GBIF validation of export from read_wi() and write_dwc()

[Rafnuss]

Hey, just wanted to check with you @peterdesmet that these validation issues from the GBIF files are not something we can do at the read_wi() or write_dwc() level.

https://www.gbif.org/tools/data-validator/5547190a-10d5-4667-af3d-dc00be586117
https://www.gbif.org/tools/data-validator/47d69d10-0142-45c5-b997-ffed38be6ef5 (doesn't seem to work for audubon)

• Coordinate rounded: Would it make sense to add round_coordinates( , digit=6) by default? (https://data-blog.gbif.org/post/issues-and-flags/)
• Country derived from coordinates: ignore
• Taxon match higherrank and taxon match fuzzy: what would be the ideal way to check/modify taxonomic discrepency between WI and GBIF? Maybe not something for camtraptor, but is there a way to modify the scientific name in a datapackage? Do you recommend the use of the specie API (e.g., https://docs.ropensci.org/rgbif/reference/index.html#taxonomic-names?)

[peterdesmet]

Coordinate rounded means that GBIF has rounded the coordinates for us, but best practice is to provide the coordinates as they are in the source database, so I would not add a default round.

Taxon match fuzzy: you can verify where it goes wrong by providing a CSV of all your scientific names to https://www.gbif.org/tools/species-lookup

[peterdesmet]

Where is the Basis of record invalid coming from in the second link?

[Rafnuss]

Where is the Basis of record invalid coming from in the second link?

The second link is the validation of the audubon file. Is this not due to the fact that the validation tool is unaware of the audubon class? (i.e., there is not validation possible for this class).

[peterdesmet]

Yes, that is definitely possible. Did you upload both files as a single DwC-Archive or separately?

[Rafnuss]

Yes, it was uploaded seperatly. Here is the one uploaded as a zip file: https://www.gbif.org/tools/data-validator/bdbd3cb4-224b-4cfc-b479-49588b64334a

here is the emf file:
https://gist.github.com/Rafnuss/5dc72f6ac42430135375d52085f9f396

[peterdesmet]

Ok, never mind, is indeed missing an EML and meta.xml

[peterdesmet]

Note: here's a published (sequence level) dataset that made use of write_eml() and write_dwc(): https://www.gbif.org/dataset/8a5cbaec-2839-4471-9e1d-98df301095dd

[Rafnuss]

Do you mean this code? https://github.com/inbo/mica-occurrences/blob/master/datasets/mica-agouti-occurrences/src/dwc_mapping.Rmd
It would be nice to have that as an article (vignette) somewhere no? Useful for me at least.

[peterdesmet]

Yeah, good idea to add this as a vignette. It would have to be for the example small dataset.

[peterdesmet]

• The status of read_wi() is discussed in What's the future of camtraptor's read_wi() function? camtrapdp#62.
• A data publication vignette is also being discussed Create data publication vignette camtrapdp#5
• Ideally, Wildlife Insights can export as Camtrap DP in the near future, at which point a read_wi() is not needed
• GBIF validation is being tested with GBIF for Camtrap DPs directly.

Closing this issue, @Rafnuss let me know if there are remaining items to be tackled.