INSPIRE - Land use in Wallonia - WALOUS 2018 in Wallonia (BE)

This data layer includes the mapping of the land cover of the entire Walloon territory for the year 2018. The data is presented here in its INSPIRE compliant version. This dual dataset (raster/vector) shows the mapping of land cover for the entire Walloon territory for 2018 (WAL_OCS__2018). It is complementary to the land use data (WAL_UTS__2018) produced jointly under the research grant WALlonie Occupation et Utilisation du Sol (WALOUS). Land cover is defined as the "physical and biological cover of the land surface, including man-made areas, agricultural areas, forests, (semi-)natural areas, wetlands and water bodies" (European INSPIRE Directive 2007/2/EC, 2007). Information on land cover is provided, for both types of spatial representation, in the Belgian Lambert projection system 2008 (EPSG:3812 – For use in LB72, see information on grid NTV2 -http://geoportail.wallonie.be/home/resources/tools/Lambert-belge-2008-L…). The spatial resolution of the raster data is 1 m set on the INSPIRE grid of the Walloon Region. The vector data is directly derived from it. The only generalisation operation applied to the vector layer is the broken corner technique, which consists in joining the stopping media of the border pixels to avoid a stair effect. The technical specifications of WAL_OCS__2018 are the result of a needs analysis procedure and co-constructions with a panel of Walloon stakeholders. Both formats are characterized by a minimum mapping unit of 15m2. This avoids, for example, the representation of all the small gaps in the forest. The WAL_OCS__2018 data details land cover in 11 main classes. However, some pixels belong to two classes in case of levelling (bridge over a river for example). These non-zero two-digit codes are detailed in the Excel spreadsheet describing the symbology available for download. WAL_OCS__2018 is the result of the fusion of GEOBIA object-oriented and pixel-oriented classifications using 2018 ortho from the Walloon region, MNH (from the 2013-2014 LiDAR survey), Sentinel 2 satellite data and a vector skeleton (built on the basis of SPW - IGN data). It has been consolidated by an automated rules approach and by manual consolidation by experts. And finally, it was validated using a random sample of 1550 photopoints interpreted independently of the classification process and analysed according to the rules of the art. Its overall accuracy is 91.5%. The implementation of the WAL_OCS__2018 data is part of the WALOUS project (2017-2020) subsidised by the Walloon region and carried out by a consortium of two universities (UCLouvain, ULB) and a public research centre of the type UAP (ISSeP).