Methodology
CarHab is a national cartographic modeling program for natural and semi-natural habitats in France. It is a machine-learning modelling using existing vegetation data, image analysis techniques and field prospection. The biotope and vegetation physiognomy maps for each department are produced by supervised modelling using a Random Forest algorithm. The National Botanical Conservatories provide the training data for the model from existing data on their territory. For the vegetation physiognomy, the modelling from satellite images (Sentinel-2) is completed by information layers from the BD Forêt® and the BD TOPO® of the IGN. For biotopes, the modelling is based on geographical information layers of environmental variables (topography, climatic data, geology, pedology, sunshine, etc.). A field survey phase is used to refine the model.
Data description
Its objective is to carry out by 2025 a national mapping of natural and semi-natural terrestrial habitats of mainland France and the French overseas territories, at a scale of 1: 25,000th. The results should provide information on the location of habitats. Three objects of study are considered : Biotopes, Vegetation physiognomies and CarHab Habitats. CarHab Habitats are the combination of information from the modeling of biotopes and that of the physiognomy of vegetation; each polygon resulting from this crossing contains information relating to the biotope (combination of the 8 edaphic and climatic parameters) and on the physiognomy of corresponding vegetation (stage in the dynamics of vegetation development).
Indicator
Species richness
Method / tool
Impact on biodiversity (in wetlands)
Geographic coverage (details)
France