Indian Reservation Sublevel Tagging


How would one tag sublevels of Indian Reservations in the US? Currently (at least in SD) they’re tagged with “boundary:administrative” “admin_level:7” like townships and that isn’t entirely correct in my opinion… but I can’t think of an alternative solution as “boundary:aboriginal_lands” doesn’t have sublevels as far as I can tell. This isn’t necessarily a problem in places where the tribe is essentially the only government in the area (Pine Ridge, Rosebud) but can be a problem where the reservation overlaps with county/state boundaries (Lake Traverse).

See Relation: ‪Old Agency District‬ (‪13839748‬) | OpenStreetMap and Relation: ‪Crow Creek District‬ (‪13836762‬) | OpenStreetMap for examples.

Really what would make sense to me is to add sublevels to the aboriginal_lands tag, since at least in the US the tribal boundaries aren’t operating on the same scale as the non-tribal boundaries. That might not be the case elsewhere though, so I’m opening up the discussion gates.

Something like this would make sense to me too. Many tribes have their own administrative hierarchies that are separate from the federal/state/local hierarchies. In many places, the relationship between the two is complex and sometimes disputed.

It doesn’t seem that trying to fit the tribal administrative levels into the hierarchy for US government administration would do justice to either side. So, it would be better if we could represent the tribal hierarchy the way the tribe has established it, independent of the US government hierarchy.

It seems that Canada and possibly Australia and New Zealand might have similar issues, although I don’t know much about the situations there.

1 Like

Thanks for bringing up this topic. I’ve been wondering if we need to change anything about how we’re representing the Navajo Nation, which is divided into agencies and further subdivided into chapters. Agencies and chapters are both geographic and governmental units. Confounding matters, the Navajo Nation has designated the CDP boundaries to serve as chapter boundaries. The agencies are currently tagged as boundary=administrative admin_level=7, while the chapters are represented only by place nodes; some CDPs have boundary=census relations as well. It seems counterintuitive that, of the Navajo Nation’s three tiers of administrative units, only the middle one would be tagged as “administrative” in OSM.

I cannot stress enough the importance of someone who is actually an enrolled member of the Navajo (or the tribe in question) to add perspective to this discussion. It can be contentious, it truly is sensitive, and over the last year or three, both in Canada and the USA, admin_level tagging in these areas (reservations and First Nations) has emerged in OSM.

Especially given the discussion about Maine and New England townships going on in a sibling topic right now, it would very likely be unwise to make assumptions about “a new way” or “the best way” to do this. Please see United States admin level - OpenStreetMap Wiki where an important quote from Wikipedia is included:

"tribal sovereignty is a form of parallel sovereignty within the U.S. constitutional framework, constrained by but not subordinate to other sovereign entities ", where a map of the contiguous US (lower 48 states) with reservation lands excluded displays.

OSM can (and likely should, if they truly exist) tag “sub levels” on these, but not without direct consultation with “the people of the land.” This is a difficult topic; please proceed with sensitivity, a great deal of open-mindedness and please do not make hasty decisions.

Edit: The linked wiki states:

“These relationships are too complex to logically shoehorn into a hierarchical scheme like admin_level=*.”

I did not originate that wiki sentence, it was Paul Norman who did (a Canadian “architect,” for lack of a better word, of deep structures and tagging conventions in OSM going back to our very early days).


This would be most helpful, but I doubt we have those people in the community. If you want to wait with mapping or (re)tagging these places, we can wait for a very long time. I suggest using other sources and OSM conventions to tag these places and to document it properly, so if/when someone comes up with a better idea we can update the tagging scheme consistently.

In the Netherlands, the water boards are separate government institutions, but since they operate outside the hierarchical chain that makes up the other governmental bodies, they don’t get an admin_level tag. Based on that logic, I would also not expect to see admin_level tags on Indian reservations. For the subdivisions of reservations I suggest not using admin_level, because using admin_level here would lead to confusion.

I agree, and yet, these tags are being added to these areas. I find this difficult, yet I note to other contributors that this is underway.

