2017-11-02 8:14 GMT+01:00 Tomas Straupis <tomasstrau...@gmail.com>: > Currently according to taginfo the most popular are: > natural=wood 4,5M > landuse=forest 3,5M > others are way behind. for example landcover=trees - 11000 objects... > > So maybe there is a point to choose one of the two popular tags and be > done with it? > >
IMHO there are semantic implications in the key, as has been said many times, "landuse" is about the human use of land. "natural=wood" reads as "is a wood". Many trees don't grow in a wood but along streets, in small groups that aren't woods, in gardens, etc. The main reason natural=wood 4,5M and landuse=forest 3,5M are used much more than any other tag is existing rendering rules in OSM carto and presets. As this topic is popping up frequently since at least 10 years, I guess there is some problem with the established tagging, hence looking at the naked numbers doesn't bring us further. If you look at the actual objects that have these tags, you'll find that many are neither "wood"s nor "forest"s according to any definition. They are simply groups of trees or areas covered with trees. A forest and also a wood are more than just trees, they are also implying an ecosystem, microclima, etc. > If anybody wants more detail - subtags could be used: wood=xxx or > forest=xxx depending on which one is chosen? Editors would remove > other tags from presets, changes will be done in the database and then > rendering and data extraction could be simplified. > what is the "thing" you want to tag with the tag to be chosen (what are the basic characteristics, that are implied by the main tag)? Trees growing there? A forest? There's also still the problem with names: typically any bigger forest with a name has smaller parts with their own names, which again have smaller parts with their own names, etc. It doesn't (IMHO) make sense to have nested same value landuses. The solution could be either "natural" objects or place objects (or yet another new key). In any bigger (named) forest you will also typically find areas which are by common interpretation part of the forest (e.g. clearings, lakes) but aren't actually tree covered (again an argument against landcover). Cheers, Martin
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk