[Imports] Should this 4-year-old import be deleted? (Ulster County, NY)

Kevin Kenny kevin.b.kenny at gmail.com
Wed Jan 23 22:56:48 UTC 2019


On Wed, Jan 23, 2019 at 4:27 PM Frederik Ramm <frederik at remote.org> wrote:
> I stumbled across this old import:
>
> https://www.openstreetmap.org/changeset/25649897
>
> (There's a few more with source=Ulster County GIS.)
>
> It contains a ton of so-called building footprints, but in reality these
> are all squares, independent of the actual building footprint. Also it
> covers only approximately half of existing buildings, and has many
> buildings where there's nothing visible on aerial imagery at all.
>
> To me this looks like a really low-quality data set that should be
> removed, at least where objects haven't been touched by mappers since. I
> did not see one single house in this import that actually matched the
> situation visible on aerial imagery.
>
> Even if you're a fan of building imports, you would likely agree that
> having this data in place will massively complicate any later conflation
> of a better data set.


For once, I agree with you about an import!  :)

I won't say that *none* of them have been touched, because I know that
I tweaked https://www.openstreetmap.org/way/304844693 when I parked
there for a hike on a less-popular route to
https://www.openstreetmap.org/node/357593337. (I didn't trace the
parking area because the aerials I had at the time were "leaves on"
images and I couldn't make it out under the tree cover. Obviously, I
didn't get the outline of the trailer quite right, either, according
to the latest New York State orthos.)

I assure you that I had nothing to do with the original import, and
would not have imported that mess!

When I discovered the problem independently, I decided to leave well
enough alone, because when I spot checked, most of the little squares
were at least close to, if not overlapping, actual buildings. I also
thought, mistakenly it turns out, that they had street addresses
associated with them, which would make them useful for navigation.
Instead, they're little boxes, with no other information. Moreover,
I've been holding off on taking any action until I could review and
decide on a suggestion as to what, if anything, should be done in
light of the http://gis.ny.gov/streets/ project. (Address point import
has been relatively non-controversial in the past, but I don't have
the time to lead an import on that scale, and I haven't been able to
spot-check the data quality.)

The import has been significantly edited in some of the villages -
check out the area near
https://www.openstreetmap.org/way/304810649#map=19/42.08528/-74.31553
for instance. Please do check that mappers haven't repaired it before
doing any sort of mechanical deletion. Aside from the places where
it's been improved, it doesn't appear to be adding very much value.



More information about the Imports mailing list