[Tagging] Mapping the Black Forest

Martin Koppenhoefer dieterdreist at gmail.com
Wed Oct 9 15:03:19 UTC 2013

2013/10/9 Tobias <cra_klinrain at gmx.de>

> > It might be a better idea to start a parallel project (e.g. with
> shapefiles
> > in a more adequate scale), e.g. starting with natural earths physical
> > dataset and adding translations and refinements. This dataset could be
> > crowdsourced and distributed in a license compatible with osm in order to
> > make them mixable.
> A parallel project, let's name it OpenGeoMap, would neccessarily require
> to merge the projects in a way that does not bother anybody.

in my understanding you would not have to merge them, you could use them in
parallel (say as an overlay, or an additional layer in rendering phase). In
the end, OSM is very detailed, much more detailed than these topographical
regions, take the "black forest" from your example: it is usually divided
into the southern, middle and the northern black forest, but according to
newer classification divided into far more sub-entities as can be seen
Still, all these subregions cannot be drawn in a spatial definition
comparable to the osm data, they are rather blurry. From looking at this
table in wikipedia it seems to depend on the geological characteristics
where they put them, i.e. this won't be surveyable by laymen and probably
other than specialist knowledge you'll also need specialist equipment to
survey (because we are talking about underground properties). As a solution
we'd copy from the experts ;-)

It depends what you wanted to do with this data, if you want to draw a
nicely curved text "Schwarzwald" on a zoom 9 tile you will not need very
detailed data, if you want to decide whether a given house is still in the
"Schwarzwald" or already in the rhine valley you'll probably find out that
also a very detailed map still might leave this up to its individual
definition of both (if its on the border the answer might depend on who you
ask, and an answer like "on the border of both" might be more reasonable
than deciding for one).

Well, you could use a rendered osm background (i.e. a slippy map) and draw
above, say in QGis or similar, some "rough" polygons and refine these
initial polygons iteratively when you notice that there are problems, but
where would you get the information from? In the end it seems more
promising to collect and reassemble the findings of experts (e.g. these
maps published in Wikipedia and based on the Bundesanstalt für Landeskunde)
and distribute them (if legally possible) as a unified dataset with
translations (i.e. you will have to create a consistent hierarchy, name and
translation columns as attributes for the geometry). You will have the
spatial reference so you could at any time merge this dataset with OSM if
you needed to.

> Regarding other natural tags like natural=desert etc and borders such a
> project would be a great idea.

also these other tags like "desert" are not yet in general usage, and they
suffer from similar problems (blurry borders, unclear definitions, not
easily surveyable).

> Regarding other data types it is not a benefit I would say, because
> keeping the data seperated would deny to relate them to each other.

see above, they will always relate to each other because of the geocoding,
and they would never fit 100% because of the different scales involved by
their nature (how they are defined / get surveyed).

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20131009/2b4bbf4c/attachment.html>

More information about the Tagging mailing list