When tagging a village/town/city, I can use place=city/town/suburb... to mark it as a town etc.

I'm not quite clear on how to use this tag:

  • If I understand correctly, the Wiki recommends: 1) putting the place= tag on a node in the center, and 2) drawing the official boundary of the place as a way and tagging it with administrative=boundary (http://wiki.openstreetmap.org/wiki/Key:place , section "Core values / As areas" ).
  • However, in the database I sometimes see places which have both an area tagged place= and a (different) area tagged boundary=administrative. One example would be Dieren (NL) (search for "Dieren" on Nominatim).

What is the point of this? Isn't this contradictory? Or are there situations where the area of the place differs from its official boundary?

So should I use place= on a node or on an area? If the latter, when would the area of a place=*, be different from its official boundary?

This question was prompted by the discussion on my answer to Way in wrong country.

asked 25 Mar '11, 11:43

sleske's gravatar image

accept rate: 25%

edited 25 Mar '11, 11:44

Does this http://help.openstreetmap.org/questions/4658/tagging-areas-as-place-vs-boundaryadministrative#4736 answer your question? I ended up asking practically the same question as you as I felt that we weren't getting far here.

(25 Apr '11, 19:29) ponzu

Putting stuff that has some extend (say if seen from above it is bigger than 1 square meter) as an area is generally better, because it stores information about position and size in greater detail. It also allows for later evaluation, which objects are contained.

Therefor I'd prefer to map place on a closed way (usually referred to as built-up area in contrast to administrative boundaries).

permanent link

answered 25 Mar '11, 12:00

dieterdreist's gravatar image

accept rate: 3%

@JoshD would you mind to explain why you don't think that mapping places as areas is fine (I saw you voted the answer down).

(25 Mar '11, 13:40) dieterdreist

"putting stuff as an area[...] is generally better": I'd agree usually, but specifically with places a "center node" provides information that an area does not, namely where the center of the place is - this is often not the geometric center, but some place like downtown, the shopping district or the townhall. This is useful for e.g. calculating distances between cities, and for showing the label in a sensible place.

(25 Mar '11, 16:51) sleske

I agree. but the "showing the label in a sensible place" can also be a problem: at the real center of a place the features are often crowded, and mappers tend to move the place-node away from this center in order to have as much information rendered as possible.

(25 Mar '11, 22:35) dieterdreist

@dieterdreist: Putting the place-node outside the place's center to help the renderer would be "tagging for the renderer", which is strongly discouraged. Anyway, it shouldn't be too hard for a renderer to move the place name a bit aside to make room for other features.

(31 Mar '11, 10:54) sleske

@sleske: nah, I feel that you are seeing "tagging for the renderers" too narrow. There is no "center" as a point in a city. In almost all cases you might opt with good arguments for several spots to declare as a "center".

(31 Mar '11, 12:24) dieterdreist

@dieterdreist: "There is no "center" as a point in a city." Well, at least where I live, in each city/town there is usually one small area that is generally considered the "center", and that people will refer to as "center". Usually it's the central market square, or the town hall, or similiar. Providing this "center" seems useful for routing engines (though it's debatable whether it should be rendered on maps). If no such notion of "center" exists for a place, obviously no center node will be necessary.

(14 Apr '11, 10:38) sleske

@sleske: Yes, you confirm, it is an area, not a "point", that's what I intended above: people are moving the node(=point) around in the area perceived as "center" to optimize rendering.

(14 Apr '11, 17:31) dieterdreist

The solution for rendering (and programmatically identifying) the "center" of any admin boundary lies with the use of the "admin_centre" role and, perhaps, the "label" role as documented here: http://wiki.openstreetmap.org/wiki/Relation:boundary Unfortunately, few mappers and even fewer renderers honor this improved schema.

Note how it suggests that you take an existing object, such as city hall or market square and add it as a role=admin_centre to the relation. That's just one reminder to keep entities to a necessary minimum. (Hint: having a node and a boundary for the same place is bad.)

(14 Apr '11, 20:14) ponzu

Why would a marketplace be an "administrative centre"? The location of the administrative centre is the seat of the administration (maybe there might be also more then one?)

(15 Apr '11, 12:13) dieterdreist
showing 5 of 9 show 4 more comments

I think we need both to make searches and nomination work correctly.otherwise what limits the boundary of a town or suburb. please see my question "how-can-i-find-and-modify-a-boundary" I have been trying to map some local suburbs and parishes but they seem to leak out miles I'm not sure why this is.It is made more difficult as there seems a very long delay (weeks) for boundaries to take effect on the nomination

permanent link

answered 14 Apr '11, 18:29

andy%20mackey's gravatar image

andy mackey
accept rate: 4%

Thank you for pointing this out. I think we are involved in a complicated dance between maintaining OSM data integrity, nomination (I assume you refer to the algorithm of building Nominatim database off OSM data) and rendering. We do not "tag for renderers" and we do not "tag for services", yet we have to keep them in mind. It seems to me most of these questions are asked from the standpoint of one or the other (and maybe a third). I don't see anyone looking at the holistic solution. Maybe the solution is to fix Mapnik. Then we can start losing the place nodes that are throwing nomination off.

(14 Apr '11, 20:20) ponzu

And again, good point about Nominatim taking forever to update and it being frustrating for projects such as yours. I saw Twain's explanation somewhere that additions to OSM data are reflected quickly, but updates take very long (by design - otherwise his server would get overwhelmed). I wonder how quickly - or if at all - he processes deletions. I deleted a town point yesterday (keeping the boundary of course), just to test the effects on nomination. Surely all this is documented - or at least discussed - somewhere. Or am I dreaming?

(14 Apr '11, 20:23) ponzu

have a look at this question, which seems to cover the same topic:

permanent link

answered 25 Mar '11, 11:56

dieterdreist's gravatar image

accept rate: 3%

Yes this is similar, but it's a bit more general. I wanted to ask specifically whether (and when) it makes sense to create an area for the place= in addition to an administrative boundary. If the exact boundary is unknown, it might be better to have at least an approximate area for a place, rather than just a node.

(25 Mar '11, 12:02) sleske

see the answer below. Place is to map settlements, this has not much to do with administrative boundaries (which cover the settlement plus all other area that is administrated from the settlement, like fields, forests, ...)

(25 Mar '11, 13:27) dieterdreist
Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here



Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "title")
  • image?![alt text](/path/img.jpg "title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Question tags:


question asked: 25 Mar '11, 11:43

question was seen: 10,574 times

last updated: 25 Apr '11, 19:29

powered by OSQA