[Talk-ca] What's up with those forests in Canada section

[Sam Dyck]

Here's my suggestion for a sort of FAQ (in wiki markup), incorporating what
James already wrote. I'm posting it here for comment because I have a
tendency to get unhelpfully passive aggressive.

The squared off sections of forest in Canada are the result of unfinished
CanVec data import. CanVec tiles are broken up into squares called the NTS
grid to better manage the data. If you see a forest that's squared off with
a empty section beside it, it's most likely that that grid has not been
imported yet.

''What is Canvec?''

[[Canvec]] is a digital product produced by the federal government that is
a combination of various federal geodata databases into 1:50000 tiles.
These tiles were converted by Natural Resources Canada into OSM XML and put
on a government FTP server for importation into OSM. After several years of
licensing discussion.

''Some of the data in a Canvec import changeset has something weird going
on (forests overlapping in lakes, islands where there don't appear to
islands, wetlands where there sohuld be lakes). Why are you importing this
garbage?''

Canvec is generally accurate, it was collected from high quality satellite
imagery collected for the federal government, and has generally withstood
our attempts to ground truth it. However there are errors and apparent
errors. Some of these can be explained by natural changes: lakeshores shift
with the years and seasons, lakes become wetlands, forests burn or are cut
down and regrow.

The simple reason we have to do this import is because Canada is enormous
and has very few people, consequently there are large areas that have a
very light human presence. For example the territory of Nunavut, the
largest subnational division in Canada, is larger than of France, Ukraine,
Sweden and the United Kingdom combined and has less than 40,000 people.
Most people in Canada live in a handful of cities a short distance from the
US border. There is a lot of blank area to fill, and so we make an effort
to import quality data, but there is a lot of area to cover, so after long
discussions we arrived at the consensus that importing Canvec data was the
best solution, providing we followed a set of practices.

''Don't you have local mappers in these communities who could check the
data?''

Most likely no. See the note about population density above. Also much of
non-urban Canada, especially Northern communities, have to rely on
satellite internet, which is both extremely expensive and has both
effective download speeds measured in kbps and small data caps of 5 or 10
GB.

''I see some issues with Canvec data, what should I do?''

If you think the data itself is in error, try and check to see if it could
not possibly be an accurate reflection of what might be at some point.
Canvec importers have been criticized for importing data, that while it
looks suspicious, accurately reflects what is on the ground. If it's an
obvious error that's easy to fix, go ahead and correct it. If there's
something bigger, talk to the mapper or post on the talk-ca mailing list.

''I see something wrong with the actual structure of the data (overly
complex ways, duplicate ways).''

These should have been fixed in the import, but sometimes things get
missed. Please go ahead and fix them.

''I found a Canvec import that didn't comply with the import policy!''

Please don't revert it, despite the appearance of wholesale importing, a
proper Canvec import takes a lot of time and effort on the part of the
importer. Canvec imports began before the current import policy, and so
some importers continued what they had already been doing unaware of the
policy. Hopefully everyone is in compliance now, but if you do see
importing incorrectly please assume good faith.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20160831/f01da63a/attachment-0001.html>