Land cover change detection analysis for the Atok area with special emphasis on chrome mining development

Abstract

The main focus of this research study was to quantify temporal land cover changes (LCC) around the Atok area in Sekhukhune District, Limpopo Province, over a period of four years. The influence of chrome mining expansion on land cover changes was mapped using Sentinel 2 imageries acquired in 2016 and 2019 using the Maximum Likelihood Classifier (MLC) and Support Vector Machine (SVM) classification algorithms. Normalized difference vegetation index (NDVI) was used to establish correlation between the change analysis results with clas sification results. The overall accuracies of the 2016 and 2019 Sentinel-2 image classifications using the MLC were 83% (with kappa value of 0,77) and 80% (with kappa value of 0,75) respectively, while those of the SVM classifier were 84% (with a 0,78 kappa coefficient) and 81% (with a 0,76 kappa coefficient) respectively, showing that the SVM outperformed the MLC by a 1% margin. The chrome mining operations grew by 1,6% from an average of 2,3% in 2016 to 3,9% in 2019, which is an increase of 314.5 hectares over the period of four years. Vegetation cover decreased by an average of 3,3% (648,6 ha) during this period. A moderate increase of 3,8% was recorded for dry areas, owing to exposures due to stripping and open cast mining in the new chrome mining operations. Further analysis of land cover types revealed the inter-relations of the change scenarios between the chrome mining class and other land cover types available in the study area. NDVI results highlighted areas of increased dry soil exposures and areas with major losses in vegetation cover, which are attributable to development of new chrome mining oper ations. Mapping of land cover change using Sentinel-2 imageries has proven robust in effec tively highlighting the impact of chrome mining activities and its expansion on the local land scape quality and environmental degradation within the area.