From OpenStreetMap Wiki
Jump to: navigation, search


I've added this proposal to the relation proposal page Relations/Proposed/Boundaries. Please add your opinion to this proposal


Not sure whether national applies to US states such as "Texas" or whether it demarks the US borders. Feel free to clarify if you know. --spaetz 07:20, 11 July 2007 (BST)

I'd say it applies to US borders. The Republic of Texas might be an unusual case. ;-) --Hawke 18:30, 25 February 2008 (UTC)
US is a federal republic so each state is under the federation, thus lower admin level. --Skippern 13:10, 3 December 2008 (UTC)


Should this just be a boundary=city key? Should we use whatever the place is, ie boundary=village/town/city? Or should this just be a generic boundary=place key with a clarification of place=city;name=Munich in addition to it? What do people think? --spaetz 08:42, 12 July 2007 (BST)

boundary=administrative, admin_level=10, border_type=muncipality?

Expanded usage

I thing we should distinguish two roles of boundaries:

  • Boundary of an area (circular way) - in this case it's boundary of one thing - which has one name (county, city, etc).
  • Linear boundary betwean two areas - in this case it's boundary between two things (counties, regions, etc.).

Clearly we need both of these usages and there are some cases some segments could be reused for both purposes. In either case there are different tags needed. And this is so different that different key might be needed to avoid confusion. AFAIK the original usage of boundary was only for areas. Which might cause problems of course. --Gorn 00:22, 20 July 2007 (BST)

National Park??

Does boundary=national_park really fit here? If the boundary tag is extended beyond political/civil/administrative boundaries, where does it end? Everything has a boundary, if you look at it that way. --inas 06:32, 13 August 2007 (BST)

Certainly, the set of things to be tagged with boundary is small. National parks are, however, political boundaries, often not corresponding to any identifiable features on the ground, even including urban areas, so seem an entirely appropriate thing to tag here. As an example of what I'm talking about, have a look at the google maps version of the Peak District National Park in England: the green bit between Manchester, Sheffield. There's no identifiable line on the ground round the edge of this, the moorland certainly doesn't stop there, so tagging this with some physical feature like natural= will not be appropriate. Morwen 09:11, 13 August 2007 (BST)
So, you are proposing essentially that the landuse (or similar tag) indicate what is actually there, and the boundary tag define the legal boundary. So for example where a state forest exists, I would use the tag forest= to mark the landuse, but if that landuse continued beyond the end of the actual legal forest area, I would use a tag like boundary=forest for define the legal boundary of the forest? Is that the idea? --inas 01:43, 14 August 2007 (BST)
Yes. Morwen 10:57, 14 August 2007 (BST)
I think boundary is adequate, too. Moreover, I'd suggest re-baptizing the tag as boundary=nature_reserve; there are lots of different proctection categories, and "national park" is just one of them. The nature reserve type could then be specified by means of an additional tag, corresponding to the respective local legislation, let's say, Naturschutzgebiet or Landschaftsschutzgebiet in Germany, or reserva extrativista in Brazil. --Ulf Mehlig 19:05, 12 May 2008 (UTC)

What happened to the boundary=national_park definition? Can't find it now. In fact I can't find a revision of this page which mentioned it. -- Harry Wood 12:21, 8 April 2008 (BST)

adding boundaries

Does anyone have an example of where I can look at how this has been done in potlatch? I think parts of my home areas need boundaries added to make the search reasonable, but I can't see any borders marked anywhere. -- Kpalsson -11:16, 21 August 2007

Use of border_types in Germany

I would suggest to use border_type=state for Bundesländer and border_type=province for Landkreise and Stadtkreise. Any comments? --Dr.Nop 10:19, 8 October 2007 (BST)

IMHO it would be good to use Eurostat NUTS terminology, see and some examples , so it would by border_type=nuts1.

I looked at this in the UK. NUTS1 does correspond to real administrative regions, but NUTS2 and NUTS3 refer to arbritrarily grouped counties and districts, and this would not be useful for tagging administrative borders. Morwen 09:24, 29 November 2007 (UTC)
for Germany NUTS and LAU classification looks good.
NUTS 1 Bundesländer (16) --> level 4
NUTS 2 Regierungsbezirke (39) --> level 5
NUTS 3 Kreise (429) --> level 6
LAU 1 (NUTS 4) Samtgemeinden, Verwaltungsgemeinschaften (1457)
LAU 2 (NUTS 5) Gemeinden (12379)
but should capture the NUTS ans LAU classifiction, but maybe in their own boundary-relations.

I propose to use the admin_level as follows:

Some "Bundesländer"/federal states (not all...) have "Regierungsbezirke" between "Bundesland"/fed. state and "Landkreis"/county: admin_level=5??? see
In some "Bundesländer"/federal states (again: not all... and sometimes not the same ones...) the "Gemeinden"/municipales might be organized in "Samtgemeinden", "Verbandsgemeinden" or "Ämter" between county and municipal: admin_level=7??? see
a little bit complicate... :-) --Mueck 16:50, 23 February 2008 (UTC)
you're right. so I added admin_level=7 and admin_level=5 for the administrative province-districts (Regierungsbezirke) and the local-admin-districts (Bezirke in einem Landkreis) in my example. --Cbm 12:31, 22 July 2008 (UTC)

Maybe better that way?

admin_level for Germany
Tag Description NUTS/LAU
admin_level=1 n/a
admin_level=2 international borders (Land)
admin_level=3 n/a
admin_level=4 federal state / province borders (Bundesland) NUTS 1
admin_level=5 state-districts / province-districts (Regierungsbezirke) NUTS 2
admin_level=6 countys / 'district-free cities' (Landkreis/Kreis/kreisfreie Städte)
admin_level=7 county- / 'district-free city'-districts (Bezirke)
admin_level=8 amt [1] (Samtgemeinden, Verwaltungsgemeinschaften)

admin_level=9 city/municipal (Städte/Gemeinden)

admin_level=10 suburbs (Stadtteile)

Use of border_types in Germany 2

In der deutschen Mailingliste wird das Thema gerade diskutiert: Nochmals grenzen und admin_level. Es zeigen sich einige begriffliche Verwirrungen und Probleme resultierend aus dem bunten Föderalismus-Kuddelmuddel etc. ;-) ICH würde vorschlagen, das ganze im unteren Bereich neu zu sortieren.

