Mapping and detecting land use/cover change in the tropical mountainous island of Tobago using remote sensing and geographic information systems

The Caribbean region has suffered from a severe deficiency in accurate and up to date information on land resources information. This study has used three Landsat Enhanced Thematic Mapper (ETM+) satellite images (2001, 2002 and 2003) for developing a land use/cover map of the island of Tobago to begin to fill this void. The images were corrected for geometric, atmospheric, cloud and cloud shadow and topographic effects. A supervised classification approach that employed a maximum likelihood classifier was applied. The overall accuracy of the output image was 94.7% and accuracies for the individual classes ranged from 85.7% for bamboo to 100% for urban. The kappa coefficient for the classified image was 0.93. Furthermore, a quantitative change detection analysis was performed, utilizing the classification output and a vectorised 1956 land use map of the island. The outcomes revealed a significant increase in urban sprawl contrasted by a significant decrease in agricultural land use over the 45 year period. The analysis indicated that this trend has emerged due to a direct exchange of agriculture for urban development.