> From: Dan Charrois [mailto:d...@syz.com] > Subject: Re: [Talk-ca] Internal CanVec conflicts > > Usually, in remote areas of the north that I've dealt with, there is > often little else already there than the Landsat lakes. And usually, in > a given tile, there is usually just a handful of lakes. > > In pretty much every case I've dealt with so far, I've replaced the > Landsat lakes with Canvec data. Landsat lake outlines have a much lower > resolution, and being derived by an automatic process themselves, are > subject to the errors associated with that. But I wouldn't necessary > erase all the Landsat data from a tile without checking first to make > sure that there will be Canvec data replacing it (I always work with my > Canvec data in a separate layer and merge things in one "feature" at a > time as it's checked). Using Bing imagery may be a good idea to check > any issues where Landsat data may exist and Canvec doesn't - even low > resolution Bing imagery is usually sufficient for the Landsat lakes. I > have yet to encounter a place where there is a Landsat lake and not a > corresponding Canvec one of roughly the same shape, but it could happen. > > I have yet to find a situation where the Landsat lake data is better > than the Canvec data.
The coastlines in BC were similar and what you have to watch out for is areas where someone has refined the traces but hasn't touched the source tag. I ended up using JOSM to search for nodes that were part of the coastline and were not version 1 nodes from the person who imported the coastline. Even though not all of this stuff was imported from a dedicated account (most of it was done many years ago) it's not too hard to find it with search strings because the importer didn't do anything in the area except for import. _______________________________________________ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca
