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 identify 14 of the most commonly used land cover data products, summarise their spatial, temporal and thematic properties, and compare their estimates of forest and cropland area. It describe the limitations to land cover data that result from their divergent predictions and discuss the implications for using this data to assess the
agricultural drivers of deforestation in Africa.