[Tagging] Tagging of topographic areas with a name

Martin Koppenhoefer dieterdreist at gmail.com
Tue Aug 6 17:04:36 UTC 2013

2013/8/6 Yuri D'Elia <wavexx at users.sourceforge.net>

> On 08/06/2013 04:27 PM, Yuri D'Elia wrote:
> > It's really a topographical information, and I feel like tagging objects
> > within or using relations might be really problematic. Just imagine what
> > kind of "spotty" tagging would you have for big mountain groups. Huts
> > and peaks would definitely not be enough for a decent boundary.
> >
> > But also drawing big areas is kind of ugly :(.
> >
> > Fortunately, the boundaries of the area are not important in themselves.
> > Nobody renders valley or mountain group borders. But we *do* use such
> > boundaries for name placement.
> >
> > I'm thorn.
> I'm attaching a crude osm file I edited quickly to demonstrate the problem.
> Valleys usually end exactly at the mountain ridges. Valleys also end at
> the border of a mountain region or at the border of another valley.

+1, valleys aren't too big usually and should be clearly defined, there is
already a proposal for ridges and it is also used: natural=ridge

to define a valley it should be enough to add the adjacent ridges to an
area relation (type=area) and add tags like natural=valley, name=* ...
renderers could in the future connect the ridged to create an (implicit)
area (e.g. to put a text inside).

For other areas other data types might be more adequate:
Some years ago on the German ML there was this interesting idea to define
(fuzzy) areas (e.g. lower scale topographic regions like "the European
Alps"). You put existing objects (like nodes, ways or relations) into a
relation with the roles inside or outside and some algorithm would
calculate an area that includes all inside and excludes all outside
objects. You won't have to be very precise with this, as this kind of rough
information is only required on lower scales where some kilometers more or
less won't change anything, just a few nodes should suffice to define
something as huge as the Alps, and you could reuse (preferably simple and
stable like peak-nodes) existing geometry.

> In the alps I would expect a mosaic which is essentially totally filled
> with valleys.


  A relation would be great to re-use existing geometry, but
> some new boundary type will also be needed to mark the end where's no
> additional geometry can be reused.

if you need explicit boundaries between 2 valleys (see above the area
relation which doesn't require to explicitly draw these, but allows to do
so if required (role=lateral).

> I also created two (inexact) mountain groups. Mountain groups actually
> form a complimentary mosaic, as you see in the file. A mountain group
> would start at the middle of a valley (which I didn't do in the example,
> but you get the point) and end at another one.

+1, usually you will have a river or stream there, as it is the locally
lowest point (i.e. the needed geometry is already there). An argument
against reusing rivers to define mountain groups is that they often add a
lot of complexity and you'd usually not need the borders of a mountain
group with the precision this allows for (adding relations augments
complexity and raises the barrier for other mappers to edit).


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20130806/2b297585/attachment-0001.html>

More information about the Tagging mailing list