1 Like

Agreed. I think we’re all eager to learn what would be most appropriate. What I really don’t want a repeat of is the unilateral edits that one mapper made to all the reservations, sometimes conflating ethnicities and governments where there might need to be more nuance.

I’ll note that the discussion and vote about boundary=aboriginal_lands involved members of more than one tribe who took opposite stances on the tag. It goes to show that indigenous peoples aren’t a monolith. There’s also the complicated situation in Oklahoma.


Exactly. I’m just after a solution.

So that map isn’t entirely accurate, let me demonstrate with an example in the area that prompted this discussion. If you’re standing at the corner of 455th Ave and 157th St and are a member of the Sisseton Wahpeton Oyate, you are in the Big Course District of the Lake Traverse Reservation and its laws and jurisdiction apply. If you stand at that intersection and are a pasty white guy like me, you are in Germantown Township, Codington County, SD, USA and its laws and jurisdictions apply. Like I said previously, I think the simplest approach would be to add levels to the boundary=aboriginal_lands tag so that where there are tribal subdivisions they can be mapped appropriately, but the simplest approach might not be the best in this case.

In the interests of keeping things positive (yet real), please allow me to convey that for over a decade, I’ve been quite keen to properly denote things like this, yet have been unable to do so (so far). I keep putting positive spin on things where, how and as I can (such as wiki entries along the lines of “there have been many tagging strategies here…many of them controversial or ineffective…” because I both want to convey that “work to do this has taken place” while at the same time, only minimal / minor resolutions seem to have resulted.

Others do this, too. Others discuss this, too. We do so again, here and now. I don’t want to be pessimistic, but this truly has been one of those (relatively rare, to be fair) issues (in OSM) where no matter what is tried, how widely it is discussed, whatever potential new solutions are posited, somebody finds reason to disagree.

An upside to this is how I’ve “spun” it in the wiki, that is to say that an approach of “let the people of the land determine these taggings” currently reigns. A downside (which I’ve never uttered until now) is that it is essentially a “free for all” which likely dilutes the meaning of admin_level=* as we define it. (As I agree with @pnorman that “shoehorning this isn’t going to work,” yet it continues, again, as a mildly-alarming “free for all,” yet “that is what it is.”)

Minh says “we’re all eager to learn what would be most appropriate,” a delightfully diplomatic way of putting it. Yet, even as I deeply wish to see forward progress on this, how this emerges and succeeds remains elusive in my mind. My best approach is to be open and listening, even as I feel somewhat deficient and frustrated that doesn’t seem enough. So while I observe more (and participate less) in this particular endeavor, I’ll remind that this has a lengthy and difficult history, with few apparent solutions. And of course, I welcome ideas to further this ahead, though, please know that what might be “obvious” or “easy” has often been found to be “wholly wrong” and “overly simplistic.”

Courage and high spirits, everyone: let’s keep chipping away at this.

I’m not aware of multi-level ones in Canada, but there might be. I’m just not familiar enough.

That proposal is a bit odd in that it was proposed in 2010 and voted in 2018. My recollection is that discussions about using the tag took place much earlier, and much of the history is poorly documented. I reviewed the chat logs (line 54400) and found discussion that occurred in October 2013 and referenced earlier talk-us@ posts, so my guess is any discussion happened in early 2012 when usage started. I was unable to find the list posts. There was additional discussion in July 2014 (line 114534).

One community member was of the view that reserves should be cut out of the admin_level=4 states, and if you were in a reserve you weren’t in a state. Myself and others were of the view that reservation boundaries were orthogonal to any admin boundaries as they can cover multiple states, multiple countries, or lie entire within a city. I am still of that view.

A different type of boundary=* was considered best, and we were using boundary=aboriginal_lands at that time, and that was considered better than trying to shoe-horn it into boundary=administrative and admin_level.

At the time, there were not many mappers who were mapping these, and there was some questioning if they should even be mapped. But then again, some prominent US community members were questioning if any admin boundaries should be mapped either.

