I was pulling some data from Acadia National Park (US) and noticed some of the lakes only exist as members of the park boundary, is this correct? This makes it hard to run a query for waterbodies in the area as some are buried under the park boundary relation.
example
2 Likes
Hi @jkoether, welcome to Discourse and thanks for asking.
Our Tag:leisure=park - OpenStreetMap Wiki says “Lakes and forests in a park belong to it, so don’t use separate multipolygon elements to exclude them!” So the answer to your question is “no.”
What both works and seems to be best for park boundaries is to simply create the park boundary as a polygon (or multipolygon as a relation of that type, if a multipolygon is topologically necessary) and then simply map items “in” the park as “in” the park. These shouldn’t have anything to do with the boundary.
Concomitantly (as a result of this), it would be correct to re-edit Acadia so these lakes aren’t part of the boundary relation, and similarly for other parks that have objects like this “mixed up” in their boundary.
If this seems like a solution to you, please click the “Solution” checkbox for this post.
5 Likes
According to the National Park Service website, these particular ponds on Mount Desert Island are not part of Acadia National Park, with the park boundary following the shoreline. While it’s unclear what historically led to this situation, there appear to be other inholdings within the park that are land-based, so I would leave these alone.
Consider evaluating only members of the boundary relation with the role outer in your query. This should give you all waterbodies contained within the park, including inholdings.
1 Like
This question bumps up against public water access law which can get a bit esoteric. In many regions water bodies meeting certain conditions are considered part of the public domain and access cannot be restricted by a private landowner. For example if someone owns a large parcel of land with a river flowing through it, they often cannot deny people from fishing or boating in the river. Even if a parcel fully encompasses a lake or pond of significant size, the landowner may not really own the lake. For this reason, GIS datasets often exclude water bodies from property boundaries.
It may well be true that technically the National Park Service does not own these ponds because Maine Law designates them a public asset. However, the purpose of National Parks and other public lands is to preserve public access, so I find this distinction to be fairly inconsequential. Most people swimming, boating, or fishing in one of these ponds will think of themselves as being “in Acadia National Park” even if technically it is not the National Park Service, but a different government agency that allows access.
So should the boundary of a park or protected area in OSM exclude water bodies like GIS datasets do? In my opinion, no. I don’t think this serves the general purpose map user well. If you told most people that they have left Acadia National Park once they have waded into the water of a pond, they would be rather confused. Perhaps for other types of land ownership that aren’t intended for public access this would make sense though.
1 Like
Apologies from drifting away from the question (and perhaps it might make sense to split this off into a different question), but separately from whether this example lake is part of the national park or not (already answered above) how does it being an “inner” for the park boundary relation makde it harder to search for? In what context are you searching (inside some software such as an editor, in a database, perhaps for rendering, or something else)?
I’m seriously impressed with @clay_c 's and @ezekielf 's replies which include knowledge of the truly-existing, subtle, often-esoteric complexities of (often federal) park inholdings and Maine water law. Nice job, gentlemen! While I’m certainly knowledgable about them, I did not consider these with my (initially glib) answer and indeed they likely form a part of the total data picture for this case.
The moral / upshot? There are frequently exceptions to a lot of rules! Even as we “discuss amongst ourselves,” OSM must be aware of such exceptions, and when necessary, point them out, including how our data might or does already and correctly encompass them.
I am using overpass to look within an area for waterbodies to generate a model for 3d printing, I’m probably not using the best query…
1 Like
So I had already removed the park boundary relation from those lakes, after seeing the more complete discussion here, should I revert it back?
“So I had already removed the park boundary relation from those lakes, after seeing the more complete discussion here, should I revert it back?”
I think you mean “removed those lakes from the park boundary relation.” Easy to say this backwards!
Based on @clay_c’s response, I’d say “yes.” If the lakes really aren’t part of the park, they should each be in the park multipolygon relation, each with role “inner.” That effectively “subtracts them” from being “in” the park. (Which has the role “outer”).
To be clear, this sort of thing happens, but not always. It’s on a case-by-case basis, and evidently, for this case, “it happens.”
BTW, I climbed Cadillac Mountain, that was way back in 1976. Great view of Bar Harbor!