Land Cover Africa 2016 - CCI ESA

Prototype high resolution LC map at 20m over Africa based on 1 year of Sentinel-2A observations from December 2015 to December 2016. The legend of the S2 prototype LC 20 m map of Africa 2016 was built after reviewing various existing typologies (e.g. LCCS, LCML…), global (e.g. GLC-share, GlobeLand30) and national experiences (Africover, SERVIR-RMCD). The legend includes 10 generic classes that appropriately describe the land surface at 20m: "trees cover areas", "shrubs cover areas", "grassland", "cropland", "vegetation aquatic or regularly flooded", "lichen and mosses / sparse vegetation", "bare areas", "built up areas", "snow and/or ice" and "open water". Two of the LC classes were largely identified thanks to external dataset: the "open water" class was based on the Global Surface Water product from JRC/EC and the "urban areas" relied both on the Global Human Settlement Layer from JRC/EC and on the Global Urban Footprint from DLR. Two classification algorithms, the Random Forest (RF) and Machine Learning (ML), were chosen to transform the cloud-free reflectance composites generated by the pre-processing module into a land cover map. The two maps resulting from both approaches are then combined either to select the best representation of a land cover class which will be part of the final S2 prototype LC 20 m map of Africa 2016 or, in case of unreliable LC class delineation, the reference layer is used to consolidate the land cover classification.