The use of WorldView-2 satellite data in urban tree species mapping by object-based image analysis technique

Abstract

The growth of residential and commercial areas threatens vegetation and ecosystems. Thus, an urgent urban management issue involves determining the state and the quantity of urban tree species to protect the environment, as well as controlling their growth and decline. This study focused on the detection of urban tree species by considering three types of tree species, namely, Mesua ferrea L., Samanea saman, and Casuarina sumatrana. New rule sets were developed to detect these three species. In this regard, two pixel-based classification methods were applied and compared; namely, the method of maximum likelihood classification and support vector machines. These methods were then compared with object-based image analysis (OBIA) classification. OBIA was used to develop rule sets by extracting spatial, spectral, textural and color attributes, among others. Finally, the new rule sets were implemented into WorldView-2 imagery. The results indicated that the OBIA based on the rule sets displayed a significant potential to detect different tree species with high accuracy.