The latest edition is also available at a stable path: dk/latest/dk.parquet. All editions together through the S3 glob s3://ftw/harmonized-field-data/dk/year=*/*.parquet (see the AGENTS.md for the DuckDB setup; plain https cannot expand *).
id differs by edition: from 2014 it is the application and the field number within it (Journalnr:Marknr), which identifies a field; before that the source names only the applicant, and that pair repeats — 5,124 keys cover 11,534 of the 678,347 fields of 2008 — so there id is a row number. A new application number is issued every year either way, so no id follows a field across editions. The older editions also carry fewer attributes: 2008 and 2009 name no crop at all, and the field block (Markblok) starts in 2016.
Area of the field, in square meters (m²). Must be > 0 and <= 1,000,000,000. (spec)
hcat:name
string
The machine-readable HCAT name of the crop (Hierarchical Crop and Agriculture Taxonomy, EuroCrops). ()
Properties that are the same for every field are stored once, in the GeoParquet file's collection metadata rather than as columns (latest edition shown; a client reading only the table will not see them):
Query the published files in place with DuckDB; nothing needs downloading first.
Provenance
This catalog is a mirror: the data is produced and licensed by Danish Agricultural Agency and republished here as cloud-native GeoParquet and PMTiles by Fields of the World. Each edition was downloaded from the source and converted with fiboa-cli 0.21.0, vecorel-cli 0.2.18: