[talk-ch] vandalism under the pretense of "simplifying"

Sarah Hoffmann lonvia at denofr.de
Thu Jun 30 20:54:10 CEST 2022

On Thu, Jun 30, 2022 at 06:30:20PM +0200, RB wrote:
> I am sorry but I don't agree with their arguments and I have addressed them
> before. Let me restate.
> Sarah makes the argument that the forest changes. Of course it does. I
> think that precise mapping in this case is a feature not a bug as it will
> in the future allow us to study the evolution of the vegetation.

No, that is not at all the point I was trying to make. I was saying that
the definition where a forest ends is fuzzy at any one given point in
time. Here is how you can test this: give the current Swisstopo imagery
to 10 different mappers and tell them to map the forest on them with the
highest precision possible. I guarantee you that you will get 10 different
results. The deviation between the results will tell you how much
fuzziness there is in a boundary and consequentely what preceision is
worth mapping. OSM is at its best when the average of the 10 opinions is
what is actually mapped.

Keep up your good work in refining the forests but maybe consider the
following points to avoid future conflicts with other mappers:

* Use a few less points. There is no hard limit here and sometimes a
  sharp bend will do well with more points, but an average 10m distance
  is probably a good rule-of-thumb for the image quality we have today.
* Try to straighten things out a bit. If the treeline looks straight on
  average, then map it straight instead of going around every tree branch.
  That does not mean that you should return to pointy angles, I just mean
  mapping the treeline a little less wiggly in the spirit of the average I was
  explaining above.
* Consider using natural=tree nodes for single trees and small groups of
  trees and also when there is a tree slightly outside the treeline
  where currently you would add a little appendix.
  (e.g. https://www.openstreetmap.org/#map=19/46.30407/6.90451)


More information about the talk-ch mailing list