On talk-de (see also here) this topic is just in discussion and it seems, that there are some problems because of unclear definitions. I would propose a new sorting of admin_levels.

admin_level for Germany
Tag Description NUTS/LAU
admin_level=1 n/a
admin_level=2 international borders (Land)
admin_level=3 n/a
admin_level=4 federal state / province borders (Bundesland) NUTS 1
admin_level=5 state-districts / province-districts (Regierungsbezirke) NUTS 2
admin_level=6 countys / 'district-free cities' (Landkreis/Kreis/kreisfreie Städte)
admin_level=7 NEW AT 7 INSTEAD OF 8: amt [2] (Samtgemeinden, Verwaltungsgemeinschaften)

admin_level=8 NEW AT 8 INSTEAD OF 9: city/municipal (Städte/EINHEITS-Gemeinden)

admin_level=9 NEW AT 9: suburbs level 1 meist Stadtbezirk genannt, mostly called Bezirk of a town / Oder Ortsteile von Einheitsgemeinden, Merkmal: in der Regel eingeschränkte Selbstverwaltung (Bezirks-/Ortsrat) / mostly limited administr./political organizations

admin_level=10 NEW AT 10: suburbs level 2 meist Stadtteil genannt, mostly called part of a town, Merkmal: in der Regel keine Selbstverwaltung, oft historische Unterteilung / mostly no administr./political organizations, often historic

admin_level=11 NEW AT 11: suburbs level 3 meist Stadtviertel genannt, mostly called quarter of a town Merkmal: in der Regel nur statistische oderhistorische Unterteilung / mostly statistical, historic


  • level 7 wird offenbar nirgends gebraucht, Teile von Kreisen sind unüblich bzw. haben meist nicht administrative Anlässe. Stadtbezirke sollten generell unterhalb der Stadt als Ganzes sein, denn es gibt kreisangehörige Städte, die auch in Bezirke eingeteilt sind (Saarbrücken bspw.)
    • level 7 seems not to be used...
  • level 8 wird in anderen Staaten oft für Gemeinden verwendet, vor allem auch dort, wo 7 und 9 nicht belegt sind, siehe Hauptliste, daher: gewisse Einheitlichkeit
    • level 8 is used for it in a lot of other countries for this
  • level 9-11: In einigen Städten sind alle drei Ebenen zu finden, meistens hat 9 die Selbstverwaltung, 10 nicht und 11 oft nur noch statistische Einteilung bzw. historisch gewachsene Baugebiete. Es kann Ausnahmen geben: in Bremen sind die Selstverwaltungen offenbar auf 10 statt 9 zu finden. Näheres müsste Stadt für Stadt geklärt werden, ebenso wie andere levels, am besten auf einer eigenen noch zu schreibenden Wikiseite ...
    • in a lot of cities all 3 levels are found

Der Aufwand der Umnumerierung scheint sich noch in Grenzen zu halten nach flüchtigem Blick via OSMinspector --Mueck 17:06, 25 November 2008 (UTC)

Habe mal angefangen, unter DE:Grenze Daten zu sammeln zu den Eigenheiten der deutschen Verwaltung in den einzelnen Ländern als Leitfaden zum Mappen von administrativen (und später anderen) Grenzen. --Mueck 19:25, 27 November 2008 (UTC)
Nachdem sich weder hier, noch in der Mailingliste oder im Forum nennenswerter Widerspruch regt und es eigentlich eine Anpassung an das internationale Schema ist, sollte man das einfahc jetzt so umsetzen, bevor noch mehr Grenzen nach altem Schema getaggt werden. Statistische Unterteilung (level 11) kann man erstmal zurück stellen, das scheint nicht so 100% einfach, weil Stadtteile und statistische Gliederung bspw. in bayrischen Städten anscheinend unterschiedlich sein können. Mal abwarten...
Because of no opposition and because it is only a changing to the international system, we should do it now
--Mueck 18:27, 29 November 2008 (UTC)

Stand aktuell: Westdeutschland außer Großraum München weitgehend umgestellt nebst anderen Bereinigungen, wenn man schon die Daten anfasst... München und Ostdeutschland mit zwei größeren Brocken rund um Sondershausen und Berlin fehlen noch. --Mueck 01:39, 2 December 2008 (UTC)

mal updaten: bin kurz drauf ganz durch gekommen. Bei Gelegenheit muss ich noch 'ne Feinkontrolle machen... Nächstes Jahr oder so... :-) DE:Grenze will ich auch noch weiterführen. Ebenso nächstes Jahr oder so :-) --Mueck 14:50, 17 December 2008 (UTC)
dann müssen wir nur noch die Ungarn, Schweden und Türken davon überzeugn ihr Schema von Nuts/Lau den unsriegen anzupassen :-P --Cbm 11:59, 2 December 2008 (UTC)
nicht dem umsrigen, dem europäischen! Aber vermutlichhat's ein deutscher Europäer erfunden? Wer weiß ... ;-) Praktisch wäre es, ja ... Aber im Prinzip müsste man erst mal für den Rest Europas nachtragen, was wo welche NUTS/LAU-Klasse ist, fehlt ja noch bei etlichen ... Bei Gelegenheit... :-) --Mueck 12:43, 2 December 2008 (UTC)

Warum admin_level 3, wenn A) Deutschland ein eigenes Sueppchen kocht, B) der Level in Deutschland nicht benoetigt wird, C) der level eigentlich grundsaetzlich ueberfluessig scheint? --traut 15:59, 16 December 2008 (UTC)

b) wäre kein Argument, c) wäre eins, ist aber HIER in der Diskussion um deutsche Grenzen nicht zu lösen, denn sonst träfe a) zu, wenn D level 3 rausschmeißt und so fürderhin ein eigenes Süppchen zu köcheln beabsichtigen täte, was mit der nun durchgeführten Änderung nicht mehr so ist, da die Gemeindegrenzen nun wieder auf 8 liegen, wo sie im internationalen Schema hingehörten. --Mueck 14:48, 17 December 2008 (UTC)

further expansion

Right, this is going well in the UK so far. However, there is weirdness in that parish and county borders are shown at the same weight and on the same zoomlevels. Because I don't want to hardcode the names 'parish' and 'county', I'm instead going with a system to indicate significance by a number.

