[Tagging] landuse=residential and named residential areas which belong together (neighbourhoods/subdivisions?)

Martin Koppenhoefer dieterdreist at gmail.com
Mon Aug 29 14:06:08 BST 2011

We have been recently discussing on the German ML about
landuse=residential. In Germany many mappers were mapping subdivisions
/ neighbourhoods [1] with landuse=residential. This led to very rough
landuse information, because in order to keep the (sometimes quite
big) area as a whole they are forced to ignore landuses that don't

I think that the information about these areas belongs to a distinct
entity and would best fit into the "place"-namespace because I see
them as subdivision of settlements (or more precise subdivisions of
what we call "suburb" in OSM).

If we could agree that the information about which areas as a whole
form a distinct unit (part of a suburb) should go into the place tag,
we would be more flexible when deciding where to apply landuse to.
Part of the current discussion results from the mixing of namespaces:
landuse is used to map settlement subdivisions which are IMHO defined
by other properties (like morphology, history, culture, typology, ...)
then just similar landuse. There is also areas which do combine
different subareas with different landuses (impossible to map this
with landuse-areas without either ignoring bigger parts of different
use or loosing the entity as a whole).

There are several questions in this context which are generally still
open in OSM:

1. does the road belong to the landuse at its sides?
-- as we don't (yet?) tag highway-areas for streets that are
represented with a centre line this is an ongoing debate. For railways
this is easier to decide because there will be landuse=railway along
the tracks. Some mappers advocate to reuse street nodes others prefer
to map these areas at the actual border of the parcels (or the
estimation of the latter). Some pros and cons can be found here:

2. What is the desired granularity for landuse?
-- Shall landuse-areas be huge areas comprising several streets and
blocks, or is it desirable to indicate the use of each lot if it is
differing from the surrounding? Or could the landuse areas even go
below single plot size?

My personal answer on these questions is currently:
To me a public road is a different landuse. Although it will not harm
if we initially approximate the landuse with bigger areas comprising
inner roads the ideal would be to have equal landuse in block-size
units. This is most easily refinable and would also give the
information of outer plot limits (adjacent to the road) so that you
could infer the public land of the road.

I would not go below the size of single plots (if there is not very
good reason and a very big plot) and I would also still adhere to the
rule of "predominant" use: if there is a building with offices in the
ground floor and 3 storeys of apartments above this would be
landuse=residential to me. If there was a whole plot with a factory
inside a residential area I would tag this plot as landuse=industrial.
Summed up the desired granularity I suggest to agree on is plot size
for landuse (but we will not necessarily map all plots as distinct
areas, it can also be bigger polygons).



[1] Due to political, cultural and economic differences I am not sure
if it makes sense to use the terms subdivisions / neighbourhoods
because I guess the precise meaning is somehow different.

More information about the Tagging mailing list