Extras in Higher Geography - Ocean, Island Group, Island, Feature, Sea, Quad, and Drainage

[Jegelewicz]

While working on the cleanup in #4809 I came across this locality:

https://arctos.database.museum/place.cfm?action=detail&collecting_event_id=11135812

and I figured - hey, while I'm here I can just georeference this thing! So I used the nifty Geolocate button, but I get nothing because part of what Geolocate refers to as Locality "Gila Cliff Dwellings National Monument" we have tucked up in higher geography. And what gets placed in locality "No specific locality recorded." is NOT a locality (see also #4807 (comment)) so Geolocate will never find any coordinates for it. While I was able to just replace "No specific locality recorded." with Gila Cliff Dwellings National Monument in locality for Geolocate to get some coordinates, it made me think twice about how we are handling everything in higher geography except Continent, Country, State, and County.

The rest of the world expects to find all the other stuff (Ocean, Island Group, Island, Feature, Sea, Quad, and Drainage) somewhere else.

Island Group
Island
Ocean/Sea dwc:waterBody
Feature - IMO everything in here is really part of locality
Quad - is just another kind of County for Alaska, but for everything else, maybe it is just a feature?
Drainage - OOF, this is a tough one, but I think it is waaaayyyy bigger than Arctos, however maybe just another kind of feature?

The things we put in higher geography that don't jive with DarwinCore end up missing at GBIF

Anyone downloading from GBIF and using Continent, Country, State, County, Locality will never see the "Gila Cliff Dwelling National Monument that we have buried in our higher geography. It's there, but who is going to look there?

AND Arctos now thinks the coordinates assigned are not appropriate because they don't match "Catron County", but Catron County IS NOT the most specific locality!

I can see a bunch of paths forward and I would appreciate other's input. I hope that we won't just continue as we are doing now, because I don't think it works well for data sharing or analysis, even using a download directly from Arctos).

[sharpphyl]

Ocean/Sea dwc:waterBody

Yes, please! #2876

[dustymc]

nifty Geolocate button

Whatever else we do, let's not let some UI or service guide our data structure.

IMO everything in here is really part of locality

I've been asking that question for many years (#1366 is one iteration). In my current view that's definitely "geography" and (like all geography) could and should be represented spatially. (I'd elevate "should" to "must be" if I could - #4755)

Anyone downloading from GBIF and using Continent,

If GBIF has something acceptable (they have something, IDK if it's acceptable to anyone) then let's just use it. Arctos is built for that sort of thing, I have spatial tools, there's no reason to continue chasing some elusive target if we could just buy into some existing ecosystem.

AND Arctos now thinks the coordinates assigned are not appropriate because they don't match "Catron County", but Catron County IS NOT the most specific locality!

There's an authority and a determination involved. If you're seeing that message then the latter isn't enclosed by the former and the data cannot be anything other than self-conflicting. (#4289 would actually address that.)

Yes, please! #2876

That does contain a proposal that I'm not sure is wrong, but #2374 is probably a generally more productive approach.

[Jegelewicz]

From Tim Robertson at GBIF

I can only reply quickly…

All we really have in terms of a backbone is GADM, and then we have EEZs and a couple of other reference layers. We basically take coordinates and look them up against the layers - nothing more. We will add more layers in time.

You can call a lookup API to reverse geocode a lat/lng into what that appears in e.g. GADM.

https://api.gbif.org/v1/geocode/reverse?lat=44&lng=-70

In haste

Tim

[dustymc]

GADM, and then we have EEZs and a couple of other reference layers

Thanks, I can get those easy enough.

ou can call a lookup API to reverse geocode a lat/lng into what that appears in e.g. GADM.

Yep, I make use of that, the problem is that we don't deal in lat/lng. What I need is "find the smallest 'authority shape' which encloses this 'determination shape'" (#3114 would do that, #4289 should be easy/natural with that kind of data).

[dustymc]

The geography is a mess because #4836, the services..

... have come up with a reasonable georeference. I think this is just symptoms of other documented problems, tentatively closing.