[Imports] Oahu, Hawaii CDP boundaries

stevea steveaOSM at softworkers.com
Sat Sep 7 17:56:17 UTC 2019


Max says:
> It may make sense to tag the boundaries as place=neighborhood or something
> like that.

Please, gentlemen and gentle readership of Imports List, let us tag correctly.  The place=neighbourhood tag (please note the "u" as this tag value uses British English) is used for places WITHIN a larger place (like a quarter or suburb which itself is within a city) or WITHIN a smaller place (like a town or a village).  It may very well be that partly because Honolulu-Oahu as a consolidated city-county (CCC) simply "contains and surrounds" anything within them that it might be tagged place=neighbourhood, so that would be a correct tag.  But if it exists, I would want any more-structured, finely-granulated hierarchy to be preserved.  For example, if it were proper to characterize a sub-area of Honolulu-Oahu as a town (maybe, maybe not), a neighborhood that is truly within that town should "aggregate with" the town.  But if it is a neighborhood which doesn't associate with a smaller (than city or county) entity (like a town or village), but rather Honolulu or Oahu itself, then THAT "hierarchical association" should be denoted properly in the tagging.  This might cause, even force a choice to select place=suburb or place=quarter as opposed to place=neighbourhood.  Please see the relevant wiki, as this tagging can have subtleties that are difficult to describe, yet when they are understood, they can logically map quite well onto a particular urban conurbation with a characteristic "Aha!" moment as you realize this.  I expect Honolulu-Oahu, with about a million people, has some of this complexity, and I'd like to see it correctly captured as OSM better tags these.  That is why we have different tags of suburb, quarter and neighborhood.

Max says:
> It would depend on whether they have a firm enough meaning to locals or
> whether they are statistical conveniences.

Very much "Yes" to the first part, although if by "statistical" you mean CDPs (they ARE statistical), then please use boundary=census (if that's what the boundary is as you enter those data).  I agree it's PART how "firm" a sense of place is to locals, but it's PRIMARILY about selecting the most correct value of place=*.  Maybe it's suburb, maybe it's quarter, maybe it's neighborhood, but these have subtle differences.  Getting them right so they convey whatever sense of hierarchy there is in the real world is important.

Max continues:
> (obviously the places are meaningful to locals, I'm talking about the exact
> boundaries)

Again, should you have and enter exact CDP boundaries (a somewhat contradictory statement, as they are statistical, and can't really ever be "exact" as they are always slightly changing) it's OK to enter them tagged boundary=census, but never with an admin_level tag.  I'll say it one more time, a very good place to start (and maybe best in the long-term, almost certainly best in the short-term) is a NODE tagged place=*.  This should be either centered in the spheroid it roughly defines, or placed at or very near a "center" which, for example, might be the commercial area of a village (however small it may be).

While Brian says:
> Being an extinct volcano, the boundaries are usually defined by rather
> dramatic topographical features.  But yet there are some CDPs (Kaneohe
> Station) that are as you say statistical conveniences used locally.  That's
> why I'm leaning towards using the postal service's boundaries as level 8
> admin boundaries and CDPs as neighborhoods where they match reality on the
> ground.

Please don't use postal service boundaries in OSM unless they are explicitly tagged boundary=postal_code.  This is done in Germany in Belgium but not the USA (or if it is, it shouldn't be, in my strong opinion).  In this instance (in the USA) I believe it is better to use a node with a place=* tag and its proper value.

> For eample, if you look at the leeward (west) coast of Oahu, the entirecoast north of Ko'olina uses "WAIANAE, HI" for its postal address, and
> there are multiple CDPs (e.g. Nanakuli, Ma'ili, Makaha) that are
> essentially neighborhoods, but with boundaries well defined topographically
> by parallel rivers and ridges running from the mountains down to the
> coast.  But then there are oddballs like you have a "Makaha" CDP and a
> "Makaha Valley" CDP just up-valley that in real usage, is one neighborhood
> just called "Makaha"

OSM in the USA doesn't really map postal addresses, as noted above, although I agree that sometimes the name of a post office can offer a sense of place-name to an area.  In that case, please use a node with a place=* tag and its proper value.  CDPs must be tagged boundary=census, not boundary=administrative and should not have an admin_level tag with any value whatsoever.  It is OK to enter these, but in my opinion, it would be better to supersede these (you COULD enter both, but that's slightly confusing to many) with nodes with place=* tags and proper values.  If the proper value for the areas you are talking about here really meet OSM's definition of place=neighbourhood, then please use that tag.  Be respectful of what higher-in-the-hierarchy entity (town, city, county...) they subordinate to as you do so, please.  And if they don't subordinate like this, then place=neighbourhood isn't really the correct tag.  Maybe place=village or place=hamlet is a better tag on that node.

Brian continues:
> Oh, and the whole shebang is contained within Honolulu County which has a
> county mayor and is administered by the government entity "City and County
> of Honolulu".
> 
> Hence I am thinking, for example:
> Honolulu County = admin boundary level 6
> Waianae (all of zip code 96792) = admin boundary 8
> Makaha+Makaha Valley CDPs = place=neighborhood boundary, perhaps with or
> without a level 10 admin boundary.

Yes, to Honolulu County and admin_level=6; this already exists in OSM as the aforementioned relation/3861844 (properly, in my opinion, with a node tagged admin_centre and a node tagged label).  But, its coterminous Honolulu relation does need to be entered (as Honolulu is a CCC), and the existing Honolulu (messy, incorrect, old) polygon (tagged boundary=census) needs to be removed, or at a minimum must have its admin_level=8 tagged removed, as this does not belong on a boundary=census polygon.  This is also a weird multipolygon as it only contains one member, if it must persist in OSM, de-multipolygonize it by making it a properly tagged polygon, please.

No, to Waianea as admin_level=8, as that clashes with Honolulu, which should be entered as a coterminous relation to 3861844, but tagged admin_level=8 and "City of Honolulu" (distinct from County of Oahu) in its name=* tag(s).  Waianea can be entered (in order of preference) either as a node tagged place=[town, village, hamlet, isolated_dwelling], a node or (multi)polygon tagged place=[suburb, quarter, neighborhood], though please have it clear in your mind — or even a note=* tag — what it subordinates to, or as a boundary=census (multi)polygon containing NO admin_level tag.  ONLY if there is a formal political structure, like an elected neighborhood council, should boundary=administrative be used on something like Waianea (with a corresponding tag of admin_level=10, or 9, if there are two of these, one at a higher level of the hierarchy).

It seems complicated.  Yet I know and feel the effectiveness at having tagged with these conventions for years, largely because I have read and understood ALL of our place=* wikis.  (There are quite a few of them).  Please absorb this knowledge and let it guide your tagging in Oahu.  Our beautiful map will be better for it, more comprehensively conveying to all who follow its guidelines, how particular Hawaiian places are named, properly.

SteveA


More information about the Imports mailing list