Wikidata could be another source for boundaries via the geoshape property, but i don’t know if that would be better than OSM from a data quality perspective.
You have a strange definition of “easy” Even just generating the list of boundaries to extract and mapping those names to the appropriate OpenStreetMap objects would probably take days of work.
I looked into Wikidata’s shape files. It looks like a lot of them originate from https://github.com/nvkelso/natural-earth-vector, which is exactly the sort of freely-licensed, collaborative, boundary shape file repository I was dreaming of. Unfortunately, that project is also apparently derelict (the last update was in 2022). And the quality of the shape files is comparable to that of GADM. Most of them seem to be rough approximate outlines rather than closely following the legal boundaries. But you might be onto something with Wikidata. If we could convince the Wikidata community to fix some of the worst shape files, it might be a good option for the future.
Also, foxes are NOT abundant in Wallonia. I really don’t get why people shoot foxes and then use poison bait to kill mice. Only 3 weeks ago I took a very friendly fellow Nature photographer to “my” foxes after work. He is in his 70s and never had the opportunity to take fox pics before so he was really grateful.
I had the opportunity to watch them many times after work and was glad to learn their behaviour.
And no, I have never seen them in a town. I heard they do live in towns in Britain, but so do parrots
possible, and that’s why i didn’t say that that part of the effort would be eady or hard, but i think you might be making it harder than it needs to be. Wikidata already cross references many iNat places with OSM ways or relations. so i’m guessing you could get most of the mappings quite quickly. it’s just the cases where places no longer exist or have been combined or are are new that would likely require some time to research. i don’t know how clean the data is here, but i don’t think it would take a whole lot of effort to make that determination.
You’re right. I didn’t realize that Wikidata cross references iNat place IDs with OSM IDs! That’s half the battle. Now if only OSM didn’t use such an onerous license we could start switching out some of these crappy shape files (and bog down the iNat servers to a crawl). Ug, what a tricky problem.
it looks like every set is going to have their pros and cons, and GADM seems to actually be better than most in many cases. for Belgium, this tool says that the resolution of the latest GADM set is 269.7 m, versus 33.3 m for OSM, but resolution varies by country.
(so the simplification of the boundaries that i talked about seems to be due to the resolution of the original GADM set itself and less likely due to anything happening during ingestion by iNat.)
I understand iNaturalis’s concerns about server overload. However, should we consider a more optimal way of storing boundaries? This is just a rough idea I came up with and may not be actually useful:
iNaturalist could break up a country’s border into smaller pieces instead of storing a whole border in a single file. This will significantly reduce server load while improving border accuracy. For example, we break up the borders into small pieces of 1°×1° based on longitude and latitude.
When calculating the country to which an observation belongs, server would just find the corresponding small piece according to its longitude and latitude, without comparing it with the borders of the entire country, especially for large countries with irregular borders such as Russia, China, India, and Brazil.
For example, if somebody made an observation at the red point, the server would just find the piece of 102-103°E, 22-23°N，and compare the point with three much smaller kml files:
Also, if an observation was made in 102-103°E, 23-24°N, there is no need to perform complex polygon calculations because this grid is completely within China.
And I guess this approach can also be used for community places.
for what it’s worth, i think the simplification that tiwane is referring to is actually applicable only in some visualizations (for example, see https://forum.inaturalist.org/t/errors-with-imported-kml/43905/5). i don’t think there’s actually any simplification of boundaries for geospatial querying purposes, beyond any simplification that existed in the original source boundaries. in other words, any problems with boundaries is likely just problems in the original source for the boundaries, and it seems like it’s probably just a difficult challenge to get a single source of administrative boundaries that covers the entire world.
China and India border each other (a very long border), at least assuming you accept that Tibet is treated as part of China (which almost all interactive maps do). So it’s nothing to do with how large or small the countries are - if you submit observations from near the border, which side they get categorized as depends on whether the data on where exactly the border runs is precisely correct.
It’s also worth understanding that even the most precise, non-simplified border data isn’t going to be perfectly correct at the level of tens of meters. In fact, a large proportion of the world’s borders haven’t even been defined that precisely, or in some cases even to the level of kilometers. And where borders were defined precisely, but before the GPS era, their exact correct course often depends on the actual locations on the ground of a series of decaying boundary marker stones whose precise coordinates may not even be recorded anywhere at modern levels of precision. (Luxembourg and Belgium are almost certainly exceptions to this though - most borders in Europe have long since been defined at high precision.)
If anyone is interested in this topic, precisely defining borders is called delination and marking their definitive locations on the ground is called demarcation. If you search for news about it, you’ll see that the process is very much still ongoing in many countries.
I don’t know anything about their quality, but a quick search found two open data sources for world boundaries with recent (last decade at least) updates: Opendatasoft’s World Administrative Boundaries World Bank’s Official Boundaries
Both of those are admin level 0 boundaries, which means just countries (and country equivalent territories like French Guiana). We need at least admin level 1 boundaries (states, provinces, and districts).
no, that’s not true. although there is a limit for how large of a file community members can load into the system for their places, iNat staff can load whatever they need to. so the boundaries for the “standard” places are as complex as they need to be in the system. (as noted above, you can see this in the greater precision in the boundaries for USA places, which come from a different source than the boundaries for most of the rest of the world.)
it’s just hard for a single source to have every boundary perfect and at a high precision for every country, state, and county equivalent in the world. and as barbetsmith noted, some boundaries just have not been defined very precisely (or have not been agreed upon), even before anyone tries to represent them in a GIS system. look at my earlier example (requoted below), and you’ll see that no source matches another source exactly for any country. even the boundary for a country like Monaco, which you would think should have the easiest of borders to define, differs between all of these sources.
I marked as Solved because the original issue is not a bug and was explained. There’s a deeper issue of accurate boundaries, but that’s beyond the scope of the specific issue here.
I was going to post the link after I saw this just for everyone else’s reference, but I honestly can’t remember what I typed in to get to the page I got to. I think I was on a page about getting a temporary hunting license as a foreigner (like for a hunting vacation or something) if anyone else was interested in searching for it.
Just a side note, I ended up on that page by chance while trying to look into the hunting laws (I am particularly interested in hunting as a means of conservation and, by proxy, interested in hunting laws of different countries). I didn’t see your post that said “you can hunt foxes in Belgium” and think “Oh, I should go hunt foxes in Belguim. How to I get a foreigner’s permit?”
Even just generating the list of boundaries to extract and mapping those names to the appropriate OpenStreetMap objects would probably take days of work.
