Morton, R.D.; Marston, C.G.; O'Neil, A.W.; Rowland, C.S.

Land Cover Map 2017 (20m classified pixels, N. Ireland)

https://doi.org/10.5285/f6b551ea-e079-4150-b13c-b5f89f5dbaa6 Cite this dataset

This is the 20m classified pixels dataset for the UKCEH Land Cover Map of 2017 (LCM2017) representing Northern Ireland. It describes Northern Ireland's land cover in 2017 using UKCEH Land Cover Classes, which are based on UK Biodiversity Action Plan broad habitats. This dataset is the Random Forest classification result from classifying a 20m pixel raster containing multi-season spectral information combined with context layers, which help to resolve spectral confusion. It is provided as a 2-band, 8-bit integer raster. The band-1 is the UKCEH Land Cover Class identifier, band-2 is an indicator of classification confidence. For a fuller description please refer to the product documentation.

This work was supported by the Natural Environment Research Council award number NE/R016429/1 as part of the UK-SCAPE programme delivering National Capability.

Publication date: 2020-06-22

Format

TIFF

Spatial information

Study area
Spatial representation type: Raster
Spatial reference system: TM75 / Irish Grid

Temporal information

Temporal extent: 2017-01-01 to 2017-12-31

Provenance & quality

The 20m Classified Pixels datasets are the most spatially detailed in LCM2017, LCM2018 and LCM2019 product range. All other datasets were derived from these. They give land cover information in 21 classes based on UK Biodiversity Action Plan broad habitats. They result from automatic Random Forest classifications of partially overlapping, rectangular classification scenes and are given as 8-bit, 2-band integer 20m pixel rasters. Band-1 is the most likely land cover class, band-2 is the classification probability associated with this class, rescaled over the interval 0 to 100. Classification scenes comprised per-season Sentinel-2 median top of atmosphere reflectance values resampled to 20m pixel resolution, derived using the Google Earth Engine, combined with regional context layers (for example: terrain, coastal proximity). Overlapping classification results were visually ranked to determine precedence. The best classifications were compiled to form seamless national products for Great Britain and Northern Ireland. Classification used bespoke software integrated with a PostGis spatial database. The 20m rasters were exported from PostGIS using open source gdal tools to give the final product in GeoTiff format. GeoTiff is open standard format for raster data and GeoTiff files can be read by most GIS software.