[OSM-talk] A message to our friends at HOT, Peace Corps etc. about Changeset Comments
Christoph Hormann
chris_hormann at gmx.de
Thu Nov 19 16:26:01 UTC 2015
On Thursday 19 November 2015, Kate Chapman wrote:
> >
> > And if on http://tasks.hotosm.org/project/1300 i read:
> >
> > "Please draw one large area outline around groups of buildings and
> > tag them landuse=residential"
> >
> > that is in violation of one of the core principles of OSM, namely
> > to map reality, what's on the ground. It instructs mappers to map
> > something that does not exist in reality based on abstract
> > geometric considerations and to give it a tag that is meant for
> > something different.
> >
> > Referring to
>
> landuse=residential is a globally used tag, so you can hardly call
> out HOT for using it.
Is it really so difficult to understand that the cited instructions are
wrong and lead to bogus mapping like here:
http://mc.bbbike.org/mc/?lon=70.356396&lat=37.310341&zoom=14&num=2&mt0=mapbox-satellite&mt1=mapnik
This might look plausible to someone from Europe or North America but
the truth is in terms of mapping reality this is not even inaccurate,
it is pure garbage as far as landuse mapping is concerned. The actual
area that would truely qualify as landuse=residential is likely only
about 5-10 percent of what is mapped here - if at all, in many cases
the criterion 'predominantly residential' is likely not met and
landuse=farmyard would be more accurate.
And to address Richard/Frederik: This is not the same as early mapping
in Europe, here we have a settlement structure which is fundamentally
different from that of rural Afghanistan. You can not reliably
identify any of these settlements on Landsat imagery. The coarse
landuse drawing is merely extrapolated from the presence of buildings
and is not connected to actually observable landuse on any scale.
But there is no sense in getting lost in this particular example - it is
just that, an example. So i repeat my suggestion to improve QA of the
mapping instructions and allowing the OSM community to effectively
correct mistakes there.
--
Christoph Hormann
http://www.imagico.de/
More information about the talk
mailing list