OpenStreetMap
Sometimes I wonder if those TIGER surveyors were just having a grand old time making up things as they drove past and casually looked out of the window of their cars…
This whole area needs lots of work - if you want to help out, that would be great!
(I have no personal or other connection to the area, but got dropped there by the MapRoulette ‘Ways Needing Smoothing’ challenge - freshly cleaned up, so most cases you get should now be real things to fix!
Discussion
Comment from Linutux on 23 August 2014 at 00:17
I just read you blog post and looked at it myself. It’s just crazy! It’s not just one single street at one single place. The map in the whole area is crap. Oh my gosh!
Comment from aseerel4c26 on 23 August 2014 at 00:35
or bing does not show reality! (in case you only rely on bing) ;-)
Comment from mvexel on 23 August 2014 at 00:46
Bing may be out of date or misaligned, but in this case I’d wager a bet that TIGER is more wrong than Bing!
We also have the JOSM ‘TIGER 2012’ overlay as an additional reality check. This is also TIGER data but from after the big TIGER improvement program MFTAIP that ran between roughly 2004 and 2008:
Comment from jumbanho on 23 August 2014 at 00:46
Yeah, sadly, the whole of Appalachia is like that, except for the few places where mappers live or visit.
Strava traces/slide, can help in the alignment if there are a lot of bikers in the area.
Comment from aseerel4c26 on 23 August 2014 at 01:45
@mvexel: yeah, I just wanted to mention the possiblity (not more) of even a made-up bing aerial image.
Comment from robert on 24 August 2014 at 09:35
Interesting thing is that it is actually topologically correct.
Comment from Pieren on 26 August 2014 at 12:38
It’s seems the TIGER 2012 dataset is much better than first TIGER imported in OSM. Is something planned by the US community about this dataset ? a new import ? conflation ?
Comment from mvexel on 26 August 2014 at 22:00
robert - that is what the T stands for in TIGER :)
Pieren - highly unlikely, for a number of reasons. The most important one is that the unique identifiers on the TIGER features have changed since the original import. Also, the original TIGER IDs have been removed from many objects in OSM. Both these things make it very hard to match newer TIGER to what is in OSM now reliably. Any re-import attempt would be far a smaller area, introducing its own set of challenges related to conflation and connecting up the newly imported data with the existing data at the boundaries of the import area. I think there is more to be gained from tools like the Battle Grid and the TIGER 2012/3 layers available in JOSM and iD.
Comment from Jonathan ZHAO on 1 September 2014 at 04:23
I found it’s very common last year when we do battle grid. It’s very weird. The transformation has no obviously rule. I thought Navteq and TeleAtlas are also originated from Tiger. But there is no this kind of thing in Navteq or TeleAtlas.
Comment from nfgusedautoparts on 7 September 2014 at 00:55
as near as i can tell, a lot of the early tiger stuff was hand drawn maps that was then entered in more-or-less the correct place without any reference to imagery. the stuff is topologically correct but off by large amounts in places. the worst i’ve seen is in WV which, of course, is part of Appalachia as jumbanho points out. best approach would be to use the tiger 2012/3 layers as reference for moving things to the right place. maybe we could find a way to load up maproulette with areas that have particularly bad correspondence?