[Talk-ca] NRN Fort McMurray area data

Steve Singer ssinger_pg at sympatico.ca
Sun Jan 18 06:55:47 GMT 2009


On Sat, 17 Jan 2009, James Ewen wrote:

> As reported previously, sometimes the RoadMatcher is thinking roads
> that intersect at 90 degrees are a match, and is leaving them out.
> There are a number of these that I'll have to go back and insert.

This is expected.  When you run roadmatcher on a dataset it divides the 
roads into three categories.  'Matched', 'Standalone' and 'unknown'. The 
unknown ones are roads where roadmatcher is unsure if there is a match or 
not.  I am only importing 'standalone' roads since we seemed to think that 
it was better to leave out the unknowns vs having many duplicate roads.
It would be nice if it could a better job matching.


> Here's where OSM data is going to surpass GeoBase. The question
> remains though, how to teach the RoadMatcher to recognize that these
> two circular routes are the same roadway?

In this case RoadMatcher actually did detect that they were the same road, 
that should have been excluded from the output. However there is a bug in 
the exclusion code my geobase2osm.py.

What happened was that road matcher matched part of the road, but it split 
the NRN segment. One part was matched, the other part was standalone.  My 
script incorrectly declared the entire segment as standalone (which is 
incorrect, it should have been excluded from the import).  I've fixed the 
script.

It looks like two geobase road segements suffered from this.

6972a46cf13841a88c89ff40c49cd498
948359a6ff024551aa1a07cec21c818b

I will manually clean this up tomorrow sometime.

Thanks for spotting this.


>
> http://www.openstreetmap.org/?lat=56.65477&lon=-111.32785&zoom=17&layers=B000FTF
>
> James
> VE6SRV
>
> _______________________________________________
> Talk-ca mailing list
> Talk-ca at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-ca
>





More information about the Talk-ca mailing list