Search

Global land cover maps are widely used for assessment and in research of various kinds, and in recent years have also come to be used for socio-economic forecasting. However, existing maps are not very accurate, and differences between maps also contribute to their unreliability. Improving the accuracy of global land cover maps would benefit a number of research fields. In this paper, we propose a methodology for using ground truth data to integrate existing global land cover maps. We checked the accuracy of a map created using this methodology and found that the accuracy of the new map is 74.6%, which is 3% higher than for existing maps. We then created a 0.5-min latitude by 0.5-min longitude probability map. This map indicates the probability of agreement between the category class of the new map and truth data. Using the map, we found that the probabilities of cropland and grassland are relatively low compared with other land cover types. This appears to be because the definitions of cropland differ between maps, so the accuracy may be improved by including pasture and idle plot categories. [ABSTRACT FROM AUTHOR]