[Talk-ca] Canvec 10 and landcover issues

Paul Norman penorman at mac.com
Fri Oct 19 19:45:22 GMT 2012


> From: Harald Kliems [mailto:kliems at gmail.com]
> Sent: Friday, October 19, 2012 11:04 AM
> To: Talk-CA OpenStreetMap
> Subject: [Talk-ca] Canvec 10 and landcover issues
> 
> Hi everyone,
> I've done some OSMInspector debugging of areas around Montreal and I've
> come across a number of newly imported natural=wetland areas, sourced
> from Canvec 10. that are clearly wrong. This, for example,
> http://www.openstreetmap.org/?lat=45.69514&lon=-
> 73.90455&zoom=17&layers=M
> is a subdivision, not wetland or wood. If you're importing Canvec data
> could you please make sure to do some plausibility checks, based on
> aerial imagery or road layout, especially in populated areas? I'm not
> sure how old the imported data is, but some areas supposedly covered by
> woods or wetlands look like they've been developed for quite some time.

Yes - one of the frequent issues is that the different thematic "layers"
(e.g. landcovers, addresses, roads, buildings, etc) are different ages and
some are very old. In BC the buildings information is horrendously old while
frequently the water and forest information doesn't agree - leading to
CanVec saying there are trees in the ocean. The recent releases are better
in this regards and I believe some zipfiles include information on the age
of the different feature data included, but the problem of areas being
described both as built up by the roads data and as forest by the other data
is still frequently an issue. It's like your mixing pictures of what the
area was like at different times, so you get confusing results.




More information about the Talk-ca mailing list