Document Type

Dataset

Publication Date

2018

Department

Biological Sciences, Geography and Geology

Abstract

Wetlands of coastal Gabon provide many ecosystems services including flood protection, water purification, and wildlife habitat. Effective sustainable management of this coastal zone is hindered by a lack of accurate wetland maps. Here we describe a novel method used to map the wetland ecosystems of nearly 100,000 km2 of wetland and upland habitat mosaic in the delta of the Ogooué River using an object-based classification of optical and radar satellite imagery based on training data collected from UAV (unmanned aerial vehicle) and a post-classification accuracy assessment using helicopter-based video. We identified 15 land cover classes, of which nine were wetland. A stratified random sample accuracy assessment of the final classification yielded an overall accuracy of 0.80. Despite the important role that wetland habitats play for wildlife and ecosystem functioning across the region, our results indicate these wetlands cover only 22% of the project area. As expected, most of the wetland habitats are found close to major water bodies, including the Ogooué River, estuaries near the cities of Libreville and Port Gentil, and coastal lagoons to the south of these cities. When considering the six Wetlands of International Importance designated under the Ramsar Convention within the project area, only 33% of mapped wetlands fall within the Ramsar site boundaries, and only 10% of mapped wetlands fall within official protected areas. Furthermore, within the Ramsar sites, only 31% of the land cover was classified as wetland. In order to better manage these wetland resources, more effective Ramsar boundaries would include the extensive wetland habitats found along the coast from Port Gentil south to Loango National Park. These data are now available for Ramsar managers to improve wetland management within the currently designated Ramsar sites, for example by protecting wetland connectivity and other ecosystem processes.

Comments

DOI: 10.18785/gclc.ds.01

Download

Share

 
COinS