Shift of TIGER data

All the streets in State if Hawaii, US seems to be shifted from the actual positions.

Fo example, http://www.openstreetmap.org/edit.html?lat=19.7237&lon=-155.0423&zoom=14 shows that the streets that might have been taken from TIGER data is shifted from areas defined from the aerial photographs. As looking at a measurement by Nokia N-810 (sorry, I don’t know which coordinate system it uses.), it seems that the aerial photographs are correct while the streets are shifted from the actual positions. Is there a way that I, or somebody, can shift all the roads in State of Hawaii so that it matches to the aerial photographs and, hopefully, the actual positions of the physical streets.

BTW, I found this map while setting up maemo mapper on Nokia N-810. I appreciate the great project!

You can check the history of the ways by pressing ‘H’ in Potlatch (the online OSM editor). If that shows that someone has (accidentally) shifted the data then a mass reverse of that edit should be possible (but requires some effort by OSM admins).

Hmm, I’ve checked the history and all the ways I’ve checked have not been edited since the import. So either the imported ways are wrongly positioned or the aerial photography is offset. The latter is common so are you sure that your tracks support the shifting of the ways?

A GPS usually uses WGS84 so be sure to set that on you Nokia N810 (really nice piece of kit btw :D)

This could be helpful information on how to deal with this:

Data from OSM, via JOSM:
Kahului 20,8921409, -156,4306475
Start run way 20,8907492, -156,4367376
End 20,9055018,-156, 4257696

Data from/: www.airport-data.com (FAA source)
156-25-49.6483W/20-53-55.1390N
-156.430458/20.898650 (Estimated)

Most streets have been drawn by user: DaveHansen as a bulk up-load from tiger.

Shift look of the order of 600 m northerly (streets drawn 600 m to the north) or slightly less in NW direction

Thanks for the information. I will walk around in Hilo to confirm actual positions of the streets, hopefully next weekend.

I have uploaded a number of tracks. I am now trying to find time to check shifts between the tiger data. I also found a related page at http://wiki.openstreetmap.org/index.php/TIGER_fixup

Yeah I’ve just added a section to that TIGER fixup page (and a related discussion on the talk page) about TIGER offsets.

If there is a problem with some kind of graded distortion of the TIGER data, then hawaii would be a very good “fix point” to help investigating it, because it is probably the Westernmost extremity of the TIGER data, which might mean the offset is at its most severe

Certainly looks quite severe compared the Yahoo imgery. Of course we have to keep in mind that we may be observing offset inaccuracy in the Yahoo imagery, but that would be interesting thing to discover in itself. GPS tracks there will help us see what’s really going on.

BTW aren’t the US gov. supposed to make a new TIGER release soon, with the census pending and all?

This thread answers a related question of mine – If there is a discrepancy between a GPS track and aerial imagery, should the map elements be shifted to match the aerial imagery? The answer appears to be “no”; the better approach is to collect a new GPS track and upload this, thereby either confirming the prior GPS track or confirming the aerial imagery (or presenting a third discrepant data set … one hopes not). The reason - positioning of aerial imagery may need correction and the GPS track data should be taken as definitive. Correct interpretation? --ceyockey

(follow-up)

However, I see now a page in the Wiki suggesting that alignment to aerial imagery is encouraged ( http://wiki.openstreetmap.org/wiki/TIGER_fixup/Alignment ). Clarification on this would be appreciated. --ceyockey

If you have information gathered on site (GPX tracks) that tell you that the imagery is off then why align to the imagery? Most important is that the data is correct ‘on the ground’. BTW, did you know you can adjust the alignment of the aerial imagery in Potlatch?

The article http://wiki.openstreetmap.org/wiki/TIGER_fixup/Alignment would seem to imply that TIGER tracks could be (in general) considered inferior to aerial imagery. I’m coming at this from the point of view of someone who has not yet invested in a GPS receiver and I am becoming convinced that a person without a GPS receiver should neither adjust tracks nor revise alignment of aerial images (no, I wasn’t aware you could revise imagery alignment). Is this a reasonable and generally supported conclusion? If yes, then I wonder whether the “alignment” article should be revised to de-emphasize the value of the action depicted in favor of how best to deal with an independent track added into the mix of a TIGER track and an aerial image. --ceyockey

mapping tools: have not yet acquired

I was just on holidays in Hawaii and apart from most of the roads on Oahu, the other islands are shifted as the previous posts implied. I have uploaded two GPS traces from Maui (Gisteq Phototrackr CD111, MTK chipset) and if you place them over the map you can clearly see the shift of all the roads.
I think the fix will require someone with uber access to make a bulk shift in the backend as the thought of making a changeset to move all the roads on the remaining islands just gives me the willies.
I have a few more traces to upload, from Hawaii and Kuaui, that will hopefully go up in the next few days as I clean them up. Same problem on those islands as on Maui.

This came up in the mailing lists recently. There are some follow-up comments addressing the degree of damage at
http://comments.gmane.org/gmane.comp.gis.openstreetmap.region.us/1828, http://article.gmane.org/gmane.comp.gis.openstreetmap.region.us/1855 and http://article.gmane.org/gmane.comp.gis.openstreetmap.region.us/1856, with an explanation at http://article.gmane.org/gmane.comp.gis.openstreetmap.region.us/1857.