; Coefficients; Components; Computer simulation; Copper; Correlation coefficient; Correlation coefficients; Data processing; Distribution; Earth and Environmental Science; Earth Sciences; Environmental assessment; Environmental hazards; Environmental Science and Engineering; Geochemistry; Geology; Geostatistics; Hazardous areas; Hazardous wastes; Hydrology/Water Resources; Kriging interpolation; Landfills; Lead; Manganese; Mine wastes; Mining; Modelling; Molybdenum; Nickel; Original Article; Outliers (statistics); Principal components analysis; Regression analysis; Regression coefficients; Reliability; Simulation; Soil; Spatial distribution; Statistical analysis; Statistical methods; Terrestrial Pollution; Toxic wastes; Variance analysis; Zinc
• ISSN: 1866-6280; 1866-6299
• DOI: 10.1007/s12665-017-6852-x

Mining activities and resulting wastes can be considered as one of the most important sources of hazardous elements in the environment. Knowledge of the spatial distribution of toxic elements in waste dump systems is necessary to assess environmental hazard and strategy. To achieve this goal, this paper investigates spatial distribution of toxic elements using statistical and geostatistical analysis. A total of 58 soil samples were collected, and the amount of As, Cd, Co, Cr, Cu, Mn, Mo, Ni, Pb and Zn was then determined at “Sarcheshmeh” copper mine waste dumps. In order to evaluate the presence of multivariate outliers, Mahalanobis distance technique ( D 2 ) was applied and the multivariate outlier samples were removed. This resulted in an increase in correlation coefficient. To reduce dimension of data set, principal component analysis was applied and four principal components were determined which indicate 83.463% of the total variance of data set. Estimated PCs together with the toxic elements maps based on the ordinary kriging display aggregation of toxic elements in some parts, and validity of predictions was evaluated using the leave-one-out cross-validation method. The regression coefficients of estimated and observed values presented the reliability of the kriging estimates. Sequential Gaussian simulation method was applied for principal components due to similar results of estimated principal components and toxic elements. The results of simulation maps are almost identical to estimated outcomes.