Open talllguy opened 8 years ago
@talllguy Great comments! I'll look into this a bit more.
We've created a testing MapRoulette challenge to address unreviewed roads in a small section north of L.A. city. http://maproulette.org/map/237
But we need to address those other things as well.
Some roads in the north county have different names than what county has them identified
There are quite a few cases where gaps between buildings are either very small (significantly less than a meter) or buildings touch at a single node.
These can be identified. What should be done with these? Should they be morphed so they don't touch?
roads now intersect buildings indicating that mapping is inaccurate
These too can be identified. The part of the building that touches the road can be shrinked away from the road. Would this solve the issue?
@wsdookadr could you provide some examples?
A lot of buildings in the county do touch, so it's fine if they share a node. Overlapping buildings isn't preferred though.
And some of the road data is still quite bad for the county. So it may be that the road is still based on the 2007 Tiger Line file import. We've got a MapRoulette challenge for roads in a corner of the county.
But if you have some specific examples we can figure out where the problem is. Thanks!
Mirroring a comment from the OSM mailing list below.
My comments are that the first issue with the sheds is something that is probably an artifact of the conflation (#67) but I haven't been following closely enough to say.
For the second bit, the overlapping roads: I think that when bad data is identified by an import is justification enough to even do the import! In this case building footprints have identified wonky TIGER streets. However, from a workflow perspective, I would think this might be caught in the validation phase (in JOSM), but the JOSM validator doesn't always catch everything.