A Hybrid Image Classification Approach to Monitoring LULC Changes in the Mining District of Prestea-Huni Valley, Ghana

Abstract

Mining and other anthropogenic activities are increasingly destroying forest cover in tropical forest areas of Africa, threating to deplete the entire forest reserves. These depletions not only affect the ecosystems but also have dire implications on global ecological balance and climate. Using Landsat 7 ETM+ satellite images, the study used a combined unsupervised and supervised classification methods to determine the rate of change of the various land use and land cover classes in the mining district of Prestea Huni Valley. The method produced very high accuracies with the least overall accuracy being 95.4272% with a Kappa coefficient of 0.9339. A change detection analysis revealed very significant loss of forest cover as a result of direct mining activities to be 96.78 square kilometres between 2002 and 2015. The results also suggested an overall forest cover loss rate of about 71.63 square kilometres per annum for the periods between 2002 and 2015 which poses a threat to the 493.55 square kilometres of forest cover left in the study area study, if proper monitoring and rehabilitation programmes are not put in place.