[Talk-us-massachusetts] Fwd: meadow & woods?

[Greg Troxel]

Jim Snyder-Grant <jim.snyder.grant at gmail.com> writes:

> I'm launched again into improving our paper and online maps for Acton
> trails.

I'm glad to hear that.  The Acton paper/pdf drawn maps are really nice
and it would be great to be able to make maps like those for Stow (and
other) conservation areas after entering the data.

> Our old maps had nicely drawn representations of landcover: forest,
> wetlands & meadows, mostly. It's easy to find wetlands data, but the other
> landcover data seems spotty. If I end up adding my own, I'd at least like
> to use tags that y'all find generally reasonable. (our map strategy is to
> start from OSM data as our base & and render different styles as needed).

There is also how to represent stone walls, but OSM has barrier=wall,
and I have been adding those from imagery and from being on the ground.
There is some leaf-off imagery (USGS large scale) which is useful for
this; bing tends to be leaf-on.  (I have also heard of scholarly papers
about using lidar to find walls and foundations, but not really read
them or tried it yet.)

My standard rant is that OSM is messy about following standards
established by professional communities.  But on the other hand one can
tag how one wants as long as it is reasonable and it's basically ok.
And I get it that you are trying to align with standards; I'm ranting
about the established tagging conventions.

A big issue in formal geography is that landuse (human geography) and
landcover (natural geography) are sort of separate.  But the way the
tags work doesn't entirely line up with that.  IMHO, in an ideal world,
there would be a set of landuse= tags that would describe everything
that humans did, and that each bit of land, often along lot boundaries,
would have a single landuse tag.  Then, there would be a set of
landcover= values, and every area would also have one of those.

Apparently I'm not the only one that thinks this:

  https://wiki.openstreetmap.org/wiki/Proposed_features/landcover

> I looked around Massachusetts using the https://overpass-turbo.eu
> interface, and looked at the tag wiki, and found a mixed bag, but I've
> ended up thinking that natural=xxxx is the way to go for now inside our
> conservation areas.

There is also the wiki.

> In Massachusetts, I see quite a few thousand uses of landuse=meadow. It's
> pretty sparse around here: I see one in Littleton and a bunch in Concord,
> so that seems OK.

As long as there's enough that no one will revert what you do, and it's
reasonable, I think it's ok.  But landuse= indicates human intent, vs
describing what is.  This is funny, becuase natural=meadow sounds
reasonable, except that meadows aren't natural and will quickly become
trees if you don't clear them every year.  So while the purist in me
doesn't like landuse=meadow, if that's what people are doing it doesn't
hurt.

> There's also even more of landuse=forest, but then the tag wiki wants
> us to save that for managed forests like tree plantations.

This is really the big problem.   I see landuse=forest as describing an
area, probably a lot boundary, that is used for forestry.  THat can be
occasional harvesting (our chapter 61 excuse) or it can be a real tree
farm.  But that's different from "there are trees here".

> The wiki database has info on a new key called 'landcover' that seems more
> like what I am looking for, but so far there's only 5 in all of MA,
> including a fine area in Stow: https://www.openstreetmap.org/way/183513179
>  with landcover=trees and natural=wood.

I put in a natural=wood recently to show an area covered by trees:

  http://osm.org/go/ZfIZ0HsQ?m=

What I like about that is that it shows an area that has trees, and
doesn't use landuse, so the fact that there is an underlying
landuse=farmland on the entire lot is not contradicted.

> Oh:  natural=wood might be good. There's a handful around here and
> thousands around the state. And natural=grassland is even less common, but
> at least it's used some, and in ways that I would expect.

So grassland is kind of like meadow, but with fewer shrubs, and more
than lawn?

Looking at this:

  https://wiki.openstreetmap.org/wiki/Key:natural

I think natural=grassland describes what I usually see, in that it's not
a hay field, and it's not short/manicured like grass, but more mowed
once a year to keep the trees out.

> So for now, that's how I plan to mark the innards of my conservation land
> areas: with natural=wood and natural=grassland. And I see that
> natural=wetland is already pretty well used. Hooray.

I think as long as you are hand editing things (even copying from
federal or massgis PD data) in conservation areas, nobody is going to
get upset.

https://wiki.openstreetmap.org/wiki/Tag:natural%3Dwood

wiki says not to use landuse=grass:
  https://wiki.openstreetmap.org/wiki/Tag:landuse%3Dgrass


The other thing to think about is not only what the tags ought to be and
then how you can render them, but how other renders deal with them.
This is somewhat unfortunate and somewhat just how things are that tags
being ok is viewed through the lens of the standard stylesheet for
mapnik on the web site.  Another thing I use that I think matters is
osmand (android), but it is easier to change.  (I can send you some
screenshots of that if you don't have an android device.)

So overall I think your plan is the right thing to do.

There have been descriptions of various things on the list, and I think
it would be great to have wiki pages linked under the massachusetts
page:

  https://wiki.openstreetmap.org/wiki/Massachusetts

One could basically say what tags you are using for this sort of
slightly ambiguous conservation land mapping and hopefully some day
pointers to any custom rendering stuff you may develop.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 180 bytes
Desc: not available
URL: <http://lists.openstreetmap.org/pipermail/talk-us-massachusetts/attachments/20161103/78b194b3/attachment.sig>