[Talk-us] Deleting tiger:reviewed=no/addr:street for routes (was: Streaming JOSM -- suggestions?)

Jmapb jmapb at gmx.com
Sun Jul 12 16:27:16 UTC 2020


On 7/9/2020 6:48 PM, Kevin Kenny wrote:
> Personally, I think even that much is overkill for deleting tiger:reviewed.
> I think that surface, lanes, and traffic controls are things that a
> mapper can notice are not mapped, irrespective of the TIGER review
> status. There are lots of hand-mapped roads that don't have the
> information!
>
> I'm willing to delete the tag when:
>
> (1) I've checked alignment against two sets of aerials, at least one
> with the leaves off. (In my case, that's almost always Maxar and NYS
> Orthos Online.)
> (2) I've added all bridges and culverts that I can identify on
> aerials. (Which always leads me down the rabbit hole of mapping the
> corresponding waterways)
> (2) I've verified that the name matches the state DOT highway map and
> the E911 address points.
> (3) I've adjusted the road class (TIGER's 'residential' can mean
> anything from a tertiary highway to a track!)
> (4) I've created route relations if the road has a ref (and removed
> the ref from the road's names!)
>
> I don't do 'lanes' very often.  I do 'surface' if the road is
> obviously not hard-surfaced (sometimes I can even see the ruts in
> aerials), and I do traffic controls only when surveying in person,
> which I always do afoot.
>
> I'd like a way to indicate that an intersection is uncontrolled. I've
> found myself returning on foot several times to the same intersection
> to look for STOP signs that aren't there, because I can't remember
> that I've checked it already.
>
> The reason that I'm so lax is that in my part of the state, TIGER is
> _horrible_ and mappers are scarce.  I chronically lack time to do very
> much about it, although I've at least checked the above information
> for all the unreviewed roads in my home county (barring some service
> ways that I'm not sure I can access legally). I work intermittently on
> a couple of neighbouring counties. There are a lot of service ways
> 'residential' ways in TIGER that are a mile or two off from the
> correct alignment or are otherwise ridiculous. At this point, in my
> area, 'tiger:reviewed=no' means 'beware: this road likely is entirely
> hallucinatory' and I kill the tag once I've verified that the
> information that TIGER provided is correct. The information that TIGER
> didn't ordinarily provide, I can leave for others (possibly including
> future-Kevin).
>
I've also been chipping away at TIGER junk in NY state (mostly Ulster
County) and I think my methodology's similar. I try to delete
tiger:reviewed=no if I'm reasonably confident that I've either confirmed
or fixed everything that the TIGER import has asserted about the road in
question, in particular:

  - The road geometry, which is often comically bad. I generally also
add the bridges and culverts (and get lost mapping streams back up into
the mountains) though I've never considered this necessary for deleting
tiger:reviewed=no. (Also, over time I've gotten a little bolder about
simply deleting the roads that don't seem to correspond to anything on
leaf-off satellite, Bing streetside, or the county maps -- especially
the ones that look like spiky stick drawings. I feel that leaving a road
I genuinely believe to be fictional is a disservice to the map.)

  - The highway=* classification -- most common problem I see here is
highway=residential for tracks, driveways, and other service roads (more
rarely residential for what should be secondary or tertiary.)

  - The access -- somewhat common to find a pubic road imported with
access=private, so if I suspect this I'll leave the tiger:reviewed=no
tag until access can be confirmed, and add a note or fixme. (It's also
quite common to find driveways imported as access=private. When
surveying, I tend to remove the private tag if the driveway isn't gated
or signed private, since access=private will prevent routing to the
house at the end of the driveway, sometimes even ending the route on a
different residential road that's physically closer to the house than
the road the driveway's connected to.)

  - The road name -- and this can be a real mess because road signs,
addresses, government maps, and TIGER often disagree. Even two road
signs a mile apart may disagree. I do my best to set name=* and
alt_name=*, and I'll often leave the extra fields from the TIGER import
(name_1, tiger:basename, etc) if they have other variations. Kevin, if
you can give some more details on your name-matching process using E911
and DOT maps, I'd love to learn.

Creating/repairing highway route relations is a special case of name
fixing I guess. I've been lax about removing TIGER's name=State Highway
X etc tags; I'll try to do better there.

Regarding the surface values, at some point Richard Fairhurst made the
specific request that adding surface=* should be part of the TIGER
cleanup, when possible. Personally I only tend to do it when the surface
can be clearly observed and the road in question falls somewhere in the
gap between paved residential and unpaved track. And I also don't
consider this necessary for deleting tiger:reviewed=no.

...Related sidebar -- What are the best practices for setting
addr:street=* for addresses on highway routes? Along NY-28 for instance,
business addresses will be signed with some variation of "Route 28" or
"Highway 28", usually prefixed with "State" or "NY".  Residential
addresses will usually just have the house number. I haven't found any
clear guidance or consistent tagging practice here.

Jason

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20200712/f7f8769b/attachment.htm>


More information about the Talk-us mailing list