[Talk-us-massachusetts] Mapping stone walls

[Greg Troxel]

Marc Sevigny <marc.sevigny at gmail.com> writes:

> Last winter, I worked on mapping stone walls in NH using LIDAR imagery.
> There has been a push in my town (Harvard) to map all stone walls, as a
> cultural and historical resource worthy of protection.

Certainly a mapper putting real things in the DB is 100% fine, of
course.  If the town wants to map them that probably raises issues *on
their end* about control over changing the dataset.  If you are just
mapping them as a private activity, not your problem, but I find that
town government's goals and OSM's goals sometimes do not align.

I would be very interested to know if this has been thought through and
there's a written plan (which would be a public record, I'd think).


What was the horizontal and vertical resolution of the LIDAR you used?
In MA near us, I think it's 1m H and also 1m V.  It seems to me that
better vertical would really help.

> I plan on using the same method used in the NH project
> https://granit.unh.edu/resourcelibrary/specialtopics/stonewalls/ to record
> them.

I went to that page and didn't immediately understand the method.  I
also don't understand licensing of the resulting dataset.

Is this just humans looking at LIDAR Shaded Relief (LSR hence) and
drawing, like we'd do in josm?  Or is there something more?  I am aware
of some other work:

  https://stonewall.uconn.edu/investigation/mapping-and-gis/#
  
that I think involves processing sort of like hill shading but intended
to find walls and building foundations.

Do you mean you intend to spend some time looking over LSR in Harvard
and draw walls?

>  I'd tag those that are added as "unverified" until ground-truthed.

I've been mapping walls, and also field checking, in Stow Conservation
Trust property.  Sometimes the signature of stone walls is really strong
and there is little risk of error, and sometimes it's iffy.  What I've
done is add "fixme=field check" when I am doubtful.  That's a little
loud.

I think it's excessive to mark everyone unverified; we don't do that
with buildings from imagery.

So I'd suggest marking unverified=yes (Is there a page/consensus on
this?) only when you are doubtful.

I would suggest fixme=field check if you are really doubtful and intend
to go check in the next 6 months.

I have found it useful to map everything I can see and even things that
are iffy with fixme, and then go hike a property and check the iffy
ones.  Sometimes they are phantom, and sometimes they are real with more
real wall extending that I didn't perceive.

I have found the MassGIS LIDAR horizontal accuracy to be excellent, when
compared to dual-frequency 4-constellation RTK.

> Does anyone have any concerns with this?  I would follow the recommendation
> found here: https://wiki.openstreetmap.org/wiki/Tag:wall%3Ddry_stone
>
> Are there downsides?  Could this clutter up rendering?

I have been adding some based on just me looking at the MassGIS LIDAR
Shaded Relief.  I don't feel like there is any problem.  I find the wall
line very thin on the standard render, and I perceive it as really
useful, not clutter.

I don't think you should worry about the effect on rendering at all.  If
there is valid data, and the render isn't pleasing to someone, the
people that to the rendering can change that.   In my view the problem
with the standard render is the other way around, how it tends to
control tagging.
-------------- 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/20210206/96eba4bb/attachment.sig>