Darwin Core mapping from processed log data rather than CartoDB

[stijnvanhoey]

Providing a Python implementation of the ETL of gull bird tracking. Please revise this translation. If ok, I'll provide a function-based approach that can be executed for the entire data set.

[peterdesmet]

The mapping to Darwin Core has been updated compared to the CartoDB SQL implementation

Updated fields

• ownerInstitutionCode (for gulls): INBO/VLIZ/UGent/UA (instead of INBO only)
• datasetID: use https
• dynamicProperties: catch_location + tracking_started_at now also included in JSON
• samplingProtocol: use DOI URL
• decimalLatitude: now always 7 decimals
• decimalLongitude: now always 7 decimals

Removed fields

• modified: was based on CartoDB, we don't use this in other datasets
• georeferencedDate: was the same as eventDate, we don't use this in other datasets
• georeferenceProtocol: not really necessary to reference article on point-radius method as the source data are coordinates.
• higher taxon fields: As there are only 3 species, these fields don't provide any filtering functionality. We keep "kingdom" for proper linking with GBIF backbone taxonomy:
  • phylum
  • class
  • order
  • family
• lower taxon fields: scientificName is sufficient. No need to indicate authorship + nomenclaturalCode does not correct (we only provide genus + species in scientificName):
  • genus
  • specificEpithet
  • scientificNameAuthorship
  • nomenclaturalCode

This reduces the number of columns from 44 to 35.