[OSM-talk] Do we care if its forest or wood? Natural world mapping ...

[Tyler]

(Sorry Tom, for the double sending, I didn't check the reply to: field)

Tom:

> I'd really like to nominate someone like Nick Whitelegg as Countryside Tsar

for a day, so he could work out the different basic features we need to

know about in the countryside and an appropriate tagging schema. Then, as

always, a combination of wiki documentation, Mapnik & Tiles at Home rules,

Xybot mischief and peer education could disseminate this sensible approach.


I'm going to go back to this because it makes so much sense to do. I too get
discouraged by the lack of comprehensible tags. I actually think that
"natural" key is a bad key. Is an artificial lake a natural=water or
something else? If it's a reservoir (and what lake isn't technically a
reservoir) is it sufficient to tag it just landuse=reservoir, and should we
tag it as man_made=water to explain that it's not actually natural?

No, clearly we shouldn't. So we could just accept that natural and landuse
are equivalent and adjust natural tagging as such (since changing everything
to landuse seems out of the question). So if we do that, then natural=wood
wood=managed or landuse=forestry or whatever becomes a reasonable way to
separate the landform from the land use

Greg:

> So, I think we need some tags that denote landcover, and some tags
> that denote legal status.


> so an area would have at most 1, preferably exactly one of:


> landcover=trees

landcover=swamp

....


>
Exactly what I was thinking (though I think just using natural as equivalent
to landcover might be the way to go at this point), and using the USGS style
landuse values [1] would be a good start covering the majority of cases (I
think just rolling them all into natural (including man made surfaces makes
sense at this point, but I accept I may be--and probably am--wrong)

and at most 1 of


> land_use=...


eh... I'm less fond of this, just because I'm not sold on there being 1 and
only 1 land use for an area but I have no supporting evidence to back up my
iffy feeling


> yes, land_use=forestry perhaps implies land_cover=trees, but in the case

of


> land_use=conservation


> I would expect a variety of landcover tags within the administrative boundary
> of the conservation area/park.


As would I, when I said solid fill earlier I mean more like hatching or even
a transparent overlay/underlay? for rendering, I'm pretty convinced about it
being a boundary=whatever issue at this point for things like parks/national
forests,DNR land, BLM land... but not convinced that something can't be both
say land_use=recreation and land_use=conservation (you can bike, and paddle
and fish, but you cant motor and litter and club baby seals)

Gustav:

> Because is see forests as something fundamentally different from a few
> trees in the corner of a park.


But would you classify them as a different landcover than say natural=wood,
wood=sparse or something. A 500 m^2 wooded area--rendering as forests do
elsewhere--inside a park is probably going to look like "hey look theres a
spot with trees" as opposed to "what is a tiny wilderness doing inside the
city park?"

[1]
http://gisdata.usgs.gov/edc_catalog/fetch_layer_docs.php?LayerName=NLCD%202001%20Land%20Cover

Anyway, I think it's all a cluster, I just thought I could pipe in to add to
the fun.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20090721/27520abf/attachment.html>