Mapping Land Cover and Land Use Changes in the Congo Basin Forests with Optical Satellite Remote Sensing: a Pilot Project Exploring Methodologies that Improve Spatial Resolution and Map Accuracy

The University of Maryland has been a USAID Central Africa Rregional Program for the Environment (CARPE) cross-cutting partner for many years, providing remote sensing derived information on forest cover and forest cover changes in support of CARPE's objectives of diminishing forest degradation, loss and biodiversity loss as a result of poor or inexistent land use planning strategies. Together with South Dakota State University, Congo Basin-wide maps have been provided that map forest cover loss at a maximum of 60m resolution, using Landsat imagery and higher resolution imagery for algorithm training and validation. However, to better meet the needs within the CARPE Landscapes, which call for higher resolution, more accurate land cover change maps, UMD has been exploring the use of the SIAM automatic spectral -rule classifier together with pan-sharpened Landsat data (15m resolution) and Very High Resolution imagery from various sources. The pilot project is being developed in collaboration with the African Wildlife Foundation in the Maringa Lopori Wamba CARPE Landscape. If successful in the future this methodology will make the creation of high resolution change maps faster and easier, making it accessible to other entities in the Congo Basin that need accurate land cover and land use change maps in order, for example, to create sustainable land use plans, conserve biodiversity and resources and prepare Reducing Emissions from forest Degradation and Deforestation (REDD) Measurement, Reporting and Verification (MRV) projects. The paper describes the need for higher resolution land cover change maps that focus on forest change dynamics such as the cycling between primary forests, secondary forest, agriculture and other expanding and intensifying land uses in the Maringa Lopori Wamba CARPE Landscape in the Equateur Province of the Democratic Republic of Congo. The Methodology uses the SIAM remote sensing imagery automatic spectral rule classifier, together with pan-sharpened Landsat imagery with 15m resolution and Very High Resolution imagery from different sensors, obtained from the Department of Defense database that was recently opened to NASA and its Earth Observation partners. Particular emphasis is placed on the detection of agricultural fields and their expansion in primary forests or intensification in secondary forests and fallow fields, as this is the primary driver of deforestation in this area. Fields in this area area also of very small size and irregular shapes, often partly obscured by neighboring forest canopy, hence the technical challenge of correctly detecting them and tracking them through time. Finally, the potential for use of this methodology in other regions where information on land cover changes is needed for land use sustainability planning, is also addressed.