Characterization of Land Contaminated by Past Heavy Metal Mining Using Geostatistical Methods

Aspects of mining operations, including production, processing, transportation, and, finally, abandoned mine workings, can undoubtedly cause varying degrees of environmental pollution and contamination. Characterization of the spatial distribution of pollutants in contaminated soil is important for risk assessment and soil remediation. This study has been carried out using both conventional statistics and geostatistical methods which are useful for site assessment, characterization, and monitoring situations where data are collected spatially. These methods are particularly suited to cases where spatial dependence structure and contour maps of contaminant concentration or other variables are needed. The aims of this paper were to determine the extent and severity of the pollution levels on land contaminated by past mining activity through using geostatistical techniques, assess sampling strategy, and interpret data for potentially contaminated land. In the study, a total of 329 soil samples was collected at 1-, 5-, and 10-m regular grid intervals of 100 × 100-m area of Carsington Pasture in the UK and analyzed using an automatic X-ray fluorescence machine for a large number of elements. Only a few elements (Cu, Pb, Zn, and Ba) were subsequently studied in detail because of their association with the mining operation. Directional and omnidirectional experimental semivariograms of elements Cu, Pb, Zn, and Ba for 10-m grid intervals, pooled data of 10- and 5-m grid intervals (S10+5), and pooled data of 10-, 5-, and 1-m grid interval (S10+5+1) showed that neither geometric nor zonal anisotropy exits in the data. The most evident spatial dependence structure of the continuity for omnidirectional experimental semivariogram, characterized by a spherical model, of each element was obtained for the pooled data set of S10+5. Pooling 1-m grid data to S10+5 shaded the spatial dependence structure. The spherical model adopted for each element embodied no nugget effect, indicating that mineralization did not occur as nuggets in the study area, but it did show different ranges of influence and sill values. Spherical semivariogram model of each element for the pooled data set of S10+5 passed the cross validation tests. Graphs of cross validation test results proved that the isotropic spherical model, with its parameters and kriging search parameters, is representative for the area under study. The values of 10,000 grid nodes established regularly over the study area were estimated by kriging interpolation technique to smooth out the contours. Kriged maps for the Cu, Pb, Zn, and Ba elements were plotted. Anomalies such as bull’s-eyes and drift did not observed in the maps generated. Contour maps of estimated element concentrations revealed the extent and severity of contamination. Spatial distribution maps of the elements might be used in the remediation studies and help decision-makers and others involved in the abandoned heavy metal mining site in the UK.