Going farther back in history, in 2010 the rendering was mis-named Mapnik and labeled Nature Reserves with little NRs and Indian Reserves with IR. This was both racist and bad cartography. [1]

The original aboriginal_lands proposal was made by a user who shortly after left OSM. I believe use started after he left. Around that time the community settled on aboriginal_lands as the best name under British English, given its use in Australia and Canada which are closer to British English than American. Indian is obviously out, as not being a preferred term anywhere, and being completely incorrect in Australia. Maybe in retrospect indigenous should have been used, but there are issues with any term. In any case, this is far from the first tag that isn’t named ideally.

What does this mean for sublevels today, over a decade later? We never thought about sub-levels at the time, although we did decide to include all indigenous reserves in the same tag, even though some are much bigger and operate differently.

I would avoid doing boundary=aboriginal_lands and admin_level. Come up with a more appropriate tag name, as there’s no usage right now.

  1. Probably not intentionally, I don’t think anyone had given much thought to that bit of the rendering. ↩︎


Thanks for looking into the history around these tags.

Yes, I remember this issue contributing to the sense of urgency for a new tag. I’m glad that at least we were able to come up with better rendering pretty decisively.

To clarify, would you suggest that we avoid using either of these tags for reservations and their subdivisions, or that we avoid using both of these tags in combination?

Also additional clarification, this is the state of the map as it is currently (both tags in combination), and I’m not advocating for its continuance (just wanted to make sure that was out there).

Also additional additional clarification, I’m not sure there are examples of boundary=aboriginal_lands and admin_level on an entire reservation. But what we ARE seeing a recent increase in are taggings of sub-regions within a reservation tagged admin_level=7.

I agree with Paul that OSM likely doesn’t want to do this. However I have documented in our wiki that it is happening.

Don’t use an admin_level key at the same time as a boundary=aboriginal_lands. It’s not an admin boundary, so don’t use a key with admin in it.

I don not believe any software does anything with this combination, or if they do, they’re likely incorrectly thinking it is an admin boundary.

1 Like

I think this is what we’re struggling with. In the real world, reservations and their subdivisions are all administrative boundaries. They would meet any test we can come up with when assessing other kinds of boundaries, such as New England townships, to pick a current example. (Maine’s reservations are considered township equivalents for demographic purposes.)

Reservation boundaries are marked by signs just like state and county boundaries. During the pandemic, many reservations set up checkpoints at these signs. Depending on the context, someone might describe their location relative to a reservation or a county, but this is not at all like a single-purpose fire-rescue district or water board. In fact, on any of the larger reservations, I suspect the counties matter about as much as a New England county.

Reservation boundaries can cross state and county lines unimpeded. I bet this causes renderer and geocoder developers a lot of grief. But this can happen with other administrative boundaries anyways. Probably this issue isn’t as relevant as the fact that there’s a parallel system of administration.

If we continue to represent the reservation boundaries as boundary=aboriginal_lands rather than administrative boundaries, then there are still several outstanding issues:

  • Mappers have periodically stressed a need for a tag to distinguish reservations from other kinds of indigenous lands, namely off-reservation trust lands and Oklahoma’s tribal jurisdictional areas (the ones affected by McGirt). Can border_type=* serve this purpose?
  • It seems logical that the reservations’ subdivision boundaries would be tagged more like the reservation boundaries than the non-native county boundaries, given the reporting structure. Can boundary=aboriginal_lands nest within other boundary=aboriginal_lands?
  • If admin_level=* is reserved for boundary=administrative, should there be a parallel aboriginal_lands:admin_level=* key? Should the numbering restart or correspond to the non-native boundaries’ admin_level=* values? Or should we rely on the status quo of adding each agency boundary relation as a subarea member of a reservation boundary relation?
  • Should reservation capitals and agency seats be tagged with capital=*, and if so, what should the numeric values correspond to? Similarly, should tribal national parks be tagged as boundary=national_park?
  • Since the CDPs within the Navajo Nation double as chapter boundaries, should they be retagged from boundary=census to boundary=aboriginal_lands aboriginal_lands:admin_level=8? Or should we map redundant boundaries for the chapters?

