[OSM-talk] Adding vmap0 road data to OSM

Christopher Schmidt crschmidt at metacarta.com
Sun Jan 6 01:09:39 GMT 2008


On Sun, Jan 06, 2008 at 12:27:25AM +0000, Bruce Cowan wrote:
> 
> On Sat, 2008-01-05 at 12:48 -0500, Will Harrison wrote:
> 
> > This would benefit the custom scenery project, since once all the
> > vmap0 road data have been added to OSM the intent is to use OSM data
> > for the scenery. But I think this could also be very helpful to the
> > OSM project, as vmap0 contains data for the entire world and it would
> > give OSM at least rough coverage of the entire globe. Is anyone
> > interested in helping? 
> 
> I can see why you'd want to add vmap0 data in places that have no
> mapping, but IMHO such bad data should not be added to OSM.
> 
> I think having no roads in Flightgear in these areas would be better
> than bad roads, saying as this bad data would have to be in OSM.

I disagree.

Many of the commercial providers use vmap0 as a 'backfill' for areas
that are not well mapped via commercial means. In fact, some of the AND
data that has been provided for China is exactly this: vmap0 data.

The data is low resolution -- the map itself was only ever designed for
1:1M scale, and even then it's up to 30 years old, so there are some
obvious problems with it for that case. Still, it's an obvious way for
users to quickly get an idea that they may be able to contribute: in
areas where there is no existing data, vmap0 can provide a reason for
people to participate.

I think that the important thing in this case is not "Do we import it or
do we not import it" -- there's no reason that having low resolution
data hurts OSM so long as it is properly labelled. Instead the question
that I see is "How do we make sure that people understand the data is
low resolution, and how do we make sure that we don't interfere with any
existing data?"

The answer to the latter seems relatively simple: limit vmap0 data
imports to areas where there is no existing coverage nearby. "Nearby" is
relative, but if you only import vmap0 data if there is no OSM data
within, say, 100km, you're unlikely to bump into any active mapping
going on (in my experience). You'd get a 'ring' around any hotspots that
would be protected.

I think that those of us in the US have discovered that even when TIGER
is woefully incorrect -- it's easier to correct than it is to create
anew. (I could be wrong on this, but I'm hoping others can back me up.)
So long as vmap0 *doesn't hurt existing mapping*, and is properly tagged
-- perhaps even such that it's not rendered in the main maps at all --
then I don't think OSM is hurt by having the data included.

Regards,
-- 
Christopher Schmidt
MetaCarta




More information about the talk mailing list