DATASET IDENTIFICATION ------------------------ - DOI: http://dx.doi.org/10.16904/envidat.205 - URL: https://www.envidat.ch/dataset/cropland-and-grassland-map-of-switzerland-based-on-sentinel-2-data TITLE ------- Cropland and grassland map of Switzerland based on Sentinel-2 data PUBLICATION ------------- EnviDat, 2021 AUTHORS --------- - Robert Pazúr (WSL) - Nica Huber (WSL) - Dominique Weber (WSL) - Christian Ginzler (WSL) - Bronwyn Price (WSL) KEYWORDS ---------- - AGRICULTURE - CROPLAND - GRASSLAND - HABITAT - MACHINE LEARNING - PHENOLOGY - REMOTE SENSING - SENTINEL-2 - SWITZERLAND - VEGETATION INDICES ABSTRACT ---------- We developed a map of cropland and grassland allocation for Switzerland based on several indices dominantly derived from Sentinel-2 satellite imagery captured over multiple growing seasons. The classification model was trained based on parcel-based data derived from landholder reporting. The mapping was conducted on Google Earth Engine platform using random forest classifier. Areas of high vegetation, shrubland, sealed surface and non-vegetated areas were masked out from the country-wide map. The resulting map has high accuracy in lowlands as well as mountainous areas. DATES ------- - Created: 2021-01-18 ACKNOWLEDGEMENTS ------------------ The following institutions contributed to finance this dataset: - WSL - FOEN /BAFU ADDITIONAL INFORMATION ------------------------ - Version: 1.5 - Type: dataset (dataset) - Language: en - License: odc-odbl - Location: undefined DATA RESOURCES ---------------- - (current version) Cropland and grassland map of Switzerland (v.1.5) (None): (1) cropland; (2) grassland; other LC classes are masked out from the map - (current version) Cropland, grassland, and shrubland map of Switzerland (v.1.5) (None): (1) cropland; (2) grassland; (3) shrubland; other LC classes are masked out from the map - (current version) Mask of unsealed surfaces of Switzerland based on 3 years Sentinel-2 NDVI 95th percentile threshold value: 0.7 (v1.5) (None): Assuming that sealed surfaces have little or no phenological response we calculated a mask of unsealed surfaces based on the 3 years (2017-2019) Sentinel-2 NDVI 95th percentile threshold value: 0.7. - (current version) Mask of unsealed surfaces of Switzerland based on 3 years Sentinel-2 NDVI 95th percentile threshold value: 0.6 (v1.5) (None): Assuming that sealed surfaces have little or no phenological response we calculated a mask of unsealed surfaces based on the 3 years (2017-2019) Sentinel-2 NDVI 95th percentile threshold value: 0.6. - GEE (Google Earth Engine) script (JavaScript) to derived the output map (v1.5) (None): The code used in Google Earth Engine to derive the cropland, grassland and shrubland map of Switzerland. Compare to the previous version the newer version includes the shadow masking procedure which might be important in the mountainous areas. - GEE (Google Earth Engine) script (JavaScript) to derived the output map (v1.0) (None): !!! Please check out the newer version in the repository !!! The code used in Google Earth Engine to derive the cropland, grassland and shrubland map of Switzerland - Cropland and grassland map of Switzerland (v.1.0) (None): !!! Please check out the newer version in this repository !!! (1) cropland; (2) grassland; other LC classes are masked out from the map - Cropland, grassland, and shrubland map of Switzerland (v1.0) (None): !!! Please check out the newer version in this repository !!! (1) cropland; (2) grassland; (3) shrubland; other LC classes are masked out from the map - Mask of unsealed surfaces of Switzerland based on 3 years Sentinel-2 NDVI 95th percentile threshold value: 0.7 (v1.0) (None): !!! Please check out the newer version in this repository !!! Assuming that sealed surfaces have little or no phenological response we calculated a mask of unsealed surfaces based on the 3 years (2017-2019) Sentinel-2 NDVI 95th percentile threshold value: 0.7. - Mask of unsealed surfaces of Switzerland based on 3 years Sentinel-2 NDVI 95th percentile threshold value: 0.6 (v1.0) (None): !!! Please check out the newer version in the repository !!! Mask of unsealed surfaces of Switzerland based on 3 years (2017-2019) Sentinel-2 NDVI 95th percentile threshold value: 0.6 CONTACT PERSON ---------------- Robert Pazúr (Swiss Federal Research Institute WSL), mail: robert.pazur@wsl.ch