[Talk-us] Rural US: Correcting Original TIGER Imported Ways

Richard Fairhurst richard at systemed.net
Mon Feb 19 13:28:27 UTC 2018

Great to see so much attention being paid to rural TIGER fixup. The majority
of my editing these days is that, and it's a massive but rewarding job.

I put together a view a while back which superimposes unreviewed rural
residentials onto the Strava heatmap. The idea is that you look for
unobscured roads, then go in and fix them. It's at
http://osm.cycle.travel/unreviewed.html .

The "unreviewed" data comes from the cycle.travel rendering database, so
there are a few optimisations (for example, it knows that all State Routes
in NC are paved) and the update frequency is not fast (every month or two).
But it's a good way of finding roads that are regularly used (by cyclists,
at least) and are currently unreviewed.

The one thing I would stress above all is that a surface= key (or
equivalent) is crucial to denote unpaved roads - especially for cyclists and
non-4x4 motorists.

In the developed world in OSM, highway=unclassified and highway=residential
are assumed to be paved roads in the absence of other information. So in
(say) NY State, if you saw a highway=unclassified or highway=residential
without an explicit surface tag, you'd assume it was almost certainly paved.
Now in Kansas most roads are unpaved, but we can't expect people to do
state-specific parsing - that way lies madness. Indeed, very few consumers
do even country-specific parsing.

So absolutely do change rural residentials to highway=unclassified, but add
a surface= tag while you're there. A simple surface=unpaved is better than
nothing, though obviously if you can be more precise with =gravel,
=compacted, =dirt or whatever, that's great. I'd ask people setting up
MapRoulette challenges and the like to incorporate this into their
instructions - thank you!

Having good paved/unpaved information will be a massive boost for OSM in
comparison to other map providers. We're already partway there. As an
example, try asking Google Maps for bike directions from SF to NYC. It sends
you down some really, really unsuitable tracks and I'm not entirely
convinced you'd survive the journey. By contrast, cycle.travel (using OSM
data) gets it pretty much right: occasionally it takes a gravel road
unnecessarily but it's pretty much always rideable.

It would be great if we could become the best map of the rural US just as we
are for much of the rest of the world.


Sent from: http://gis.19327.n8.nabble.com/USA-f5284732.html

More information about the Talk-us mailing list