[Talk-us] Creating a multipolygon of a body of water across state boundaries

[Brad Neuhauser]

It's one body of water, so I would create one multipolygon for the whole
reservoir. Data providers should be able to deal with things crossing
administrative boundaries, it happens with all sorts of objects.

Note that the user who did this made chunks of areas over the water--maybe
to avoid the 2000 node limit, and/or to avoid creating a multipolygon. I
did some of that five years ago, too ;). It would be pretty easy to convert
to one multipolygon, though: cut the area ways where they join, add the
shore portions to the multipolygon as outer ways, and delete the ways that
cross the water. Make sure all the outer ways join each other--a small gap
could cause rendering issues. All tags should be on the multipolygon and
none on the outer ways.

Then islands can be added as inner members of the multipolygon. In
addition, islands can be tagged place=island, especially if they have a
name. Sometimes I also see them tagged with landcover (ie natural=forest,
etc)

Cheers,
Brad


On Wed, Jun 18, 2014 at 10:02 AM, Will Skora <skorasaurus at gmail.com> wrote:

> I noticed that some islands within Pymatuning Reservoir
> ttps://www.openstreetmap.org/#map=12/41.6074/-80.5203 , a large reservoir
> that
> cuts across the Ohio and Pennsylvania borders for 8 miles, weren't
> visible on the main map layer and was going to trace them.
>
> The islands were tagged natural=land, what I recall was an outdated
> tagging scheme. The water is consists of multiple adjacent closed ways
> tagged 'natural=water'
>
> I wasn't sure whether natural=coastline was the correct current
> tagging for them, so I checked the wiki page and noticed that for
> waterways and bodies of water (that aren't the ocean), to use a
> multipolygon relation with tagging the islands as 'inner'
>
> My hunch is that if I create the multipolygon relation across multiple
> states, some state extracts [OHIO, PA] may not contain any of the
> reservoir if I were to create one multipolygon relation for all of the
> Reservoir. Is this correct?
>
> What's the recommended course of action to satisfy consumers of OSM
> data and mappers?
>
> Regards,
> Will
>
> _______________________________________________
> Talk-us mailing list
> Talk-us at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-us
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20140618/8bcbdb7f/attachment.html>