Countrywide classification of permanent grassland habitats at high spatial resolution

European grasslands face strong declines in extent and quality. Many grassland types are priority habitats for national conservation strategies. Countrywide, high spatial resolution maps of their distribution often lacking. Here, we modelled the 20 permanent at level phytosociological alliances across Switzerland 10x10 m resolution. First, applied ensemble models to provide individual habitat types, using training data from various sources. Copernicus Sentinel satellite imagery variables describing climate, soil topography were used as predictors. The performance these was assessed based on true skill statistics with a split-sampling data. Second, combined into countrywide most second likely type, respectively, an expert-based weighting approach. map type via independent testing dataset comparison predicted habitat-type proportions extrapolations field surveys. Most had useful excellent predictive (TSS ≥ 0.6). For grid cells maps, either ecologically closely related or representing two along nutrient gradient. same omission errors. We found good agreement between estimated area raised bogs appears be underestimated, while dry showed highest agreement. This work highlights potential earth observation fine temporal broad scales, thereby providing foundation diverse applications. A particular challenge remains capturing transition nutrient-poor nutrient-rich grasslands, which is highly important biodiversity conservation.