Fig 1.
Modular framework to find, harmonise, evaluate, and publish reference data.
Table 1.
Land cover classes defined in the WorldCereal reference data repository.
Table 2.
Data sets that were collected and harmonised (data sharing policy “private” means only shared for internal use within the WorldCereal project).
See S2 Table for more background on these data sets and S3 Table on the recommended citation and data licence.
Table 3.
Distribution of observations (in %) by data policy (public vs private), geometry (point vs polygon) and year of observation for datatype FO and CV (top) and datatype AC and FD (bottom).
Fig 2.
Spatial coverage of the public FO and CV data, holding 593,820 labels, in the WorldCereal reference database (in orange, lines indicate country borders and green colour indicates main maize and wheat areas according SPAMM 2010 version 2).
Fig 3.
Distribution of LC (a) and CT (b) labels of the public FO and CV data (numbers are percentages of total numbers of labels).
Table 4.
Confidence scores of all data sets (private and public).
Fig 4.
Breakdown of confidence scores over the different criteria for FO datatype crop type mapping.
Fig 5.
Screenshots of the WorldCereal data portal showing the landing page (a), a global view of publicly available data (b) and the details and associated metadata of each contributed data, including the spatial coverage of the data shown on the map (c).