I'm presently modifying the t@h stylesheets to do this.

so the system is, in the UK

  • admin_level=6 on county borders;
  • admin_level=8 on district borders;
  • admin_level=10 on parish borders

i've left gaps so we can infill if needed. this is intended as only a temporary solution until such time as we get per-country stylesheets, which are Quite Hard, so I appreciate this might end up lasting longer than hoped for. Morwen 21:06, 28 November 2007 (UTC)

Where do UK unitary authorities fit in this? Are they counties (I guess so), but many are effectively cities Chillly 11:41, 9 December 2007 (UTC)
admin_level 6, I would say. Morwen 12:07, 9 December 2007 (UTC)

Very local use

I've used boundary=college and admin_level=10 for local use, which is likely to help a custom rendering that I might do. See [Durham#New_tagging_system], I hope it won't horribly interfere. - LastGrape 14:26, 5 December 2007 (UTC)

Hmm, how about amenity=university? --Hawke 22:55, 22 February 2008 (UTC)

left:X right:X

Wouldn't it make more sense to use the same digits for left:X and right:X than the admin_leve rather than names that might not be appropriate in some places. Currently:

left:parish, left:district, left:region, left:province, left:state

But how do these work in countries with departments, cantons, arrondissements, deelgemeenten, prefectures, etc. ? Why no just use something like

left:4, left:6, left:8, left:10

with the same scale system as admin_level? --Moyogo 00:31, 5 February 2008 (UTC)

Currently, nothing is done with these tags, but it would seem logical to use this system for naming what is either side of the border. --Edgemaster 20:05, 6 February 2008 (UTC)
Relations is a better solution, as it removes the need for left/right names. Also this way the border only need to be tagged with highest admin level. --Skippern 12:54, 2 December 2008 (UTC)
Yes, relations are better for it. And yes, I also prefer an admin_level at border itself additional to the admin_level inside relation. But others have another opinion... --Mueck 21:31, 2 December 2008 (UTC)

The practice of left/right is damaging to the approval of changes to boundaries. We are now holding a vote about how to implement Maritime borders, but the question about left/right is becoming a serious issue. During the RFC face, people objected because the lack of left/right, and now during the vote, people are rejecting it because it contains left/right. I totally agree that left/right is a bad way of doing it, and relations are a much better solution. But left/right have been in practice for some time, and is not likely to go away like that, even though relations have taken over the purpose of left/right. If it was up to me, left/right should have been removed long ago. --Skippern 14:05, 3 March 2009 (UTC)

Admin_level values for specific countries

Discussion of existing admin_level values should go here, as should new countries, where there is doubt about which level to use for a boundary


Similar to Germany I'd like to propose:

  • admin_level=2 - National border (Staatsgrenze)
  • admin_level=4 - Federal states border (Landesgrenzen)
  • admin_level=6 - County borders (Bezirksgrenzen bzw. Städte mit eigenem Statut)
  • admin_level=7 - Special Divisions in county of Liezen, Styria (Politische Exposituren im Bezirk Liezen)
  • admin_level=8 - Municipalities (Gemeindegrenzen)
  • admin_level=9 - Divisions of cities (Stadtbezirke)
  • admin_level=10 - (Katastralgemeinden)
    • Katastralgemeinden sind keine administrativen Einheiten, bei den Wiener Stadtbezirken sogar nicht mal grenzidentisch laut Wikipedia-Karte --Mueck 17:10, 25 November 2008 (UTC)
  • Ich würd die schon als Verwaltungsgrenzen sehen (Einteilung des Grundbuchs); dass die Grenzen mit der höheren Ebene nicht immer übereinstimmen ist wahr. -- A uller 13:12, 2 December 2008 (UTC)
Richtig, sie sind von einer amtlichen Behörde, haben aber trotzdem nix mit Verwaltungseinteilung (administrative) zu tun... Es darf aber gerne boundary=Kataster erfunden werden ;-) Wie auch boundary=Forstbezirke etc. --Mueck 21:34, 2 December 2008 (UTC)


The following is proposed:

  • admin_level=2 for the border around Belgium
  • admin_level=4 Region border
  • admin_level=5 Community border
  • admin_level=6 Province border
  • admin_level=7 Administrative arrondissement border
  • admin_level=8 Municipal border
  • admin_level=10 Suburb and Locality border
  • -or-
  • admin_level=10 part-municipalities (=deelgemeentes(nl); section/ancienne commune/sous-commune(fr))


Second proposal (by AiNikolas , but modified by Logictheo ) To see the first set of changes made by Logictheo click here.

  • admin_level=2 - National border (Εθνικά Σύνορα)
  • admin_level=3 - no current use
  • admin_level=4 - Periphery border (Περιφέρεια)
  • admin_level=5 - no current use
  • admin_level=6 - Prefecture border (Νομός)
  • admin_level=7 - Municipality border (Δήμος)
  • admin_level=8 - Municipality district border (Δημοτικό Διαμέρισμα - now called Τοπικό Διαμέρισμα)
  • admin_level=9 - Village, settlement +/- neighborhood (Χωριό, οικισμός +/- συνοικία)
  • admin_level=10 - Neighborhood ("Συνοικία"), (if not implemented in level 9)

Kallikratis administration reform

There are major changes in Administrative structure of Greece (Kallikratis) (english), since many Municpalities have been degraded to "Unities" (Δημοτική Ενότητα, πχ Triandria Municipality in thessaloniki) and districts (διαμερίσματα) are now considered "communities" (Κοινότητα).

