Remote sensing of land cover change within granite quarries

M.Sc. (Environmental Management)Abstract: Granite quarrying, like any other mining activity in South Africa, contributes significantly to the economy and job creation in the country. However, the adverse environmental impacts caused by quarrying activity cannot be neglected; therefore it is important to monitor and manage such impacts. Remote sensing offers powerful tools for efficient means of mapping and monitoring environmental impacts related to mining activities, due to the fact that it is cost-effective, covers large spatial areas, comprises of data at various spatial resolutions and provides multi-temporal data that are suitable for time-series analysis. The aim of this study was to evaluate the use of Landsat and ASTER data in monitoring land cover changes caused by granite quarries between Rustenburg and the Madibeng Local Municipalities in South Africa. Data used in the study included a series of Landsat data from 1973 to 2015 and ASTER data from 2001 to 2013. A supervised classification was run on multispectral and normalized difference vegetation (NDVI) index derived from each data. Classification accuracy was assessed using samples extracted from temporally coincident Google Earth images. A post classification change detection method was adopted to quantify changes in granite quarries over time. Landsat-based classification yielded an overall accuracy of 75% with the kappa coefficient of 0.71, while the overall accuracy of classification from ASTER data was 79% with the kappa coefficient of 0.75. The slightly better overall accuracy and kappa coefficient obtained from ASTER was due to the higher spatial resolution it has, especially in the visible and near infrared regions of the electromagnetic spectrum. Landsat data from the year 1973 to 2015 revealed a significant increase in granite quarries (an increase of 907.4 ha) and quarry lakes within the quarries (an increase of 2.1 ha). A considerable decrease in bare land (a decrease of 119 ha) and vegetation (782.1 ha) was also observed during the same period. ASTER data analysis between 2001 and 2013 revealed an increase in granite quarries (an increase of 531.9 ha) and quarry lakes (an increase of 2.2 ha), while a significant decrease in bare land (a decrease of 62.1 ha) and vegetation (a decrease of 467.7 ha) was observed. NDVI results obtained from Landsat data indicated variation in green vegetation cover with mean, NDVI value ranging from 0.15–0.54 in the 1998 image followed by the 1973 image (0.17–0.43), then 2015 (0.09–0.34) again in 1986 (0.09 – 0.25). Mean NDVI values for ASTER data showed a decrease in green vegetation cover from 2001 to 2013 with 95% of granite quarries in 2001 having mean NDVI value between 0.25–0.3 and 90% of granite quarries in 2013 with mean NDVI of below 0.11. The findings of this study showed that remote sensing techniques can be used successfully in detecting land cover..