{"ob_id":42320,"computationComponent":[{"ob_id":42319,"uuid":"66b1679c46ef49fdadc0c0c1a590ce63","title":"Computation for Greenland 1980 and 2010s landcover grids from Landsat 5 and Landsat 8","abstract":"the tif grids were produced using Google Earth Engine. All summer Landsat imagery was filtered by metadata, followed by topographical correction, resulting in a best-pixel mosaic for Greenland's periphery. Band ratios (NDSI, NDVI, NDWI) were computed and stacked with visible, NIR, and SWIR bands. A principal component analysis was conducted, retaining the first six principal components as bands, which were subsequently classified using a K-means clusterer and refined with a supervised random-forest classifier and a slope threshold was applied to discriminate shadows from dark water bodies more effective.","keywords":"","inputDescription":null,"outputDescription":null,"softwareReference":null,"identifier_set":[]}],"acquisitionComponent":[{"ob_id":12363,"independentInstrument":[],"instrumentplatformpair_set":[{"ob_id":4806,"platform":"https://catalogue.ceda.ac.uk/api/v2/platforms/12358/?format=json","instrument":"https://catalogue.ceda.ac.uk/api/v2/instruments/12365/?format=json","relatedTo":{"ob_id":12363,"uuid":"6b99193d62cd44fa9239d3b02991887d","short_code":"acq"}}]},{"ob_id":27068,"independentInstrument":[],"instrumentplatformpair_set":[{"ob_id":11996,"platform":"https://catalogue.ceda.ac.uk/api/v2/platforms/8246/?format=json","instrument":"https://catalogue.ceda.ac.uk/api/v2/instruments/7872/?format=json","relatedTo":{"ob_id":27068,"uuid":"812b78639c2740f98090d73e5a8f938e","short_code":"acq"}}]}],"identifier_set":[],"responsiblepartyinfo_set":["https://catalogue.ceda.ac.uk/api/v2/rpis/203926/?format=json"]}