[Talk-us] understanding administrative boundary relations

Fri Sep 4 01:56:40 UTC 2015

On Thu, 3 Sep 2015 18:51:50 -0500
Toby Murray <toby.murray at gmail.com> wrote:

> The area you linked to has no boundary data in OSM. You don't state
> this in your email but by pulling up the TIGER shapefile, it looks
> like you are wanting school district boundaries? I see one runs
> through the area you linked to, in the shapefile. School districts
> were not imported into OSM. The only things that were imported from
> TIGER is roads, state, county and city boundaries.

Interesting. There are lots of linear features that are not roads or
the boundaries mentioned. The water feature I described is one.

And, yes, I am looking at school district boundaries right now. But
really, the question is about any other set of relations that are
supposed to tile the map. For example, I knew about Sunnyvale's
relation, which you mention below, but there are over 350 cities in
California and there do not seem to be that many relations, or I cannot
find them (despite some banging of my head against OverPass), or both.

I was hoping that city boundaries would follow conventions like
"admin_level"="8", "place"="city" and so on. More fool I. Most of these
"rules" definitely seem to be honored in the breach.

> Administrative boundaries in OSM have always been a tricky subject.
> OSM thrives on information that can be verified by someone standing on
> the ground, looking around and seeing something that can be put into
> the map. Administrative boundaries are (usually) not that way. They
> are imaginary lines drawn on the map. Sometimes they follow physical
> features but often they don't. So the only source to verify or update
> them is to go back to the imaginary line drawer and ask for an update.

I had not heard this perspective. I am not sure what to say, other than
that though it is a "tricky" subject, it is not really something that
can be ignored. Too many things in real life depend on these
boundaries.

> Because of this, I think boundaries in OSM tend to deteriorate in
> quality quicker than other features. Sometimes people modify a way
> that is part of a boundary relation and don't realize that they are
> affecting the boundary. I have done a lot of work fixing up boundaries
> (mostly county) across the country and there are definitely a million
> ways to break them.

It has occurred to me that there will probably need to be a "boundary
watcher" tool, which can let an interested group know about it when a
boundary gets broken in some way. And I have started playing with the
python libraries for accessing OSM data with this in mind.

> 
> If you want an example of an admin boundary in OSM, here is the
> Sunnyvale city boundary relation:
> https://www.openstreetmap.org/relation/112145
> 

Yep. Knew about that one. Here is the table I am keeping my "meta-data"
list of relations that I am interested in:

mysql> select * from osm_relations;
+----+----------------------------------------------+--------------------------------------+-----------------+
| pk | url                                          | name                                 | place           |
+----+----------------------------------------------+--------------------------------------+-----------------+
|  1 | http://www.openstreetmap.org/relation/112145 | Sunnyvale, CA, USA                   | city            |
|  2 | http://www.openstreetmap.org/relation/148838 | United States of America             | country         |
|  3 | NULL                                         | Sunnyvale Elementary School District | school district |
+----+----------------------------------------------+--------------------------------------+-----------------+

As you can see, the relation for the SESD still needs to be defined.
And there are some others.... TBD.

- ray

> Toby
> 
> 
> On Thu, Sep 3, 2015 at 5:40 PM, Ray Kiddy <ray at ganymede.org> wrote:
> >
> > Hello -
> >
> > I am on a quest to learn more about how administrative boundaries
> > can be managed as relations. I have a bit of experience with these
> > things, but I am discovering the limitations of my knowledge also.
> > Which was the point, actually.
> >
> > I would like to be able to suggest that governmental entities could
> > manage their district geo data with OSM. I am interested in seeing
> > why this does not work now and what can be done. Well, and it would
> > be interesting to find out why so few cities in California actually
> > seem to have a relation. Or perhaps I am missing it.
> >
> > I know that TIGER data was imported into OSM, but I am seeing some
> > disconnects. To be precise:
> >
> > Using QGIS, I can load the vector files (SHP) from the following as
> > two different layers:
> >
> > ftp://ftp2.census.gov/geo/tiger/TIGER2015/ELSD/tl_2015_06_elsd.zip
> > ftp://ftp2.census.gov/geo/tiger/TIGER2015/EDGES/tl_2015_06085_edges.zip
> >
> > Using JOSM, I can see the "Sunnyvale East Channel":
> >
> > https://www.openstreetmap.org/#map=19/37.364537971457935/-122.02110206207291
> >
> > I cannot see how to get the id of this way from JOSM and its tag
> > info seems to be:
> >
> > boat=no
> > intermittent=yes
> > name=Sunnyvale East Channel
> > scvwd:FACILITY=2026
> > scvwd:ROUTEID=20260
> > waterway=drain
> >
> > And in QGIS, I can see the same feature (removing empty TIGER
> > fields):
> >
> > wkt_geom        LINESTRING
> >         -122.0213200000000171 37.36408999999999025
> >         -122.02097200000001465 37.36464099999999178
> >         -122.01992900000000475 37.36629399999999634
> >         -122.01963900000001217 37.36675400000000025
> > STATEFP 6
> > COUNTYFP        85
> > TLID    618169892
> > TFIDL   229597201
> > TFIDR   230278901
> > MTFCC   P0001
> > HYDROFLG        N
> > RAILFLG N
> > ROADFLG N
> > OLFFLG  N
> > EXTTYP  N
> > GCSEFLG N
> > OFFSETL N
> > OFFSETR N
> > TNIDF   39083667
> > TNIDT   409312163
> >
> > But there seems to be no connection between the feature in OSM and
> > the TIGER data. So, TIGER data was used to define new features? But
> > perhaps TIGER id data was not merged onto existing features?
> >
> > I am certainly not seeing "tiger:tlid"="618169892" associated with
> > this object in OSM anywhere.
> >
> > So, if I want to give the Sunnyvale District the relation that
> > defines its boundaries, I cannot use TIGER data to find those
> > lines? Or rather, I must use the TIGER data and find the line in
> > OSM and set up the connection myself?
> >
> > Ok.... Any other suggestions?
> >
> > thanx - ray
> >
> >
> > ps:
> >
> > My early stumblings are in my diary:
> >
> > http://www.openstreetmap.org/user/rayKiddy/diary
> >
> > Please excuse any ignorance on my part. I know a bit about the GIS
> > practices of the state of California. I have a very small bit of
> > experience with Santa Clara County. I have a smidgen of knowledge
> > about the city of Sunnyvale. And I have more exposure to the
> > Sunnyvale Elementary School District, having once been on the
> > Board. And I develop database software and am interested in mapping
> > applications.
> >