Data for: Implications of land cover classification on soil carbon monitoring – comparing LUCAS and Swedish soil inventories

The dataset contains two data files (LUCAS_BlockParcel_compiled_F.tsv and InventoriesSE_F.tsv) underlying the manuscript comparing the Land Use/Land Cover Area Frame Survey (LUCAS) and the Swedish national soil inventory on agricultural soils. Both data files are compiled files: LUCAS_BlockParcel_compiled_F.tsv is based on LUCAS and the Swedish Land Parcel Identification System (LPIS), while InventoriesSE_F.tsv is based on the Swedish crop and soil inventory on arable land (SLU environmental monitoring and assessment; agricultural landscapes programme) and the Swedish forest inventory (SLU environmental monitoring and assessment; forests programme), which covers semi-natural grassland. LUCAS_BlockParcel_compiled_F.tsv: 93 columns, 188 rows Raw LUCAS land cover and 0-20 cm topsoil dataset for the years 2009, 2015 and 2018 were obtained from the European Soil Data Centre. We filtered the dataset to include only sampling points classified as cropland or grassland in the Land Cover column, resulting in 187 resampled locations in Sweden. Agricultural blocks and parcel database are obtained from Block and parcel database were downloaded from the Swedish Board of Agriculture (https://jordbruksverket.se/e-tjanster-databaser-och-appar/e-tjanster-och-databaser-stod/kartor-och-gis, accessed in July 2025). We created LUCAS soil sampling points layers using GPS coordinates and projected to SWEREF99 TM to do spatial join with block and parcel database. The geospatial block database (jordbruksblock) delineates farmers’ agricultural blocks based on land types (ägoslag), while the parcel database (jordbruksskifte or skifte) further divides the blocks into parcels that contain attributes of crop type, growing area and type of support, etc. (Widén-Nilsson et al., 2025). Because the routine for updating the block database and the year it represents changed in 2009, the block from the following year should be used to match the crop data for 2009 onward, and any unmatched records should subsequently be matched with the block dataset from the same year (Widén-Nilsson et al., 2025). Thus, we spatially joined LUCAS sampling points from 2009, 2015, and 2018 with the block database to obtain land type attributes. Before 2012, the parcel database had very limited spatial coverage of agricultural land. However, crop information for parcels applying for support was available as text files. These files could be matched with the block database using the block ID to identify which crops were grown within each block. Crop data were added as follows: for 2009, we extracted all parcels associated with each LUCAS sampling location using Block IDs and calculated the crop composition per block based on cultivated area; for 2015 and 2018, crop type attributes were added directly via spatial join with the parcel database. The geodesic distances for the same point between sampling rounds were calculated. Spatial data processing was conducted in ArcGIS Pro with arcpy. Changes in soil organic carbon (SOC) content between sampling year 2009, 2015 and 2018 were calculated. InventoriesSE_F.tsv: 25 columns, 1734 rows The Swedish Soil and Crop Inventory on arable land began its second sampling cycle (2001-2007) with the recording of precise geographic coordinates, and a decade later (2011-2017), the same locations were resampled for 0-20 cm soil depth, resulting in 1,821 revisited sampling points. The Swedish Semi-Natural Grassland Inventory sampled soil at 0-10 cm depth, with 63 revisited sampling points available during the two similar inventory periods (2003-2009 and 2013-2019). To enable a direct comparison of SOC change estimates focusing on mineral soils, we further excluded sampling points located on organic soils, with high carbonate content (SOC > 16%, CaCO3 > 5%) and those missing particle size data from the two national inventories. SOC was reported as a percentage of soil mass, and changes in SOC were expressed as the absolute difference in percentage throughout this study. Data has been generated as part of SLU's environmental monitoring and assessment from the Soil and Crop Inventory (Agricultural Lanscapes) and the Swedish Forest Soil Inventory (Forest Landscapes) programs.