[Tagging] tree rows vs individual trees

Tobias Knerr osm at tobias-knerr.de
Sat Feb 9 19:15:07 UTC 2019


On 09.02.19 15:23, Tom Pfeifer wrote:
> "Tree rows ... This approach can also be combined with individually
> mapped trees for further details."
[...]
> IMHO this violates the one object - one OSM element principle. Either I
> choose the coarser approach to map a way for the row, or I refine it to
> individual trees, but should not use the row anymore.

Because the two feature types exist at different levels of abstraction
(a tree is *part* of a tree row), I do not see this as a violation of
one feature, one element.

Instead, I consider it comparable to mapping building:part areas within
a building=residential outline within a landuse=residential, or mapping
amenity=parking_space areas within an amenity=parking.

> If a renderer wants to cluster any trees that can be done algorithmically.

Writing an algorithm to reconstruct tree rows from individual tree nodes
is probably possible, but it's more complex than what OSM renderers
usually bother with – even for features that are much more significant
than tree rows. Checking whether a tree node is part of a tree_row way,
on the other hand, is far easier and only requires relatively standard
OSM tooling. So the latter seems like the more practical solution to me.

Tobias





More information about the Tagging mailing list