[Talk-us-massachusetts] MassGIS landcover dataset sample

[Greg Troxel]

hobbit at techno-fandom.org writes:

> Wow, that looks pretty busy.  Trying to jam *every* little patch of trees
> into the database seems almost like overkill, despite the present state of
> RI and NJ and others ... On the what-to-do-about-parks angle, I would think
> a reasonable approach might be to map the major treed areas that extend
> outside the boundaries, in a relation with the major internal park non-wood
> features carved out like Tom P. did with the Fells, and then at least in
> Carto you get a lot of green *with* the official boundary visible as the
> darker line through it.  Personally I think that looks fairly straightforward,
> and still carries the idea that the woods aren't just about the park parcels
> without necessarily making *everything* "green from space'.

The database is about semantics.  As I see it, either landcover is legit
or it isn't.  I don't think it's valid to argue that landcover should
only be added in moderation to get specific rendering.

If having accurate landcover everywhere results in a displeasing render,
that means the renderer should be changed.

I do think that it is probably best to merge deciduous/evergreen, which
will greatly reduce object count, and perhaps omit small objects.

> The kicker I learned last week is that those inner carvings can't be relations,
> or at least it's inadvisable, which came as a surprise.  Ways it is, then.
> Otherwise, I don't think that would be all that hard to draw in the supersets
> of real wood landcover, even with iD working in small viewports.  No large-
> object copying needed.

Yes, that's all complicated.  But landcover tags shouldn't be on
non-landcover objects, so this shouldn't be a practical problem.

> I thought "conservation" had some very special particular meaning in MA?  I
> saw discussion of that go by in my big catch-up but didn't really absorb it,
> and  https://wiki.openstreetmap.org/wiki/Massachusetts/Conservation  sort of
> just muddies the issue.

What do you find confusing about what is under Tagging?

Specifically, "purpose is to indefinitely preserve" leads to
landuse=conservation and "legal prohibition against development" leads
to boundary=protected_area.

> I wonder if it would be possible to take a selection of GIS woods data and
> make a layer with a kind of "fuzziness" with an area threshold, and then drop
> out the tiny patches below that threshold to leave the larger significant
> areas of real woods asserted?  I suspect that no automated process will ever
> deliver the "right thing" without a lot of post-intervention, though.

That is what I am thinking:

  first, deal with ponds and swamps getting in the db.  This is a big
  deal separately.   Secondarily streams (harder to automate).

  figure out how to deal with the mismatch in the swamp areas in the
  massdep hydrology dataset (which match my on-the-ground observations)
  and "palustrine forested wetland" (which are bigger than what I see on
  the ground, and maybe merge some of those (PFW but not swamp ==>
  natural=wood).

  set deciduous/evergeen/faux-swamp to one tag

  merge all adjacent polygons with same tag, deal with inner rings

  select polygons with some minimum area

Then that can be QCd for "does it seem that way on imagery with very
high probability", perhaps town by town.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 194 bytes
Desc: not available
URL: <http://lists.openstreetmap.org/pipermail/talk-us-massachusetts/attachments/20220727/bb7297a5/attachment.sig>