So in our mapping to the levels, we have to adjust levels 7, 8, 9 and 10, i.e. to include the Unities. This is going to create some chaos with the already inserted boundaries. For example, if we insert the Unities at level 8, then all the former Districts which were previously mapped with 8, have to be shifted to 9. This is a problem, because at the moment according to tagwatch there are 839 entries with admin level 8, 320 with admin level 10 (and only 22 with admin level 9). If we decide, we can enforce an automatic shift of the levels in Greece, which should be doable through the API. As another level of granularity I would propose the "Electoral Districts" (εκλογικά διαμερίσματα) (e.g. look at the official electoral division of the 2nd Municipal Community of Thessaloniki)

Anyway, the simplest proposal would be

  • admin_level=2 - National border (Εθνικά Σύνορα)
  • admin_level=3 - Decentralized Administration border (Αποκεντρωμένη Διοίκηση) - optional
  • admin_level=4 - Periphery/Region border (Περιφέρεια)
  • admin_level=5 - (Former) prefecture border (Νομός) - prefectures were abolished in 2011
  • admin_level=6 - (New) regional unit border (Περιφερειακές ενότητες) - the majority of regional units coincide with former prefectures, but some prefectures were split
  • admin_level=8 - Municipality border (Δήμος)
  • admin_level=9 - Municipality Unity border (Δημοτική Ενότητα)
  • admin_level=10 - Municipal Community border (Κοινότητα)
  • admin_level=11 - Electoral District border (εκλογικό διαμέρισμα), Village, settlement +/- neighborhood (Χωριό, οικισμός +/- συνοικία)

--Radiobuzzer 23:20, 8 May 2011 (BST)

United States

How to handle:

  • school districts
  • postal codes
  • telephone area codes
  • PLSS divisions
    • Sections
    • half-sections
    • Quarter-sections
    • Half-quarter-sections
    • Quarter-quarter-sections
  • census-designated places (do not have own government)
  • city-defined districts & neighborhoods

Place boundaries were recently imported in Ohio. This included census-designated places, as admin_level=8 (the same as incorporated cities & towns). I think CDPs should be level 9 or 10. They don't have any legal status, or represent the jurisdiction of any government. CDPs are created by the Census Bureau for statistical purposes only. They can be somewhat arbitrary. Often, the existence of a CDP isn't even known by many of its residents. Furthermore, CDPs exist in and are not exclusive of townships (level 8). Therefore I believe they should be tagged at a level of lesser significance. Vid the Kid 07:17, 30 July 2009 (UTC)

Religious authority boundaries


Very basic and there are some exceptions to this model (for example Gibraltar and The Falkland Isles) boundary=episcopal; denomination=anglican

Anyone care to comment on what the equivalent structuring for other churches is? ShakespeareFan00 00:20, 24 February 2008 (UTC)

Why not use admin_level=8 for Deanery? --Hawke 16:05, 3 March 2008 (UTC)

I would suggest the following for Anglican Communion:
My understanding is that while primacies are not always coterminous with countries, they are in religious terms what countries are in policital terms, and that not all parishes are necessarily part of a deanery. That said, I am technically an apostate and would stand to be corrected. Chriscf 16:07, 3 March 2009 (UTC)
Sometimes parishes group into benefices -- typically a group of parishes sharing a priest. It might make sense to bump archdeaconries and deaneries down one place to accommodate this. ras52 17:17, 19 May 2011 (BST)

Roman Catholic

Draft based on some French Dioceses organisation. Must be discussed...

boundary=religious_administration religion=christian denomination=roman_catholic

FrViPofm 10:05, 21 September 2009 (UTC)

A very similar draft based on the Hungarian system:

See more (in Hungarian): WikiProject_Hungary/Katolikus_Templomok

--Borazslo (talk) 08:18, 18 July 2016 (UTC)

Judaism: Eruvin

May I solicit ideas on how to map and tag eruvin [3]? See, for example, [4], [5] and [6] for the three eruvin in my city. -- Gridlock Joe 21:40, 17 June 2010 (UTC)

Because it may include things that serve other purposes (actual walls, poles, etc.), I'd tag those objects as usual, tag eruv wires as barrier=wire, and put them all in a relation tagged type=boundary + boundary=eruv. AM909 11:26, 2 March 2011 (UTC)



Proposed UK schema

  • European Parliament
  • UK (Westminster Parliament)
  • Scottish Parliament/Welsh Assembly/NI Assembly
  • County/Unitary authority members
  • District Council seats
  • Parish/Town council seats - Are the equivalent to individual wards..?

ShakespeareFan00 17:32, 25 February 2008 (UTC)

If feasible, it would be nice to have this in a manner which could mesh with the US system(s), but that might be difficult unless we have someone very familiar with both, and several states in the US.
  • Congressional districts (435)
  • State Senate districts (Legislature, for Nebraska, which has a unicameral legislature)
  • State House (or Assembly) districts
  • City Council (Aldermanic) districts
Hawke 18:42, 25 February 2008 (UTC)
In the UK, the noted levels already duplicate the existing admin_level values that exist - the areas map directly to each other. admin_level would probably be extended to 11 to cover wards - although the boundaries for wards change often and the datasources are nearly impossible to get freely. (This idea has been trashed about on the mailing lists several times in the past) --Edgemaster 18:46, 25 February 2008 (UTC) says "The area which a council covers is divided into one or more electoral divisions - known in district and parish councils as 'wards', and in county councils as 'electoral divisions'. Each ward can return one or more members - multi-member wards are quite common."
I believe these "electoral divisions" and "wards" do not map to the existing admin_level values...a ward for electing a particular seat of the council has no bearing on how the council governs, and is exclusively for election purposes (and therefore not really an administrative boundary). However, I am not in the UK and therefore not really qualified to comment on it, beyond what I can find in Wikipedia. --Hawke 21:53, 25 February 2008 (UTC)
There are specific rules for some Westminster constituencies (the most explicit regarding the City of London, Orkney and Shetland), however, there is no guarantee that any constituency boundary falls within any other boundary - e.g. the Redditch constituency contains not only Redditch district but also some villages to the south in Wychavon, most of the rest of which is represented by the Mid Worcestershire constituency. The only boundaries that are almost certain to be avoided are the ward boundaries (since this is the basis for distributing the ballot boxes), and to a lesser degree the county boundaries in England (which the Boundary Commission is averse to crossing). In the US, things get a lot more complicated, due to their propensity for using meaningless numbered districts rather than geographical ones - with redistricting every ten years or so, you may find yourself represented by someone different each time, and long-term elected members regularly find their constituency changed. Except perhaps in California, where the system is rigged to keep US Representatives in their seats. Chriscf 15:54, 3 March 2009 (UTC)
boundary=political is supposed to be electorial borders, so maybe it rather should be used together with border_type=constituency in this case? I do not know how how these constituencies are devided, just a suggestion. boundary=political is approved but not documented, maybe somebody can document it? --Skippern 10:37, 17 August 2009 (UTC)


Proposed schema - boundary=postal

In order of descendign mangnatitude..

