[Talk-us] misuse of the landuse=forest tag for national forests

stevea steveaOSM at softworkers.com
Sun May 12 23:12:16 UTC 2013

>If you have an area that cannot grow trees, due to altitude, 
>inadequate groundwater, or having exposed rock rather than soil (as 
>with many mountaintops), then, in what sense is it a managed forest? 
>I am not talking about areas that are temporarily treeless due to 
>the trees having been harvested.

I am talking about how "landuse=forest" as a tag has come to imply 
the semantic of "managed forest."  I'm not saying this is not ALSO 
tangled up with "usually implies a landcover of trees" -- it IS 
tangled up exactly like that.  We agree "harvested of trees, and so 
treeless" is still OK to call "landuse=forest."  Why?  Because we 
agree that the area so denoted with that tag is "a managed forest."

Additionally, if an area "above the treeline" which will never grow 
trees is ALSO part of the inside of a polygon of "managed forest" it, 
too, should continue to remain "landuse=forest" for exactly the same 
reason:  it is an area within managed forest.  Perhaps it is also 
roadless and perhaps it is also the even higher-protected class of 
wilderness -- OK, that's fine, but we frequently (and correctly, I 
argue) add wilderness areas as "overlay polygons" within 
landuse=forest areas without confusion.  And that's correct, and even 
mapnik renders these in a certain way that can be seen (NR over dark 
green with trees).

What many (all?) seem to agree on is that landcover could very well 
be improved.  Yes, in both use of the tag AND in rendering support!

What some seem to disagree on is that "landuse=forest" is the correct 
tag for areas which are "managed forest land, whether with or without 
trees."  In my opinion, it means exactly that, as that is how it 
seems to be extensively, if not exclusively used.  At least in the 
USA where my (wider) focus is.

Then, Clifford Snow and Martin Koppenhoefer get into it:

>2013/5/12 Clifford Snow 
><<mailto:clifford at snowandsnow.us>clifford at snowandsnow.us>
>For administrative purposes and area is designated a National 
>Forest. In practice parts of the National Forest have no trees. Show 
>nature=* make sense if we want to show what occupies the land. 
>National Forest boundaries should be in OSM,

Martin replies:

>yes, but not as forests but as kind of protected area.

A "landuse=forest" truly, logically and de facto is "mapped" (in the 
logical sense) as a "forest (national) which is a protected area." 
Tags, including:

landuse=forest, boundary=national_park, boundary:type=protected_area, 
protect_class=6, protection_title=National Forest, ownership=national 
and name=Name of Forest

as I mention in my previous post exactly capture all of these 
semantics, and properly.  Additionally, wilderness areas and THEIR 
tags (again, in my previous post) capture THOSE semantics.  There 
seem to be no problems with these, except a complaint that some 
people don't like large blocks of green represented by mapnik's 
interpretation of landuse=forest rendering as "dark green with tree 
icons."  However, I still maintain that the landuse=forest tag is 
accurate as it is used, and if that rendering is not to your liking, 
please file a bug with the trac system of mapnik.  We are using the 
tag "landuse=forest" to denote national forests, not to mean anything 
particular regarding the landcover within them.  The word "use" (in 
landuse) strongly implies this.  The distinction landCOVER (from 
landUSE) is not lost on me, nor should it be on anybody else.

Would I like to see more extensive use of the landcover tag with 
mapnik/standard rendering support -- perhaps even with fancy 
halftoning algorithms for "80% scrub" or "50% trees"?  OF COURSE! 
But we are only in the early stages of getting there.  Let's not 
dismantle established semantics while we do so.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20130512/e502b1d1/attachment.html>

More information about the Talk-us mailing list