ABSTRACT
This chapter provides a comparison of four land cover data sets to determine which product is the most suitable for applications such as food security and monitoring of agricultural expansion. The land cover products compared are: the Global Land
Cover Map (GLC-2000), the MODIS land cover product (MOD12V1), the SAGE cropland database, and the AFRICOVER data set from the Food and Agricultural Organization (FAO). The data sets were first converted to the same resolution and the legend classes were then reconciled. A comparison was then made of the overall agricultural areas for the different data sets together with official national and subnational statistics. This was followed by a spatial comparison of the land cover maps; areas of high disagreement were identified between each land cover product and AFRICOVER. The results showed that uncertainties in the cropland distribution in African countries are very high. For example, MODIS had the tendency to underestimate cropland cover, whereas the GLC-2000 tended to overestimate cropland cover in those countries that are located at the northern transition zone of subtropical shrubland and semidesert areas. In this area, MODIS and SAGE showed a relatively similar cropland distribution. It was also demonstrated that even though overall cropland areas in administrative units (e.g., FAO national statistics) are not so far apart, the spatial distribution of these can vary and a high level of uncertainty exists when a spatial comparison is undertaken.