  • Province
  • Postal Area
  • Postal Area Subdivision
  • Postal District
  • Postal Sector
  • Postal Unit

Any comments? In particular I'D like to know how this coudl be mapped to a hireachy. ShakespeareFan00 17:49, 25 February 2008 (UTC)

First Nations

In Canada some areas of land are designated as Indian Reserves, held in trust for First Nation bands. Some of these lands cross provincial borders and even cross the U.S. border (see Akwesasne )

Suggestion: First Nations and Indian Reservations should be boundary=administrative; admin_level=1; border_type=first_nation; as they are international.

Should borders like this really be administrative? Certainly using admin_level=1 for this just looks wrong to me... --Eimai 17:22, 25 June 2008 (UTC)
I agree with Eimai. If they are international borders, they should be treated as admin_level=2 just like other international borders. --Hawke 05:48, 26 June 2008 (UTC)

I would suggest that we remove these from admin_level=1, and use something else instead. We now have a boundary=national_park, which arguably has some administrative function. In the same way as the various equivalents of "National Parks" in other countries, these cross various administrative boundaries, and should probably be considered as being orthogonal to county and state-level borders. For example, the Brecon Beacons National Park includes land in no fewer than nine counties. boundary=indian_reserve perhaps? Chriscf 11:57, 3 December 2008 (UTC)

It can also apply to other native people, such as the Masaai in Kenya/Tanzania or the Samii in Norway/Sweeden/Finland/Russia. boundary=native_reserve or boundary=native_nation is probably better. But I fully support removing them from boundary=administrative. Let admin_level=1 be reserved for supernational administrative borders such as the European union. --Skippern 13:15, 3 December 2008 (UTC)
I agree with Eimai ... Skippern. In spite of the great respect I can have for those Nations and Peoples... There are other solution, like the Skippern's one. FrViPofm
As a Cherokee, I find boundary=indian_reserve and boundary=native_nation to be intrinsically racist. While admin_level=1 is probably not right, I believe First Nations lines are administrative boundaries, the admin_level=* of which must reflect the degree of sovereignty agreed upon by treaty between the nation in question and the United States (or other potentially subjugating force). For example, the Cherokee Nation would be admin_level=2, whereas The Confederated Tribes of Grande Rhonde would be somewhere closer to the 3-5 range, and even smaller nations that got more heavily screwed over by the United States might fall in the 6-8 range. Yes, I realize this means Canada and the US and their member provinces, states, ridings and counties, would get turned into swiss cheese by most renderers, but are we looking for a map that looks like every other map, or a map that is accurate and objective? --Paul Johnson 04:11, 10 August 2010 (BST)
I see your point Paul, but still a different tag than boundary=administrative should be chosen. How to actually solve this for not being racist, exclusive, eliminating, or whatever negative effects they may have, is open for debate. I agree that their borders should be rendered in some way on maps. When travelling in Kenya I would have loved to see where I passed the borders of the Masaii territories, when travelling in Brazil I would love to know when I pass through or close to the various native populations there, etc. This can have just as much interest for tourism, economy, history and much more than a pure mapping point of view. --Skippern 08:03, 10 August 2010 (BST)
Perhaps a non-numeric value for boundary is best, given the variety of local contexts. Paul, can you suggest a boundary=text that would be respectful and accurate? Are any of the terms used in this article [7] suitable? Rw 01:48, 15 February 2011 (UTC)
Yes, I can. boundary=administrative, admin_level= equal or less than 4, depending on the contents of their treaty with the respective invasive powers that be. Indian nations have set administrative boundaries and are sovereign in, at a minimum, the same degree Scotland is sovereign from the UK, though some are more sovereign than that (particularly ones that cross the US-Canada or US-Mexico boundaries). Paul Johnson 21:20, 14 July 2012 (BST)
I have been drawing some local borders and have also been trying to figure out the best tag for them. From reading this BIA FAQ, it sounds like the legal status of (federally recognized) tribes is similar to that of US States: they are sovereign governments, but are subject to Federal laws, and (like States) cannot legally declare war, issue currency, etc. From this perspective, admin level 3 or 4 seems most accurate. This applies in the US only, of course--- for all I know the Akwesasne boundary should have one level on its Canadian side and another on its US side. --WimL 02:47, 18 July 2011 (BST)
So - what are people using now? I just took the time to draw Morongo's border and don't know how to tag it. AM909 11:32, 2 March 2011 (UTC)
boundary=aboriginal_lands is the current concensus. I'd leave off admin_level all together as aborginal_lands is really orthogonal to administrative in many cases. I'm updating the page accordingly to remove the suggestion of admin_level=1 Pnorman 21:29, 26 November 2012 (UTC)

no use for admin_level=3

  • Is there any need for level 3? I don't see any helpful difference here. Even where it is applied, it could be shifted one level down there, compared to other countries. I'd recommend to move everything else one level up. What was the purpose of this level? --traut 16:35, 24 July 2008 (UTC)
  • Why bother? It can be useful, as for whatever reason a government would come with yet another administrative level. That why level 1 is not used either. One day the EU might become a real "country", that day we just start using level=1 instead of mass changing everything. Furhter more it saves the people who make the stylesheets for mapnik/osmarender/kosmos/whatever, time. --Skywave 16:54, 6 August 2008 (UTC)
  • It's actually all country specific. In Greece admin_level 3 was abolished, but I guess it's now replaced again as admin_level 3, by municipalities. We have 10 levels I think, and it gives enough room for all decisions various governments make. Logictheo 10:07, 30 August 2008 (UTC)

opengeodb does take ten levels as well. But all of them are currently in use:
1) continent e.g. Europe
2) country (state) e.g. Deutschland
3) federal state e.g. Bayern (Bavaria)
4) Regierungsbezirk e.g. Oberpfalz
5) district (Landkreis)
6) community (Gemeinde)
7) town/city (Ort)
8) quarter (Ortsteil)
9) street/road
10) single place (e.g. house number)

Not all of those levels are available everywhere - but they do require all levels, while some of the current boundaries appear to be always unused. --traut 15:11, 11 September 2008 (UTC)

