Distribution modelling of vegetation types based on area frame survey data

摘要: AIM: Many countries lack informative, high‐resolution, wall‐to‐wall vegetation or land cover maps. Such maps are useful for land use and nature management, input to regional climate hydrological models. Land cover based on remote sensing data typically the required ecological information, whereas traditional field‐based mapping is too expensive be carried out over large areas. In this study, we therefore explore extent which distribution modelling (DM) methods predicting current of types (VT) a national scale. LOCATION: Mainland Norway, covering ca. 324,000 km². METHODS: We used presence/absence 31 different VTs, mapped in an area frame survey with 1081 rectangular plots 0.9 km². Distribution models each VT were obtained by logistic generalised linear modelling, using stepwise forward selection F‐ratio test. A total 116 explanatory variables, recorded 100 m × 100 m grid cells, used. The evaluated applying AUC criterion independent evaluation dataset. RESULTS: Twenty‐one had values higher than 0.8. highest value (0.989) was Poor/rich broadleaf deciduous forest, lowest (0.671) Lichen heather spruce forest. Overall, found that rare VTs predicted better common ones, coastal inland ones. CONCLUSIONS: Our study establishes DM as viable tool spatial prediction aggregated species‐based entities such scale at fine (100 m) resolution, provided relevant predictor variables available. discuss potential uses utilizing large‐scale international surveys. also argue predictions from may improve parameterisation earth system