Evaluation of three classifiers in mapping forest stand types using medium resolution imagery: a case study in the Offinso Forest District, Ghana

Abstract

Loss of Ghana’s natural forests has been counteracted by plantation establishment and development. As at 2003, Ghana had a plantation area of about 97,000 hectares comprising different tree species. With the rapid expansion of plantations in the country, it is anticipated that major managerial challenges will arise due to insignificant technical personnel for monitoring and management. The application of GIS and remote sensing will be a powerful intervention and tool for monitoring and managing these forest resources in the area. The aim of this study was to determine a suitable method of mapping the different forest stand types using medium resolution images in the study area. Three classifiers were examined for their suitability in mapping the different forest stand types in the area (maximum likelihood, spectral angle mapper and decision tree). The results showed that using maximum likelihood classifier and ASTER imagery, different forest stand types can be accurately mapped with an overall accuracy of 88.50%.

Key words: Geo-referenced, classification, algorithm, spectral, mapper.