Level 1 is actually unused. In fact it is actually much too large, and still inapplicable worldwide due to the impossibility to find equivalences between various international organizations or common markets. In addition they may even overlap (e.g. in Europe: the EU, the Council of Europe, the CommonWealth, the Francophonie, the Shengen Space, the EFTA, the Central European Free Market Association).
If you try to limit yourself to geographical continents only, it's impossible due to overseas of France, Spain, Portugal, Netherlands, UK, USA : do you want to split these countries despite they are a single UN member ? Now consider Turkey, Kazakhstan, Egypt : will you split them even though they form a aingle contiguous area spanning territories on several geographical continents?
There are really a LOT of international organisations, some are very small and limited to 2 countries (e.g. Switzerland + Liechtenstein form a single customs union).
It would be just simpler to create relations for grouping countries in collections, without adding any ways as members, but only listing the country relations as members.
Now how many international organizations will you list ? There are tons of multinational treaties, plus treaties grouping not only UN members but also other international organizations or customs unions (e.g. the EU itself is a member of the Council of Europe), or only smaller regions of specific countries.
Every country has its own specific internal organization. And there's NO clear distinction between international treaties (each treaty has its own conditions of memberships or participations for observers; this will become even more complex when the members of each treaty can also define parts of their territory as being excluded from the application of the treaties : this also includes the EU, where alsmot all members have exclusion areas).
I suggest not mapping them. It is still possible to use the OSM database to generate derived maps (e.g. in SVG files stored in Wikimedia Commons and referenced in Wikipedia articles describing them more completely).
The OSM database however will remain useful to list the smaller subareas with specific updatable features that can be used as a the source for regenerating those updated maps.
Verdy_p (talk) 07:10, 12 June 2012 (BST)


In Brazil the following have been used, and I am continuing that use:

  • 4: State, e.g. between Minas Gerais and Espirito Santo
  • 8: Manincupal, e.g. around Bela Horizonte

Shared boundary

How shall we tag a boundary that is used for many levels? E.g. a level-6 border that is also a part of a level-3 border? Should relations be used for this, or can we cope with it by using keys?

  • We should, perhaps, have used a numbering scheme that allows adding the levels together:

100000= national; 10000 = state; 1000 = district ...

Then the levels could be added and still make sense: 111000 is a shared national, state and district border. We could use a smaller radix, but binary codes for such things might be too easy to make mistakes. Chillly 14:21, 8 September 2008 (UTC)

I was under the impression, that the reason for the admin_levels is, that lower numbers of borders are automatically also borders for all numbers higher than that. E.g. a level-2-border is also a border for level 3,4,5...; a level-7-border is automatically also a level-8-, level-9- and level-10-border. -- A uller 17:21, 8 September 2008 (UTC)
No, there are several places where this isn't the case, I think a few were mentioned somewhere in the mailing list some time ago. --Eimai 17:44, 8 September 2008 (UTC)

Instead of coming up with binary representations, using relations to combine all boundary pieces around one administrative area together seems more logical to me. Then you just need to look at the relations a boundary is a part of. --Eimai 17:44, 8 September 2008 (UTC)

Hmmmm, a relation the size of The Russian Federation ...
That's no problem. The solution for this are relations as members of relations. (JOSM can easily handle this procedure) --Cbm 06:41, 9 September 2008 (UTC)

Boundary stone

any idea, how to tag boundary stones? My idea is to tag them boundary:boundary_stone. In my area, there are boundary stones which are still in use, so the historic tag (the search guided me there) doesn't fit well. -- Schusch 08:45, 29 September 2008 (UTC)

  • I like this idea too. I would make it a little broader with boundary=marker, type=stone or type=sign. When people map boundaries sometimes the only clue is a sign such as "Welcome to ...". Tagging the sign helps identify a definite place that is on the boundary. A boundary stone is, perhaps, such a place too and can be historic. Chillly 18:38, 30 September 2008 (UTC)
    • I've been using boundary=marker + ref=* + year=* for a while now. My tagged ones are not historic, although most are old (1843). They are still maintained by official committees of the respective countries. The competing tagging is historic=boundary_stone, but to me that's just a more limiting tagging. It's easy to add historic=yes (and defunct=yes) to boundary=marker, but impossible to denote active use on the historic=* variant. Having more tags to tell the type of marker (marker=stone/wood/metal) sounds logical too, and I notice marker=stone is already used a good number of times. Ldp 23:20, 9 January 2010 (UTC)
      • boundary=marker + marker=stone/sign/beacon/pile/buoy/* + ref=* is including something like 99% of all border markers in the world, and includes most of the relevant information. The node should of course be part of the way boundary segment and that way be part of any boundary relation. It might be necessary to do a ref:norway=* + ref:sweeden=* for boundary markers along that border, etc as the two countries might have different references for the same marker. --Skippern 23:49, 9 January 2010 (UTC)
        • They do not always stand exactly on the official border. We have cases where the border runs down the center of a river and there are 2 identical markers on each shore. In other cases, a marker used to be on a road, but the road has been enhanced/widened, and is just not the same as it was in 1843. In this case, the marker was placed at the side of the road, some meters away from the actual border. So no, they should not be defined as always part of a boundary way, although most will be. Ldp 23:53, 23 June 2010 (UTC)

UK admin_level=2?

Why is it that the UK has admin_level=2 for borders of Wales/England/Scotland? The rest of the world only uses 2 for state borders, so it makes sense if the UK could follow that principle and use admin_level=3 or 4 for the Scotland/Wales/England borders and 2 only for the UK border... --Eimai 14:09, 20 November 2008 (UTC)

Isn't UK (United Kingdoms) a union of countries? If so, than the right way is to tag admin_level=2, if this is not the case, than change it to 3 or something else. --Skippern 13:08, 3 December 2008 (UTC)
England, Scotland and Wales are all separate countries so they should be tagged as such (IMHO) which would be admin_level=2. This does however raise the question of whether the UK as a whole should have its border tagged as admin_level=1 --InsertUser 22:50, 2 September 2012 (BST)

UK counties

The way I've generally been tagging things has been to use 6 for traditional and preserved counties in England, 10 for parishes, and 8 for anything in between (districts, unitary or otherwise). We could do with having a separation between:

