Discussion:
[Imports] Spartanburg County SC road centerlines import
Mike N
2018-10-22 03:20:04 UTC
Permalink
This is a proposed import of road centerlines for Spartanburg County SC,
based on county GIS data. This will include a systematic review of all
roads in the county and qualify to remove tiger:reviewed tags.

https://wiki.openstreetmap.org/wiki/Spartanburg_county_road_center_line_import
Mike N
2018-10-22 12:39:18 UTC
Permalink
Thank you for your comments. Answers inline.
Post by Mike N
This is a proposed import of road centerlines for Spartanburg County
SC, based on county GIS data.   This will include a systematic review
of all roads in the county and qualify to remove tiger:reviewed tags.
https://wiki.openstreetmap.org/wiki/Spartanburg_county_road_center_line_import
A roads import! 🙂
There's a few lanes that are weird. lanes=7 for a 6 lane road. It's
weird that some roads have lanes on some parts, not all (e.g. "Hollywood
Street"). Maybe try to make it consistant? JOSM validator has found a
handful of topology errors. There's ~100 examples of roads that aren't
connected properly (nodes on top of each other, but not connected).
You seem to be defaulting to "highway=residential" a lot (e.g. if you
dohn;t know another, or turning 'Gravel' into 'highway=residential
surface=gravel'). I don't know a lot about tagging in the USA, but isn't
there (wasn't there) some problem with the TIGER data using residential
too much?
The 'lanes' and highway type were experimental to see what useful
information could be mined from the source data. I agree that they are
all but useless for OSM's purpose. 95% of the work will be checking for
geometric alignment and name from the background image layer in the
editor. For example there have been many projects where sharp
intersections have been realigned for safety to create right angles.
And streets have been renamed for E911 purposes.

The one case where I see direct access to converted data is a new
residential subdivision - where a new group of roads would be copied
from the reference data and connected to existing data. Those would
nearly all be residential. So I didn't take the time to go back and
remove lane attributes from the raw data.

Defaulting to residential was not totally wrong for this county in
the same way it was wrong out west. The most likely mismatch would be a
new unclassified road into an industrial area - but those will likely be
single roads, and thus be as easy to hand trace and assign the correct
classification as to copy from the reference layer.
Can you link to your discussion with the local community, how/where did
that happen?
This was mostly verbal discussion with another community member, as
well as one of the meetups at
https://www.meetup.com/Open-Street-Map-upstate/ , and using some of the
ideas presented by Clifford Snow in his "Discover Rural America"
presentation at
.
The link the tasking manager project doesn't work.
Corrected.
I'm a little unclear about one big question: What are you doing with the
existing data in OSM? Existing OSM data seems to have nearly identical
locations to this new data. You're just going to update existing OSM
data? Do you know how much existing OSM data needs to be updated?
All existing data will be reviewed. Most of it will add the surface
attribute and lanes if visible from imagery and remove the
tiger:reviewed attribute. So nearly everything will be modified.

Stepping back to the big picture - although many hours have been
spent improving the road network in that county, OSM is the last source
I would use when planning a trip to an unfamiliar part of the county.
There have been other US projects in which a group would go into a "fast
growing region" and review all roads, adding surface and lane attributes
to improve navigation. The end goal of this project is similar. When
combined with some additional planned work such as address points, OSM
will be suitable as the primary reference source when planning a trip
through that county.
Rory McCann
2018-10-22 18:56:15 UTC
Permalink
Hi Mike.

Thanks for the answers, that clears things up. Buuuut....
On 10/22/2018 5:00 AM, Rory McCann wrote: >> I'm a little unclear about one big question: What are you doing with
the existing data in OSM? Existing OSM data seems to have nearly
identical locations to this new data. You're just going to update
existing OSM data? Do you know how much existing OSM data needs to be
updated?
  All existing data will be reviewed.   Most of it will add the surface
attribute and lanes if visible from imagery and remove the
tiger:reviewed attribute.   So nearly everything will be modified.
I'm sorry, maybe I'm having a brain fart, but I'm still confused. It
sounds like you're going to look at all existing OSM roads in that
county and manually review them? Just going through and fixing them up
and removing tiger:* tags, and keeping the existing roads in OSM? That
sounds great. But that's a regular map-a-thon, not an import. What do
you need this new data for? If I'm reading you right, this new data from
the county won't be used at all? Right?

You're not going to *replace* the existing OSM data with this new data,
right? You're not going to delete the existing OSM data, right?

If you (& friends) are going to fix up the roads, you don't need to talk
to this list. Just go ahead and do it! That's not an import. Just
tracing from the imagery you created from this data isn't an import.
That's just using a new imagery source. You can just go ahead and do that.

If you want to find new roads that aren't in OSM, load OSM & this new
data into postgres, and look for roads in the new dataset that are far
(>10m?) from anything in OSM. Should be quicker than humans looking at
all. 😉 (Do you know how to do that?)
--
Rory, who is still confused.
Mike N
2018-10-22 23:00:20 UTC
Permalink
Post by Rory McCann
Hi Mike.
Thanks for the answers, that clears things up. Buuuut....
On 10/22/2018 5:00 AM, Rory McCann wrote: >> I'm a little unclear
about one big question: What are you doing with
the existing data in OSM? Existing OSM data seems to have nearly
identical locations to this new data. You're just going to update
existing OSM data? Do you know how much existing OSM data needs to be
updated?
   All existing data will be reviewed.   Most of it will add the
surface attribute and lanes if visible from imagery and remove the
tiger:reviewed attribute.   So nearly everything will be modified.
I'm sorry, maybe I'm having a brain fart, but I'm still confused. It
sounds like you're going to look at all existing OSM roads in that
county and manually review them? Just going through and fixing them up
and removing tiger:* tags, and keeping the existing roads in OSM? That
sounds great. But that's a regular map-a-thon, not an import. What do
you need this new data for? If I'm reading you right, this new data from
the county won't be used at all? Right?
You're not going to *replace* the existing OSM data with this new data,
right? You're not going to delete the existing OSM data, right?
If you (& friends) are going to fix up the roads, you don't need to talk
to this list. Just go ahead and do it! That's not an import. Just
tracing from the imagery you created from this data isn't an import.
That's just using a new imagery source. You can just go ahead and do that.
If you want to find new roads that aren't in OSM, load OSM & this new
data into postgres, and look for roads in the new dataset that are far
(>10m?) from anything in OSM. Should be quicker than humans looking at
all. 😉 (Do you know how to do that?)
There will be 50 to 200 streets of new data used for new
subdivisions. I suppose that I could have created sets of data for
"These might be renamed", "These might be imported" , "These might be
adjusted" , "These might be deleted" (Because a diff doesn't identify
which one is right) , then not bothered to mention the additional review
which would indeed just be a local project.

If this is deemed not to be an import, then we will begin immediately.
Loading...