[Tagging] landcover=trees definition

[Martin Koppenhoefer]

sent from a phone

> Am 10.08.2015 um 12:29 schrieb Jean-Marc Liotier <jm at liotier.org>:
> 
> To me, it seems that mapping this area as a combination of landuse=residential and landcover=grass would be most fitting.



just do it


> I have thought about using the landuse=residential + natural=grass combination instead, but those lawns do not strike me as natural.


I think that "natural" should not be read as "made by mother nature" but rather as some kind of typically natural feature (in particular things that can be seen as named entity, like a lake, a "piece of forest" that has a name, a spring, a peak, a bay, a beach, a dune, a tree, a ridge, a mountain pass, an island, a vulcano ... and these can also be nested), and these could also be man made in exceptional cases (like an artificial hill could get a natural=peak the same as a natural hill). Unfortunately some values have been more or less established, since I came up with this idea, that don't fit into this way of reading it, in particular grass, mud, sand, but would fit into landcover, and "water" was already an odd one (but with water=x refinement it does fit). So maybe natural has realistically to be seen as a mix of different concepts (could be sorted in preprocessing according to the value).

For not very high scale mapping, secondary (and tertiary?) landcovers might add further detail (like trees and what grows below them, or grassland with scattered bushes and trees where you don't want to map single trees and bushes).



> 
> What do you all think ? Is this a good illustration of the need for landcover=* ?


parks, especially the bigger ones, are also good examples, where people like to micromap, and where it's hard to sustain that every group of trees is a little forest ;-)


cheers,
Martin