National scale maps of parent material properties, terrain and soil natural capital units at 50 metre resolution for Great Britain, 2020

We derived 9 new map layers by processing several third-party spatial data layers, including: (1) Ordnance Survey Terrain 50 Digital Terrain Model (OS Terrain 50 DTM); (2) British Geological Survey (BGS) Parent Material Model (PMM); (3) BGS Depth to Groundwater Model; (4) A unified peat map for Great Britain; (5) Soil organic carbon concentration (SOC) maps for 0-30cm and the years, 2000, 2004, 2008, 2012, 2016 and 2020 from the EcoDataCube; and (6) Locations of quarries and dumps circa year 2020, according to Open Street Map (OSM).
All maps were harmonised to 50m resolution raster grids under the British National Grid coordinate reference system to the same position and extent. For data already in raster format, layers were reprojected to British National Grid and resampled to 50m. Vector-format maps, including the BGS PMM variables we used and the peat map, were rasterised.
Next, through a mix of aggregating values into a smaller number of classes and combining different map layers together, the following 4 new map layers were produced: (1) Textural composition, (2) Physico-chemical parent material characteristics, (3) Subsurface drainage controls, and (4) Geomorphon landform classes
A circular moving neighbourhood analysis window with a 1km diameter was applied to the first 3 maps to smooth out artefacts and sensitive linework in the underlying BGS PMM layers. Next, a decision tree was applied to assign 13 soil natural capital units, defined by combinations of categories across these 4 component maps. To support the accuracy evaluation of our maps, we derived an additional 4 maps of landscapes and terrain indices from the OS Terrain 50 DTM, giving us a total of 9 maps of soils, parent material and terrain for Great Britain. Each of these maps was clipped to a vector layer that represented a "true" land mask. This mask layer was derived from the OS boundary high water mark and spatial inventory of water bodies to exclude coastal and inland areas of open water. Full details of all processing steps are included in the Supporting Documentation file included with the dataset.
Initial checks on the basic structure of the dataset were performed. These include steps to verify that all 9 maps shared the same and correct coordinate reference system, grid resolution and spatial extent; consisted of values within expected ranges; had appropriately defined no-data values; and could be loaded into R, QGIS and ArcMap to test interoperability.
We compared our subsurface drainage map with the Hydrology of Soil Types (HOST) map (63 % agreement); our textural composition map against the NATMAP topsoil texture map for England & Wales and a topsoil texture map for Scotland, derived from the National Soil Map of Scotland (53 % agreement); and our physicochemical properties map against equivalent classes from the NATMAP and National Soil Map of Scotland datasets for England & Wales and Scotland, respectively (72 % agreement). Full details of the evaluation results can be found in the Supporting Documentation file included with the dataset.
In an effort to remove small artefacts and blur sensitive linework from the BGS parent material maps, we applied a moving neighbourhood smoothing algorithm with a 1km diameter circle. This has blurred out some finer spatial detail, rendering our maps unsuitable for sub-field scales. We would therefore recommend that users apply our data at whole-field scale or across larger domains, such as river catchments or counties up to full national scale.