TIGER fixup

From OpenStreetMap Wiki
(Redirected from TIGER desert)
Jump to: navigation, search
WikiProject United States openstreetmap.us Projects Tags TIGER fixup
US Local chapter Users Data Imports Resources Category
Lightbulb.png TIGER 2012 is available for comparison, as a tile layer, and is typically superior to any unfixed older TIGER data in OSM.
Lightbulb.png The TIGER Edited Map ([1]) shows which areas of data have been changed since the initial import

The TIGER data are of variable quality. Here are some common problems that you'd find in unedited TIGER data:

  • entire roads that no longer or never did exist (especially in rural areas)
  • Only one way to represent a two-carriageway highway
  • Wrong road classifications (especially in rural areas)
  • roads that cross on different levels get intersection nodes
  • railway and roads moved so they intersect when they do not
  • whole subdivisions misplaced by dozens of meters
  • missing subdivisions (may be available to be traced with TIGER_2012 Imagery)
  • highways missing a "One Way" Tag
  • bridges and tunnels are not added
  • "braided" streets
  • missing ref on highways
  • using obscure or outdated names for Places rather than common names

Each way has a tiger:reviewed=no tag. The practice to remove this tag varies. see: this discussion on newbies mailing list for some discussion. Some users remove it after armchair mapping an area, others not until they do a personal, on the ground, survey.

The term "TIGER desert" has been coined referring to areas where little or no editing of TIGER imported data has been done for an extended period of time — or at all since import (period of inactivity measured in years).

Road Alignment

In general, alignment is pretty poor
Probably the biggest problem with TIGER data is that they were originally created for taking census surveys, not for making maps. This means that in many areas the roads aren't aligned with their true position, but if you imagine the TIGER database as a sketch for people walking around with clipboards, then it has all the roads laid out roughly correctly. The work then needed is to drag everything into its correct place and add more nodes to smooth out curves. See TIGER fixup/Alignment for HOWTO information on re-aligning roads, and a list of counties with widespread problems.

Wrong highway classifications

Most highways imported from TIGER were classified as highway=residential. In many instances, highway=residential is not the most appropriate tag. This problem is most common in areas with little editing since the TIGER data were imported and/or in rural areas.

OSM typically has four highway classifications (tertiary, unclassified, residential, track) where TIGER had one class (A40 group, especially A41). Thus many of these imported highway=residential ways should instead be:

Tertiary and unclassified roads are assumed to be paved in Western countries, so make sure to add a surface=* tag if the road is in fact unpaved.

Connectivity along county borders

Fixing county border connectivity problems at this HOWTO
The majority of county border connectivity errors have been fixed, including partially overlapping ways. In many instances, ways crossing county boundaries may stop at one side of the county border, when they continue into an adjacent county, or the gaps in the ways at the county border are so large that simple connectivity checks will not identify them. Visual inspection of county borders may still identify these sorts of errors.

There are particularly serious data problems along county lines, where different TIGER datasets were imported. County lines show up as administrative borders at zoom level 9 (e.g.). The main problem is that each county import didn't "know" about adjacent neighbours, so created duplicate nodes where the end of a road meets the start of its continuation (or "twin brother") from the next county. This even affects major highways so prevents most Routing between counties. Other problems include overlapping roads and border ways - for example, when a road runs along the border it will have been uploaded twice in exactly the same place - once from one county and once from the other.

There is a HOWTO available on fixing this problem.


Routing from a freeway onto an overpass - HOWTO fix this
The original TIGER data have no concept of whether two roads are connected or not, so that was all synthesised during the import into OpenStreetMap. For grade-separated roads this causes problems when routing. A HOWTO article is available on fixing this problem.

Divided roads

Many Interstates and other major roads only have one carriageway in TIGER, when in reality, they have two one-way carriageways ("divided road" or "dual carriageway"). A HOWTO article is available on fixing this problem.

