[talk-au] PSMA Administrative Boundaries

[Andrew Harvey]

Good to see there is a lot of interest in this. Local knowledge is going to
be key to ensuring success if we undertake work to bring this into OSM.

On Fri, 31 Aug 2018 at 16:45, Ewen Hill <ewen.hill at gmail.com> wrote:

> I would like to aski if it is possible to
> 1. add the ability to have, possibly at [*5*], the Aboriginal nations
> e.g.
> https://commons.wikimedia.org/wiki/File:Map_Victoria_Aboriginal_tribes_(colourmap).jpg
> and
>


> 2. swap  *6 Local Government Area (eg Shire/Council) *and *7 District or
> Region (e.g Perthshire, Fitzroy, Canning, Greater Sydney, Greater
> Melbourne, etc.) *as normally these areas are large than a singular LGA.
>

Despite what is on the OSM wiki, admin_level=7 is only really being used in
ACT at the moment, Greater Sydney isn't really an admin boundary. I can
only comment for NSW, but we have regions like Hunter, Central Coast,
Illawarra, South Coast, New England, Eastern Suburbs
https://en.wikipedia.org/wiki/Regions_of_New_South_Wales but I've been
tagging these using place=district.

On Fri, 31 Aug 2018 at 17:21, Andrew Davidson <theswavu at gmail.com> wrote:

> On 31/8/18 16:23, Andrew Harvey wrote:
> > (not sure why the split across 9 and 10...? but it looks like no data
> for 9
> > and all suburb/localities are marked as 10, a separate discussion but
> > perhaps they should be 9 and smaller neighbourhoods be in 10?)
>
> Level 9 appears to have been intended for non-ABS derived suburbs and
> level 10 for ABS. What's happened is 10 has been used for suburbs.
>

Given ABS has pretty much nothing to do with how suburbs/localities are
defined, not sure that makes sense going forward.


> Do we have any data for bounded localities smaller than suburbs? I don't
> know of any jurisdiction that defines sub-suburb areas, so I don't see
> the point of moving away from 10.
>

Maybe as informal places which we can use place
https://wiki.openstreetmap.org/wiki/Key%3Aplace tags for, but I'm not aware
of any which justify using boundary=administrative.


> I know that some people don't like having the admin boundaries in OSM. I
> find that having the suburb/locality boundaries is useful (downloading
> data by area and geo-referencing in Nominatim).


I feel this too, it's not data that we are really creating or adding value
but I can see having it in OSM can make it easier to use other OSM data as
well as helping to contributing to a global dataset with consistent schema.


> The LGA boundaries less
> so as most people don't really think very much about their local council
> and it's odd to see it appear in an address for example. But that said
> if you do the suburbs you might as well do the LGAs.
>

I hear you, but they are surveyable (usually street signs are branded with
the LGA and signs telling you when you enter the more rural shires) and it
is handy to be able to use OSM to easily check the LGA boundaries and have
all the attached metadata (website, phone number etc).

On Fri, 31 Aug 2018 at 17:53, Warin <61sundowner at gmail.com> wrote:

> My 'problems' with the admin boundaries are;
>  where they use another way that is also a road/river. And then some
> mapper comes along and improves the road/river .. but buggers up the admin
> boundary.
> then I come along and 'fix it' using the LPI Base map .. and put that
> source on the way.
>

I've seen this problem too, especially for new mappers who want to change a
road, add turn restrictions etc. it's too easy to break relations like the
admin boundary. I think this is just the cost of having this data in OSM
that occasionally it will break and need fixing. Hopefully editors will
improve to reduce the risk but might be something we have to accept.


>
> 2 things on my wish list ;
> that admin boundaries don't use other 'close enough' ways already in OSM.
> This would reduce the number of times that the relationships are broken.
>
that admin boundaries have a source tag on there ways to say what the
> source is. This means when I come along and fix a broken one my source
> statement will not be confusing when looking at  the relationship as that
> relationship will have no source tag.
>

No problem with that.

Generally I wouldn't treat these boundaries as coordinates carved in stone
that absolutely must be fixed in OSM. I think it's fine if there they are
out my a few meters if it makes sense to line them up with a road or river,
especially if those boundaries are defined by the road or river.

On Fri, 31 Aug 2018 at 22:17, cleary <osm at 97k.com> wrote:

> - I agree that it is inappropriate to map administrative boundaries on the
> same way as natural phenomena such as rivers. The boundaries are usually
> where the river flowed sometime in the past when an early survey was done.
> The boundaries do not change but the rivers often do. I think it best
> always to map administrative boundaries separate from natural features.
>

I'm on the fence about this, I think it depends how they are defined, if
the suburb boundary is defined to follow a road, then the admin boundary
should use that road as part of the relation, same for rivers.


> - It would be possible to change suburbs from level 10 to level 9 if
> others find that useful but I think that sub-suburb localities such as
> neighbourhoods are not "administrative" nor (to my knowledge) are they
> defined by a government authority so I am unsure what level 10 could be
> used for, if suburbs are made level 9.
>

I agree, these lower levels are likely not administrative, so could simply
use the place tag.

On Fri, 31 Aug 2018 at 22:21, Lee Mason <lee.mason at outlook.com.au> wrote:

> The state (and probably more broadly) already has a comprehensive naming
> of localities and suburbs from surveys, but mostly from the GeoScience
> Australia place names dataset. I would think that when integrating PSMA
> boundaries, it would be important to preserve these place nodes which more
> accurately pinpoint the locations of smaller communities (even if it is
> only a cluster of a few homes), which would not necessarily be the centroid
> of the relation.
>

Absolutely, as you point out that's where the place tags come into play
https://wiki.openstreetmap.org/wiki/Key%3Aplace, the admin boundaries can
exist completely independently to places. I'd only mix them if they really
are the same thing. Generally this might be true for place=suburb, but
probably not for place=town,village,hamlet etc.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-au/attachments/20180831/4978c75c/attachment-0001.html>