Understanding agricultural drivers of deforestation through remote sensing: opportunities and limitations in sub-Saharan Africa
This working paper (PDF) by the International Institute for Environment and Development (IIED) highlights opportunities and limitations for understanding agricultural drivers of deforestation through remote sensing. Effective monitoring of deforestation and cropland expansion in Africa requires reliable estimates of land cover area. However, continental scale land cover datasets generated solely or partially through remote sensing technologies show large differences in the extent and spatial distribution of forest and cropland. In this working paper, 14 of the most commonly used land cover data products are identified, with summarised spatial data, temporal and thematic properties, and estimates of forest and cropland area are compared. The limitations to land cover data that result from their divergent predictions are described and the implications for using this data to assess the agricultural drivers of deforestation in Africa are discussed. TA key finding found was that disagreement between data products is apparent at continental, national and regional scales, being particularly elevated at the smallest spatial scales. Forest area estimates agree most strongly in areas of dense forest and where forest is clearly not present (eg deserts). Agreement between estimates of forest area is poor in areas of woodland and savannah. Crop area estimates show greater disagreement than for forest area, and may be particularly uncertain where agriculture is small scale and spatially scattered as is dominant in much of Africa. Data products agree better on locations where agriculture is limited rather than widespread.