  • Regions of England
  • Traditional counties - known as "ceremonial" in England and "preserved" in Wales
  • Unitary areas - districts in England and counties in Wales
  • Non-unitary districts in England
  • Divisions of unitary areas - often related to older districts (e.g. Carmarthenshire has Llanelli, Carmarthen and Dinefwr)
  • Wards
  • Civil parishes (not church parishes, which would fall outside the remit of "administrative" boundaries)

Working upwards from 10 takes us up to 4, which would mean pushing the constituent countries up to 3. I'd be interested to hear some more opinions on this. Chriscf 14:24, 12 January 2009 (UTC)

Traditional county data is available in OSGB or lat/lon format for England. Scotland and Wales are still being worked on. Owain 12:29, 13 August 2009 (UTC)

Ok, After some thought about what the concepts above represent I think I have a reasonable solution. (This is only a suggestion. I think it makes sense given the problems faced). The reason it doesn't quite work with the above model is that you are mixing up administrative areas with areas of representation. Ie Wards are not administrative areas, they are just an area represented by a councilor. Parishes are administrative areas. Subsequently I'd be in favour of representing the two separately:

  • Tag:boundary=administrative to represent administrative bodies area of influence for parishes=10, districts/boroughs=8, counties/unitary authorities=6 etc. (although you might want to demote that to 7 so you could add historic counties at level 6).
  • Tag:boundary=political to represent areas of representation: Wards=8, Electoral divisions=6, National Assembly Areas = 3/4, Constituencies=2, European Election Areas=1.

When I was choosing those numbers, I was attempting to marry up the values with the area of administration: ie District councilors for a ward work in the area covered by admin_level=8. MPs work at the National level (ie the whole UK) and hence are in level 2.

The problem I can see with this is how to represent the admin_level for political boundaries. Should we have admin_level:political or political_area. There are pros/cons for both. Essjayhch 11:11, 15 May 2011 (BST)

Clean up - move to 'Tag:boundary=administrative'

This page need a clean up, the information about admin levels should be taken out and put on a boundary=administrative page together with a admin_level=* page, this page should contain general information about boundaries and have links to different value pages as boundaries make up more than administrative boundaries. --Skippern 18:24, 10 February 2009 (UTC)

boundary=administrative by far the most prevalent usage though (tagwatch). So at the moment Key:admin_level redirects here to Key:boundary. Strange way of organising things? Maybe. But it's a good way of making sure everyone is quite literally "on the same page", so this situation really isn't that bad. If we do as you suggest, and split it out, we'll end up with lots of links (and people) going here and expecting to see the admin level list, for a while at least. Would need to fiddle with lots of links. You can go ahead and do it if you want, might be better actually do a page move, and then re-create this page (preserve more wiki editing history) ...but I can think of more urgent wiki cleanup tasks. -- Harry Wood 18:41, 10 February 2009 (UTC)
Yes, but at the moment this page contains ONLY boundary=administrative (and admin_level=*) and nothing about boundary=national_park and the hardly used? national, political and civil boundaries, and will only be confusing if we add documentation of boundary=maritime or what the outfall of Maritime borders will be. Don't we want people to know about all the commonly/recommended usages of a tag, or is it OK to assume that boundary=administrative is the only usage of boundary=*? --Skippern 18:51, 10 February 2009 (UTC)

It's a fair amount of wiki fiddling to fix this (at least half an hour) but maybe it is worth doing. If you can be arsed, these are the steps:

  • Move 'Key:boundary' to 'Tag:boundary=administrative'
  • Search 'What links here' and change any direct links to 'Key:boundary' to 'Tag:boundary=administrative'
  • Repeat those two steps for FR:Key:boundary and Tr:Key:boundary
  • Having moved everything, change the 'Tag:boundary=administrative' page. Make it use Template:ValueDescription intead of Template:KeyDescription and other tweaks
  • Create the 'Key:boundary' page (replace redirect). This will be fairly short, pointing to the important 'Tag:boundary=administrative' page
  • Repeat those two steps for FR:Key:boundary and Tr:Key:boundary

If you're going to do it. Do all of these steps. Don't leave it half done. Maybe I'll do it myself some time.

-- Harry Wood 12:40, 19 March 2009 (UTC)

politically/nationally disputed boundaries

There are some disputed boundaries in the world, for example the Ilemi Triangle - it would be nice to have those boundaries being rendered differently, maybe as dotted or dashed line... --katpatuka 05:50, 22 February 2009 (UTC)

Please, pay your attention to this discussion. Dinamik 22:43, 26 November 2012 (UTC)

Possible new value: boundary=historic

I've been editing in Ohio today and I came across Boundary Line Rd. I know that this road name refers to the Greenville Treaty Line, which is still mostly visible in roads and treelines (and even some township & county boundaries) today. It would be nice to be able to draw this historic boundary in OpenStreetMap. But how should that be tagged? This wiki article doesn't really discuss anything other than current administrative boundaries. So I suggest boundary=historic for such things. I'm not sure how the admin_level would be applicable to that. General-purpose renderers can ignore it. If anyone else likes this idea, then I'd say a proper proposal page is in order. (I'm not quite brave enough to write one myself yet.) Vid the Kid 02:46, 31 July 2009 (UTC)

