Since October 2007, Bristol Conservation and Science Foundation (BCSF) has been working on a pilot project based in Anjouan to develop the best approach to forest management for the Comoros. The project is working in international partnership with Durrell Wildlife Conservation Trust (Durrell) and the University of East Anglia (UEA), and locally with the Comorian Government, the University of the Comoros, and the Centre National de Recherche et Documentation Scientifique (CNDRS).

The project has three main components: (1) working with local communities to assist them in finding and implementing ‘win-win’ solutions that protect forest, biodiversity and livelihoods; (2) setting up a biodiversity and habitat monitoring system that will help to delineate protected areas; and (3) creation of a local NGO to ensure sustainability of the actions.

The aim of the biodiversity component of the project is to establish a long-term field-based monitoring scheme which will provide data on the distribution and levels of biodiversity to inform conservation planning. To be able to map biodiversity distributions and delineate potential protected areas, up-to-date and high resolution land cover maps are required. The production of land cover maps by ECDD is being conducted with the support of Planet Action, Darwin Initiative, Tim Brewer of Cranfield University, GEF’s Programme of Work on Protected Areas (PoWPA), Geo Eye, and Oliver Hawlitschek of Ludwig-Maximilians-Universität Munich.

eCognition object-based classifications were conducted for the three islands of Anjouan, Grande Comore, and Mohéli using the 5m resolution Rapid Eye imagery, 30m DEM, and a derived NDVI layer. The problem of cloud cover was resolved using data from classifications conducted using 15m LandSat imagery by Oliver Hawlitschek of Ludwig-Maximilians-Universität Munich in 2008.

The next steps within the process of producing land cover maps for the Comoros will be:

- Accuracy assessments of the final land cover maps
- Improvement of the classifications if required, using Landsat, GeoEye, and Google Earth imagery. These classifications will then be used to fill the areas of cloud
- Publication of the land cover maps nationally and internationally (by August 2012).