[Talk-us] Importing highway surface tags

Josh Doe josh at joshdoe.com
Fri Dec 21 02:33:37 GMT 2012


On Thu, Dec 20, 2012 at 5:03 PM, Adam Franco <adamfranco at gmail.com> wrote:

> Hello All,
>
> I'm pretty new to the OSM community and have gotten involved through a
> hobby project analyzing road geometries<https://github.com/adamfranco/curvature/wiki>.
> I live in Vermont, a state where more than half of the road-miles are
> unpaved graded-gravel. For the benefit of my project and hopefully the OSM
> community I have been working outward from my home tagging highway
> surfaces <http://www.openstreetmap.org/user/Adam%20Franco/edits> as
> gravel, asphalt, or concrete based on direct observation as well as looking
> at high-resolution aerial imagery where available.
>

Welcome! I lived in Vermont for many years, and can certainly attest to the
poor nature of some of the roads. Sometimes the asphalt roads were in worse
shape than the dirt ones, except for those washboard roads.


> What I am interested in doing is bulk-tagging the surface of roads based
> on public data where no surface tag has been manually set. Before I begin
> this project (and pester my department of transportation for data) I wanted
> to check with the community about the permissibility and feasibility of
> this effort.
>

Glad to see you're interested, and especially glad that you came to the
list to get the community involved.


> A few questions:
>
> * Has anyone located a good source for state or national road surface
> data? The TIGER data doesn't seem to include surface information as far as
> I can tell.
>

I've seen condition tagging on many Massachusetts roads, which I believe
came from the MassGIS import. I believe that was a wholesale import and
replacement of existing data at an early point, rather than data being
conflated.

* While the devil is always in the details, I'm assuming that a relatively
> safe import could be done by only adding tags to ways that don't have
> surface tags already. Are there other considerations I should be thinking
> about related to not worsening the OSM data?
>

What you would do is automatically add the tag if none exist, and then flag
any conflicting items for later manual review. However by far your biggest
issue is going to be conflation of this data. First, you'll need to match
roads between the two datasets, which is a significant challenge. The
bigger challenge would be to automatically split the OSM roads at the
points where the surface data changes in your dataset. I've never seen
anyone do this. The more doable approach is using the dataset as a second
layer with JOSM or Potlatch to manually conflate. You might even be able to
write a JOSM plugin which would help with this, perhaps extending the
conflation plugin I started a while ago. Also Potlatch has some
functionality thanks to Cyclestreets to help with this sort of thing:
http://wiki.openstreetmap.org/wiki/Potlatch_2_merging_tool


> * Is this a project that the OSM community in Vermont, the broader region,
> or nationally (assuming data is available) would support? I'd rather not do
> a lot of work to prepare it if there is no desire for inclusion in the data
> set.
>

I would say absolutely, but if you truly want automatic bulk conflation,
you've got a long hard road in front of you unless you have some mad
skills. Even then you'd need to demonstrate it to the community that it
works without bugs. Using JOSM or Potlatch for manual conflation seems like
an easier path, if not as exciting or fast.

Also, think about how to show this data. Consider creating a custom map
rendering that shows these surface tags, as it will be good motivation to
contributors.

I'm sure others will chime in with more thoughts.
-Josh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20121220/85e40294/attachment.html>


More information about the Talk-us mailing list