Simplify ways with excessive nodes

Before running 'simplify way' in josm on a TIGER-Imported way
After running 'simplify way' in josm on a TIGER-Imported way

Highways in some rural areas were imported with many nodes. JOSM and Potlatch users can use the Simplify Way Feature to eliminate excessive nodes without losing any characteristics of the road. This feature is very useful in rural areas where roads tend to be straight. After using on a way, be sure to verify that no characteristics of the road were lost.

Braided streets

Example of braided street problem

Some streets in the TIGER upload were "braided", where the one-way highways of a dual carriageway criss-cross back and forth across each other, sometimes going in the wrong direction for the side of the street on which they run. The ones that were identified shortly after the upload were fixed using the osm-unbraid tools. The following counties are known to have remaining braided roads:

  • Polk County, Florida (mostly clean now)

Please Note: "braided" refers only to the situation where the ways are indiscriminately on the wrong side of the street for their direction. It is a reasonable and well-used technique to bring the ways of dual carriageways back to a single point at intersections to facilitate and simplify the mapping of control devices and turn restrictions. In other words, you may see situations that look similar to the picture, but are correct because the ways both stay on the correct side of the center divider between intersections. In all cases, like any other edits in OSM, be careful not to lose any existing correct data contributed by your fellow mappers.

Useful tools

See the TIGER 2012 page for details of a tile set which you can load in as a background while editing.

The TIGER Edited Map ([2]) shows which areas of data have been changed since the initial import

osm.cycle.travel shows unreviewed rural residential roads (in blue) that have been ridden over by Strava users, implying that they should be reviewed.

MapRoulette (maproulette.org) is a micro-tasking/gamification system with several challenges related to TIGER fixup.

In Potlatch 1, you can choose to "Highlight unchanged TIGER data" (in the options dialogue). This will give a pink border to roads which have not been changed since the initial import. Again in Potlatch 1, if you hover the pointer over a way where the tiger:reviewed tag is set to "aerial", the name will be followed by a *.

In JOSM, unreviewed TIGER data (that is, with tiger:reviewed=no) used to show up with a yellow-ish background. It no longer does.

Other handy fixup plugins include the WayDownloaderPlugin

You can jump from one slippy map to another, keeping the same position, with the mapJumper bookmarklet. Two ways :

  • Just drag the mJLite bookmarklet on your tool/favorite/bookmark.
On a well formed url map (with lat, lon, zoom), by clicking on your mJLite bookmark, you will be led (in a new window) to the mapJumper hub page where are proposed survey links.
  • Just compose your bookmarklet here, with the preset import or survey. Paste the code in a new bookmark. Name it, let's say, mJ[Tiger] (you can get other mapJumpers).
On a well formed url map (with lat, lon, zoom), by chosing the mJ[Tiger] bookmark, give the shortkey for the Tiger fixup service (tf). Validate and jump.

These old tools were useful once, but are no longer available (could be inspiration for new developments?)

  • TIGER Battlegrid used to show yellow patches where most fixup work is needed, based on comparison with TIGER 2012.
  • Duplicate nodes map used to show a map of this specific problem, which is a good indicator of problematic imported data.
  • The To-fix tool used to have a have a 'tigerdelta' task queue (was running here) which looked at comparison issues with the TIGER 2014 dataset. The Mapbox data team (e.g. those based in peru) were using this in 2015 to fix many errors. See the commentary on this issue: https://github.com/mapbox/mapping/issues/100. It's not clear if there are more detectable deltas to queue up, or if they fixed them all.
  • Waychains analysis ([3]) is no longer updated so not really useful. It used to reveal some bugs in the major interstate connectivity (hopefully mostly fixed now anyway)

Automated Edits

In December of 2012, emacsen ran an automated bot to clean up the TIGER abbreviated streetnames on the Eastern half of the US, along with cleaning up some of the spacing on the Western half of the US. Code at https://github.com/emacsen/tiger-expansion

In 2013-02 jjaf.de did wikipedia=* correction in Mechanical Edits/jjaf.de#WIWOSM TIGER import fixup.

See also