Historical data does not belong into the "today" database. This can go into a separate database for overlays, see Proposed_features/4th_Dimension/Archive. --Lulu-Ann 10:06, 31 July 2009 (UTC)
I wasn't referring to former boundaries that are no longer relevant. What I meant was, boundaries that are of historic significance, and still apparent today. The only examples I can think of, however, are boundaries between Public Lands surveys in the US, which would include the already-mentioned Greenville Treaty Line. Still, someone else has asked about tagging PLSS section lines. They're a more specific division than the boundaries of the Public Lands surveys themselves, and they even appear on many official current-day county maps. So, clearly, they're not just old, forgotten lines from the past. They're current-day boundaries, and while they have little to no impact on our daily lives, they're still here and they have deep historic roots. Vid the Kid 01:29, 10 August 2009 (UTC)
Historical boundaries have an important usage today, even if they are no longer administrative borders. For example they may be still used by the successoral administrations to respect some old international treaty, with specific legal rules applicable to only parts of their territory. But also for drawing historical maps, or for cultural reasons (e.g. the old provinces in the kingdom of France, that are bound to nobility titles, even if these titles confer no privilege today, are still wellknown and often related to natural regions).
There are also other kinds of administrative subdivisions or groupings that do not respect the normal administrative hierarchy (an example in France is the groupments of communes in EPCI's, that sometime span communes of several departments or even several regions; same thing about France's newer definition of "Pays", under Loy Voynet, created as a tool of cooperation between various kinds of local administrations, sharing some common project and federated in an non-profit association or groupment of economic interest, sometimes half-public with participations from the private sector, in order to manage some project, notably for environment and cultural/touristic promotion).
Another example concerns transnational groupments of local administrations, that want to share some project (for example urbanism, utility provision, ways of communication, transport systems, common economic expansion around an airport).
Another example is the existance of multinational entities (most often, international airports or harbours, with facilities shared between several countries, and special fiscal regimes). Another example is the existance of tax-free areas, or areas that use another currency than the currency of the country to which they belong (some examples near the borders of Switzerland and Germany): they are not necessarily exclaves, but instead have to live with the fact that another country than their own will better supply some facilities (such as security and emergency services, accesses by road in mountainous areas during winter when the national road has to be closed or is almost unusable for extended time because of natural events or risks...).
A last example is the mapping of linguistic areas (this can be very complex in Belgium or in some cantons of Switzerland, with a linguistic line splitting different levels of administrative areas...)
In many of those cases, you'll find references to very old treaties and local arrangements across administrations or across today's countries, and in national or state laws (consider the case of aboriginal areas in US, Canada, Brazil, and even in France's oversea collectivities like Wallis-and-Futuna, where you'll find kingdoms, or in New-Caledonia due to the protection of civil and traditional status by the Constitution, which also recognizes distinct hierarchies of collectivies and goverments, some of them managing their own exclusive territory, or in the oversea departments of Guyane and Mayotte).
The various status of territories in each country often contain many exceptions (this is true in almost all countries except possibly a few microstates). Transregional boundaries are a necessity and do not fit very well with the simple hierarchic subdivision assumed by "admin_level=*"... For this reason, the numeric "admin_level" will only be useful for the major subdivisions, but even in this case, you will need sometimes to subqualify this level by adding another tag to indicate which kind of subdivision is implied (when this is not the main hierarchic subdivision).
And finally, even in this case, two separate hierarchies may coexist (we have the example of electoral subdivisions, there are other similar cases), which still should be easily parsable to find subareas. I strongly suggests that, instead of locaing subareas only by the next higher "admin_level", we still allow a bounded area to list explicitly its list of subareas in directly in its defining relation.
So I strongly support the inclusion of child boundaries with the "role:subarea" tag (but NO requirement that the subareas are perfect hierarchical subdivisions: it should be allowed that a subareas belongs to two parents of the same administrative type/level). However JOSM still complains that we cannot reference mutipolygons of "type:boundary" with "role:subarea" as members a multipolygon of type "boundary" (it currentkly only accepts silently as members: ways with "role:inner/outer"; and a single node with "role:admin_centre", excluding larger definitions of the admin_center as a relation containing multiple nodes or ways; it does not correctly accepts boundary segments with a "role:inner/outer" and does not easily allows creating subgroups of separate boundaries into collections created also with type "boundary" but without meaning that these groups are subdivisions, even though these collections would easily idenify the areas of the same/assimilated type : e.g. collections for regions or for COM's or for sui-generi countries in France; collections of bodies making the state of Netherlands, within the kingdom of Netherlands, which may be either provinces or special communes, and the collection of the 4 states making this kingdom).
The Karlsruhe schema still lacks such better modeling. Verdy p 22:25, 11 August 2011 (BST)


While browsing some boundaries in the UK OSM data, I wanted to ask about the use of descriptive text and notes in the name tag.

I would sort of expect this thing to have "name=West Midlands" because that's it's name, rather than "name=West Midlands (county)" which is currently in the OSM data. The fact that it's a county is surely encoded into the admin_level?

Ojw 12:14, 1 August 2009 (UTC)

Maybe that's to disambiguate the county from a formal or informal regional designation that may exist with the same name? I'm sure California has both a "Los Angeles" boundary at admin_level=8 and a "Los Angeles County" boundary at admin_level=6. Or at least it should, if properly mapped. Vid the Kid 01:34, 10 August 2009 (UTC)

Boundary rendering

Here's a test of the boundary rendering in Mapnik and Osmarender. Shown are boundary=admin with admin_level=1 through 11, and the other mentioned values for boundary=*.

Mapnik appears to render the boundaries the same at various zoom levels, while Osmarender changes the rendering and visibility. Michael Z. 2011-03-01 18:09 z

Boundary rendering.png

Boundary shared with other tags

I have seen a few places where a boundary=* tag is on the same way as highway=* tag (or railway=*). This seems to confuse quite a number of tools and renderers. (e.g. Potlatch 2 shows these ways as a boundary, not a highway or railway.)

Is this a valid combination, or would it be better to indicate the boundary part using a relation, including the various ways that make it up? --Ebenezer 21:16, 1 October 2011 (BST)

It is perfectly valid. Those renderers should not be confused and will be corrected because this is really common to share ways, as this generates maps that are easier to edit and update.
So file your bugs to the maintainers of these renderers. — Verdy_p (talk) 07:13, 12 June 2012 (BST)

Property (land parcel) boundaries

How should one code the boundary of an individual land-parcel/property? ie the boundary around a residential dwelling including garden etc? Any ideas? Seems as though boundary is the logical tool for the job. PeterIto 21:38, 26 December 2011 (UTC)

May I suggest boundary=other and border_type=property? --Skippern 23:42, 3 September 2012 (BST)

Role for populated settlements

With developed administrative boundaries and with active use of subarea role its very easy to build tree of administrative territories without sophisticated geospatial queries. So how about to make next step and explicitly mark settlements in boundary relation of the lowest level? If we mark city/town/village/hamlet/etc. boundary (closed way or multipolygon marked with place tag) in relation with role "settlement" (for example) it will allow us to build address tree up to the settlement level with ease. Yes, this will be redundant, but it will allow us to gradually move from use of tags addr:country/addr:region/addr:district/addr:subdistrict/is_in and probably will be less error-prone and much more machine-friendly.

boundary=yes for muti-type boundaries?

Some ways can be just a boundary (not also a highway, etc.), and they can be a boundary for multiple places, some administrative and some not. In such cases, perhaps boundary=yes can be used?