Show number of records with valid coordinates about gbif-dataset-metrics

[peterdesmet]

For definitions of coordinate issues, see http://gbif.github.io/gbif-api/apidocs/org/gbif/api/vocabulary/OccurrenceIssue.html. Based on the download I have, this is how hasGeospatialIssues is interpreted:

Issue	hasGeospatialIssues
Zero coordinate	?
Coordinate out of range	true
Coordinate invalid	false
Coordinate rounded	false
Geodetic datum invalid	false
Geodetic datum assumed WGS84	false
Coordinate reprojected	false
Coordinate reprojection failed	false
Coordinate reprojection suspicious	?
Country coordinate mismatch	true
Presumed swapped coordinate	?
Presumed negated longitude	false
Presumed negated latitude	false

I find it odd that coordinate invalid is interpreted as false (see #22). Maybe even bad enough that we'd have to parse the issues ourselves. The advantage of doing so is that we can use the API or the download.

[peterdesmet]

After really digging into this issue, I'm wondering if the end result won't be exactly the same as georeferenced records... 😵

http://api.gbif.org/v1/occurrence/count?datasetKey=7f9eb622-c036-44c6-8be9-5793eaa1fa1e&isGeoreferenced=true

IS_GEOREFERENCED has no definition though.

[peterdesmet]

Also, GBIF does correct negative/swapped ones and does transform geodetic datums it understands (example). So what you're getting in decimalLatitude/Longitude is (mostly?) WGS84. Oddly, hasCoordinateIssues doesn't raise a flag when the reprojection failed.

[peterdesmet]

@bartaelterman @niconoe, please read this issue on GitHub... I've changed some previous comments. After a long search through all the fields, IS_GEOREFERENCED is maybe the easiest solution. The only disadvantage is that we don't know exactly what it is based on: to do that, we'd have to parse decimalLatitude, decimalLongitude and search for specific issues. What do you think?

[peterdesmet]

Have thought about this a bit further, and I think our strength is in showing more than a true/false distinction. On a bar chart we could show:

• Good coordinates
• Coordinates with serious issues
• Coordinates not provided

That way the user has an idea of what he is getting, and the publisher has an idea how well he's doing. We can do very similar overviews for other metrics (taxon, date, elevation), maybe even including precision. I'll create separate use cases.

[peterdesmet]

Regarding last comment: this is all described in issue #23. Closing this one.