Issues with government trail data

I posted this to the mailing list as well, but since this announcement came on the forums I posting here as well.

I find it very hypocritical that land managers are complaining about OpenStreetMap trail data when their own data is full of inaccuracies. Perhaps Utah is different, but just around the area where I live in Colorado I have found hundreds of errors in government trail data, including cases of:

  1. Trails being mislocated by up to a third of a mile.

  2. Trails that do not exist, or are in such poor shape as to not be considered trails, in the government data.

  3. Official trails (as evidenced by signs and other government documents) that do exist, but which are not in the government data, e.g.

  4. Trails whose on the ground signage as to name, and number do not match the government data. e.g.

  5. Trails whose on the ground signage as to allowed usage do not match the government data.

  6. Duplicate trails e.g.

  7. Topological issues in government trail data, e.g. two trails that are connected in reality, but are not in the government data.

  8. Government data including trails that are posted “no trespassing”

In some cases the government trail data is over ten years out of date!


Unless we are very careful, attempts to “improve” OSM with government data is likely to lead to a degradation of the quality of OSM!

There seems to be a reluctance on the part of land managers to admit they have a problem, or do anything about it.

The problem is so bad that a local television station did an investigative report on it:

I gave a presentation at GIS in the Rockies on the matter:


All good points, Mike. I split this out to a separate topic for conversation so it isn’t lost in the Updates from OSM US thread.

The main goal of the trails stewardship initiative is to increase coverage of trails with either an operator= tag (indicating an officially designated trail) or an informal=yes tag (indicating an informal,not officially designated trail) as well as appropriate access tags. Government datasets can help with this, but definitely shouldn’t be taken as a perfect, authoritative source of truth. Comparison of multiple data sources and field surveys is always best. If geometry for a certain trail in OSM is more accurate, then we certainly shouldn’t replace it with less accurate geometry from a government dataset.


I have seen all of the issues you mention in data sets from USFS, NPS, BLM, county GIS data, and of course the Census Bureau (TIGER). All the data from these sources needs to be taken with a grain of salt. Sometimes the data is quite good. Often it is not.

As @ezekielf says, the best approach is to verify the data with other reliable sources such as aerial imagery, street-level imagery, GPS tracks, and direct first-person surveys.

The point about official trails that are not in the government data sets is an important one. I recently visited the Moss Wash OHV Area near Kingman, Arizona. Although BLM does publish a map of the area and has trail signage and map kiosks on site, none of the trails in the area are present in the BLM Ground Transportation Linear Features data set.

The fact that a trail is not in the official data set is not a clear indication that the trail is “informal” or otherwise restricted from public use.

Anyone using a government data set must be aware of its limitations.


@ezekielf Thanks for the reply.

I wouldn’t assume that if a trail isn’t in the government data that it is “informal”, their data is missing some newer trails (but documentation exists that they themselves authorized the construction of, as well as trails that have existed for decades). Also, it is possible that some other government entity, e.g. a city, is maintaining the trail, and hense it isn’t in the USFS/BLM/etc. data. But yes of we know for a fact that a trail is “informal”, I am all for tagging it as such. The more information the better.

I have found cases where these are not accurage as well (don’t match the signs along the trail).

  • Field survey is best, yes. - take geotagged pictures of all signs
  • Strava Global Heatmap
  • USGS 3D Elevation Program data - I was told by the USGS that horizontal accuracy is “better than one meter”
  • GPS traces in OSM - keep in mind that old traces may no longer be accurate if the trail has been rerouted.
  • Agency’s own online documentation (often doesn’t match their GIS data)
  • Street level imagery
  • Aerial/Satellite imagery

General comment - this whole process, that is conflating in allowed usage, etc, would be a lot easier if the government data was spatially accurate.

No, neither would I if that was the only available source I could find and I was unsure of its accuracy. However, if I’m referencing several data sources and none of them show any evidence of an officially designated trail then I feel like it is possible to be reasonably confident that the trail is informal. If this is proven wrong by field survey in the future, then the informal=yes tag can be removed. If I was armchair mapping without local knowledge and the trail was already tagged with note=Official data for this area is out of date and does not include this trail. It has been verified by field survey as of {DATE}, then I would trust the previous mapper and not add informal=yes.

I’d consider that land masses are vast and government workers doing data collection are not as vast in numbers to be able to keep up with the ever growing trail networks out there. As an OSM editor you can put trails in that are new and edit what is old, then inform the government about the updates.


Hi @tekim! I appreciated watching your video about the errors in gov trail data that you’ve discovered. Definitely important to keep those in mind while mapping and for our tasking manager instructions.

I want to emphasize what @ezekielf is saying: The efforts of the Trails Working Group at this point is to determine operators of trails to the best of our ability using multiple data sources and keeping the best geometry. In fact, the Utah campaign really is an attributes-only mapping project with the goal of adding tags beyond highway=path so the data can be more useful.

Another effort we’re working on, that may be up your alley, is our Digital Trail Stewards idea where locals who CAN get out and do field surveys are heavily involved in all the trail mapping campaigns (you and I can both do this for our Colorado stomping grounds).

This link may have been posted earlier, but there’s a lot of information here if you haven’t seen it and want to dig in, or just reach out to me!: Trails Stewardship Initiative | OpenStreetMap US

Thanks so much for watching and for being open to what might be new information (even I was surprised by the level of error in the Dillon Ranger District). While you and I can use OSM or a third party app when we are out on the trails, certain government folks, such as wildland fire fighters (e.g. some people in my family), are provided the “official” data, so getting the government to make corrections to their data is of particular interst to me (as is keeping errors in their data out of OSM).

While more tags would be great, there are cases where the attributes (tags) in the government GIS data are wrong (or at least don’t match their own signs and written documentation). Also, in some cases the government data is missing trails that are clearly official trails, again as evidenced by their own signs and written documentation. In those cases a remote mapper would probably tag the trail as informal=yes, which would be wrong. Some of these trails have existed for decades, while others have been approved by the land manger and constructed by the land manager, volunteers and contractors over the past five to ten years and is not reflected in the GIS data.

Yes, we need more of this. I have a lot of trails on my list that require an in person visit because they are so remote that they have no Strava “heat”, don’t show in the 3D Elevation Data, and don’t show in imagery (but they are in the governement data). I suspect some don’t really exist.

I am going to review it again. Thanks.