User talk:Rudolf/draft landcover
landcover=wood ... prefer trees?
Shouldn't we avoid direct adoption of remote sensing principles?
Idea of using landcover concept seems to be good and completely functional. But I can see certain flaws in direct adoption of it from Remote Sensing science. In Remote Sensing, concept of landcover is single-layered. I mean, usually, there are non-intersecting areas with only one class assigned to each area. But there is one important case that doesn't fall into this principle: wooded areas. When tracing some satellite imagery, it's usually impossible to tell, if trees are forming natural ecosystem (layered forest with trees, shrubs and grass, for example) or not. We have massive thread about that in Russian community forum, and people are reasonably debating usage of "natural=wood" tag for trees, planted among buildings within city limit. Many people intuitively feel there is a huge difference between natural ecosystem (part of real forest) and group of trees growing in the middle of some lawn (regardless of being planted or naturally growing, but not forming any kind of natural woodland ecosystem). So, I think, landcover class for area with wooded vegetation should reflect only one elementary feature - trees. That's why I'd suggest different wording for it: landcover=trees. And this core tag could be extended by others, telling if it's managed vegetation, what's the purpose of it and so on. Otherwise, we'll have similar debates about natural/non-natural wooded vegetation and so on. --BushmanK (talk) 19:56, 8 January 2016 (UTC)