Rangelands vegetation mapping at species composition level using the SPiCla method: SDM based pixel classification and fuzzy accuracy - A new approach of map making

Vegetation maps have been made since centuries. The vegetation cover was represented as homogeneous mapping units (polygons), representing different vegetation types, where each type consists a combination of different plant species (floristic composition). More recent, with the use of satellite imagery, the polygons have been replaced by pixels with similar content as the polygon maps. In both approaches, field-observations were linked to the mapping units (polygons or pixels) often resulting in a complex of different vegetation types per mapping unit. In our new approach field data (sample points) on presence and abundance of individual grass species are spatially extrapolated based on a set of environmental layers, using the species distribution modelling approach (SDM). When combined, each pixel will contain its own set of information about the vegetation structure and its floristic composition. This new methodology (SPiCla) results in a very accurate and detailed vegetation map at pixel level, allowing extraction of very detailed, accurate and easy to update spatial information on e.g., forage production and quality (palatability) for rangelands management. As no exact boundaries exist, but only gradients, we introduced fuzzy accuracy. The resolution mainly depends on the resolution of (or one of) the environmental layers used, scale of interest and workability. The methodology is generic and applicable to any other region in the world.