[Talk-ca] Canada Import chart

[Frank Steggink]

Hi Sam,

>     I think I've done enough import for now. 9 1/4 sheet of data. The
>     1/4 sheet was a section near Quebec City (021L14) with a lot of
>     data and a lot of cleanup necessary.
>
>
> When you say 'sheets' im assuming you mean NTS tiles right?
Yes, that's correct. In my parlance sheets are big files, but tiles are 
small files (like the OSM tiles). Maybe sheets can be seen as large tiles ;)
> Here's where my thinking is.   We break down the process to 3 steps...
> 1 - Data conversion:  by converting the geobaseNRN directly from the 
> gml file, and skip roadmatcher, and make the complete .osm files.  We 
> then make these full tiles available.
> They dont need to single tiles, you can group them how you like (just 
> keep to the grid boxes)
>
> 2- Data copyover (rather than say 'import') as its manual.
> This way, our focus in on just getting the data available.  Local 
> people can be working on the manuall copying over of the roads that 
> they want.   So if they were the one that did the initial imagery 
> tracing, they have the option to swap or not.  and would be able to 
> attach the ref tags and relations needed.
>  
> 3 - Roadmatcher results.  Having these available as an .osm file, so 
> users can copy might be better than importing.
Regarding the roadmatcher results: in most cases I've modified the OSM 
files I end up with after running RoadMatcher. The reason is to correct 
obvious errors, copy bridges (which are often missing in the existing 
OSM data), and various other reasons. I'm not sure if anyone would be 
interested in those files. Since you've expressed multiple times that 
you hope that people make them available, what is the exact reason for that?

About the general process for the import: this sounds feasible to me. 
Hopefully more people would be inclined to help. I found Roadmatcher 
often very frustrating to work with, and I often end up retiring roads 
which are already in OSM, so Roadmatcher will ignore them. For the final 
result this doesn't matter, but having to do so much manual work, 
devaluates the use of Roadmatcher.
> So this way, our efforts are in the actual conversion.  At this point, 
> only 4 or 6 people in Canada know how to actually use RoadMatcher and 
> geobase2osm.  (have a GIS background) Everyone else knows how to use 
> JOSM, and can open up a .osm file and start copying over the data.   
> It's not essential that we record WHO is copying over the data, as 
> long as 1 person in each tile area is listed as a contact to figure 
> out WHO is doing the import copy.   Usually, it would be 1 person per 
> tile, but for busy areas it would be 2 or more.
I think it would also be good that after the copying (or maybe during 
the copying) the data is connected. You mentioned in the past that 
existing data shouldn't be touched, and I don't know your current stance 
(haven't followed the import discussions too closely), but eventually 
someone will link the NRN data with the old OSM data. When this is done 
during the import, it is easier to keep track of what has imported. 
Although is a lot of work to do, it is worth it in the end. In case 
you're not sure if a connection should be made, you can see in the 
original Geobase data if there is a connection.
> So far I haven't looked at CanVec data yet. It takes a while to figure 
> out how exactly it is constructed, which is hard to do with a quite 
> demanding day-job, and with only a few hours left in the evening. I 
> actually wanted to look at the NHN data first, and import some of it. 
> I think I'll contact Yan Morin for info about that. Regarding the 
> Geobase data: do you know if land cover is also available to us, and 
> if the quality is good enough to be imported? If we manage to import 
> that, we'll really be able to show off how forested Canada is :)
Apparently Richard had this idea already. It is a great addition. 
Although the data can be out-of-date, this will also happen when we add 
data of the current situation today, and review it several years from now.

In order to get it straight: only the roads come from Geobase, and the 
rest is CanVec? There is a lot of overlap in the datasets, and for a 
casual observer of these discussions they look to be the same.
> (fyi, on an aside) the contours.osm converted file and the SHP files 
> will be available in the 'extra' folder of the canvec2osm (data 
> conversion script).   I just got word about the "OSM Atlas" project, 
> and to get contours, the shp files are needed, to make the PDFs.
Weren't you interested in creating an atlas last year? It seems someone 
finally heard your wish, and had time to implement it :)
> Do you think i should make ALL the SHP files also available for all 
> the data (right now im deleting them, after conversion)?  So then we 
> have a origional copy of it, rather than just my converted version.   
> (since the data gets updated, the older version might not be 
> available) to make the DIFF shp file. (for what was added since last 
> time).
I don't know. I'm not familiar enough with all the data. Focusing on the 
roads will already keep me busy ;) BTW, the files I referred are the DEM 
files from Geobase, and not CanVec contours. I can read them, and output 
some nice bitmaps, but they're not reprojected, and tiled up.
> I know that it's a tedious task... hey, why do you think im working on 
> many things at once (also schedualing meetups and such), as well as 
> tackeling the new Ontario datasets, among other things?   .. but these 
> next 2 weeks im trying to stay on task :-)
You should definitely keep up your work too. Without you blazing a 
trail, the import would go much slower :)

Regards,

Frank