Application of artificial neural networks for classification of uranium distribution in the Central Rand goldfield, South Africa
Source: Environmental Modeling and Assessment, Volume 10, Number 2, June 2005 , pp. 143-152(10)
Abstract:Mine tailings generate significant environmental impacts and contribute to water pollution. The Central Rand goldfield, South Africa is replete with gold mine tailings which have contributed significantly to water pollution as a result of acid mine drainage (AMD). Water quality is affected by mine tailings and spillages, especially from active slimes dams, currently reprocessed tailings, as well as footprints left behind after reprocessing. The release and distribution of uranium from these sites was studied. Correlation matrices show a strong link between different variables as a result of AMD produced. Principal component analysis (PCA) was used to identify very influential variables which account for the pollution trends. Artificial neural networks (ANN) using the Kohonen algorithm were applied to visualise these trends and patterns in the distribution of uranium. High concentrations of this radionuclide were detected in streams in the vicinity of the tailings dumps, active slimes and reprocessing areas. The concentrations are reduced drastically in dams and wetlands as a result of precipitation and dilution effects.
Document Type: Research article
Affiliations: 1: School of Chemistry, University of the Witwatersrand, P. Bag 3, WITS 2050, Johannesburg, South Africa, 2: School of Chemistry, University of the Witwatersrand, P. Bag 3, WITS 2050, Johannesburg, South Africa, Email: firstname.lastname@example.org 3: Department of Analytical Chemistry, Faculty of Science, Masaryk University, Kotlarska 2, 611 37, Brno, Czech Republic,
Publication date: 2005-06-01