I would try to avoid numeric values. For admin levels there are reasonable numeric standards for the higher level boundaries, but I haven’t seen that for reserves.

aboriginal_lands:admin_level seems like a confusing beast of a tag, so maybe we can do better.

I would avoid that. We’re a spatial database, telling us something is a subarea of another thing is duplicating information.

Unquestionably yes, if they’re the same type of park. national_park is describing a certain type of park, which most US national parks are, but so are state parks, Canadian provincial parks, and I think even some county parks. It wouldn’t surprise me if some tribal parks are the same - or might be nature_reserve.

The CDP should be gotten rid of and only the nation mapped, imo. If for some reason we wanted to map census boundaries for their own sake, it would need another relation with all the same members.



The latter. (While the former is an important consideration, it isn’t a solution or proof). There are renderers which display admin_level at checkbox-per-value levels, one is - Tuiles pour contrôle de données / Tile service with dataa checks where the red areas shown are admin_level=7. This often means an MCD, roughly “township” in USA-ese (very rough) in the Lower 48. These are found as an agreed-upon / consensus-reached in about 1/3 of US states. We see here (now) the red areas in the Four Corners / Arizona and Dakota regions relatively newer (last year or three) sprinklings of “these.”

If these data are from the Census Bureau, then for all good reasons, let’s tag them appropriately with boundary=census. If they are not, (in my opinion, I believe matching Paul’s) they should not have anything to do with how we currently use admin_level=*. The end (of that topic). Paul is right, these would only get confused / tangled up with admin_level=* (tagging and values) as we now use them (oops, already happened) and we wish to set things right, as these have simply evolved as something different. That works for me, as it is reality and so we must.

Next up (imo) would be to develop a new syntax (key:value structure) for denoting these as what they are. I say “similar semantics go on here that go on in admin_level” but the numeric value scheme the latter uses simply isn’t the same thing. Imagine descriptions (at a first draft) as like (Wikipedia’s habit of) first-level divisions and second-level divisions. We are starting at the level of a nation and talking about subdivisions, so let’s invent (develop) a way to do that. It’s not a terrible distance ahead to do so, it is a small bit of development. There might be British English (OSM tradition) placeholders or emerging-as-de-facto values like local_council or divisional_government, I’m spitballing.

Getting this untangled from admin_level=* seems quite correct. I cannot strongly recommend enough a lot of latitude and flexibility continuing to inspire this, as there are no doubt hundreds or thousands of these entities which will emerge in their own unique ways, having both structure and not having structure (can be a tough one, look at us struggle so far…but we are improving). These things do evolve organically, we really have no choice but to help that forward ahead. We can do this. In fact, I think it actually has the potential to bring world peace. Well, I might exaggerate, but I’m standing as tall as I can.

Not so fast. Unless I’m misreading their GIS website, the Navajo Nation has adopted the Census Bureau’s CDP boundaries as their own chapter boundaries (the third-order boundaries in the Navajo system). The data may come from the Census Bureau, but that’s the source of truth for these tribal administrative boundaries.

I don’t have a photo of a chapter entrance sign to share, but the chapters are used for geographical reference. For example, Lake Powell Navajo Tribal Park is described as being “located within the Le Chee Chapter of the Navajo Nation.”

@pnorman wants to avoid numeric levels. There may be some wisdom in this. It took us years to come to the realization that there are subdivisions beneath reservations, so what’s to say we aren’t overlooking some differences in the number of levels in each reservation that would frustrate any attempt at national consistency? On the other hand, the chapter boundaries are, practically speaking, spatially equivalent to local municipal boundaries, so maybe something should reflect that. And without some explicit scale, it’s going to be up to renderers to figure out how to make each increasingly local subdivision consistently show up only at higher zoom levels.

1 Like