[Talk-ca] OSM Geobase import: giving a try
Steve Singer
ssinger_pg at sympatico.ca
Mon Feb 16 01:19:52 GMT 2009
On Sun, 15 Feb 2009, Frank Steggink wrote:
> Hi Steve,
>
> In what way would it limit us? When we'll receive a new dataset from Geobase?
> Or do you hint towards other datasets which are linked to the NID? In that
> case that additional data can't be linked to existing database, because that
> doesn't contain the NID attributes.
Mostly when we receive geobase updates.
Consider what should happen when as geobase releases road name data for
provinces that have already been imported. We will just need a script that
finds the OSM way that is tagged with a particular geobase:uuid and add in
the name tags from the geobase update. If we don't have the uuid in the OSM
data we'd need to figure out some other method.
Also we think about what happens when a road crosses two of the artificial
tiles we dividing the import process across, if each road as a unique NRN
id you can easily excluded the nids already imported in the adjacent tile
from the current one.
I only reason to not include it would be to save a few bytes of space.
> I actually wanted to use PostgreSQL/PostGIS, but it seems that no OSM data
> can be exported from it. At least not when OSM data was previously imported
> through osm2pgsql. Maybe it will work to keep the Geobase data in Postgis,
> export as OSM, and then apply some postprocessing (fixing in JOSM).
I've been exporting OSM data from PostGIS without issues.
I've added some more detailed steps at
http://wiki.openstreetmap.org/wiki/Geobase_NRN_-_OSM_Map_Feature
I export the OSM data from PostGIS as a shapefile, I only use this as an
input to road matcher (I only need a few fields). I don't think the data
from osm2pgsql can be used to regenerate .osm files though.
> I'm going to try to generate buffers around the imported road data from OSM,
> and will use it to clip away any Geobase data (NRN) which is entirely within
> a certain distance from the OSM data. And then export it to OSM. I'll need to
> work it out further, using my test area. I think that a buffer of 10 or 20
> meter is enough.
I thought about using buffers initially but then I came across road matcher.
10-20 meters tends to work for OSM data from GPS traces but the roads from
low-res imagery (often of highways in rural areas) often can be off by 100 meters.
> One final question: what is the accuracy of the Geobase data? Is it worse,
> comparable, or better than typical GPS tracks? The roads I've drawn so far
> match the Yahoo imagery quite well, although usually there is a slight offset
> of a few meters.
The Geobase data comes from different sources. Some of it is from GPS,
other data is from Vector files and some from imagery. My guess is that all
of it is more accurate than the low-res pictures Yahoo provides in Northern
Alberta.
My guess is that most of the geobase GPS data is better/comparable to the
typical OSM mapper GPS data, the vector data doesn't seem to be as good (and
I've found instances in Ontario where the geobase vector data shows
roads/intersections that don't exist). The geobase documentation lists
accuracy/precision somewhere but I can't recall what they claim.
>
> Regards,
>
> Frank
>
Steve
More information about the Talk-ca
mailing list