[Talk-ca] ACQTECH vs. Valdate vs. OSM User imput for CanVec HD

Fri Mar 13 20:01:41 GMT 2009

Hi all!

What I've been told (safe source) about canvec/geobase data origine and accuracy...

About 5% of the available datasets (usually populated area) have been produced digitally and have an accuracy better then 25m. 
So, almost every thing you'll find in canvec have been scanned from 50K paper map.  The paper map where produced using 50K aerial photographies and their accuracies were between 25 and 100m (or worst in some case)! 

BUT, over the last years

- All the road network has been acquired using high accuracy GPS (90% better than 10m).
- Every datasets that had an accuracy worst than 30m have been geometricaly corrected.  

Then, the resulting canvec (Federal) datasets have an accurate road network, and the rest of the content should have an accuracy between 10 and 30m.

However, if you are standing on the bed of a stream/river with your GPS unit (and having a good satellite visibility), I would say you're probably more accurate that the available data!

Anyway, I would propose that everyone uploads their GPS traces even if some exists for the same way because it might helps to see what geometry is right.   

Daniel

Date: Thu, 12 Mar 2009 13:29:10 -0400
From: kevinfarrugia at gmail.com
To: talk-ca at openstreetmap.org
Subject: Re: [Talk-ca] ACQTECH vs. Valdate vs. OSM User imput for CanVec HD

I think he's aware that we can't use Google Earth, nor did he ever say to.  He was just saying that his GPS tracks met up with the actual path when overlayed on the Google imagery, meaning that his tracks are probably more accurate.

About the waterways - I think many of the waterways (smaller ones) were just traces of aerial imagery onto the topo maps.  Waterways are too much of a pain to do by GPS on such a large scale so I couldn't see them being too accurate, unless you get a municipality's watershed layers which I've found to be very accurate and detailed, but those aren't available to the public or under the CC so it doesn't matter anyways.  For all we know the federal data could have been aerial-->paper topo-->digitized, meaning it could be very innacurate.  I don't know what the metadata says since I'm at work and can't (shouldn't) access it, so I don't know if it says in there what the source is.  I think a user who knows the area should be the one who decides what should happen to the waterway or at least compare the bot loaded data with the aerial imagery that's already available to decide which is nearest to the actual location.
-Kevin (Kevo) 

p.s. sorry for sending twice sam, i always forget to press reply all.

On Thu, Mar 12, 2009 at 1:16 PM, Sam Vekemans <acrosscanadatrails at gmail.com> wrote:

Hi, 
Ya we (the OSM community) dont use Google Earth (because the map isn't free, you cant take a printed map of Google Earth/Google Maps and print it in Books. .. you can with OpenStreetMap.org

By submitting your original GPX tracks to OSM, you then help build the larger map of the world, the more original GPS tracks that get added to the area, the more likely and easier it is to accurately tell what map features are more accurate.  

The map that your making (yes, it is useful) however, the OpenCycleMap currently available, is also available to download as a garmin map from Cloudmade 
http://downloads.cloudmade.com/north_america/canada#breadcrumbs

& the OpenHikingMap is currently in progress which takes the OSM map features and tones down the highways so then all the trails stand out more.  Also, all the things relevant to Hiking will also be shown.

http://wiki.openstreetmap.org/wiki/Hiking_Map

Here's that the area of discussion looks like on the cyclemap layer
http://www.openstreetmap.org/?lat=49.58793&lon=-114.5186&zoom=17&layers=00B0FTF

But of course, if your not into contributing to OSM directly, im sure Simon would be able to help out again, and import the latest version of your .mp map when it's ready :)

Cheers,
Sam Vekemans
Across Canada Trails

On Thu, Mar 12, 2009 at 9:53 AM, Calgary Trail Maps <calgarytrailmaps at shaw.ca> wrote:

The current OSM data is from my maps, which are based on my tracks, other than the ATV bridge which someone else added.  You can see them clearly in Google Earth.  I'm working on cleaning this area up for my next release and I'll clean it up a bit more.

John

From: samvekemans at gmail.com [mailto:samvekemans at gmail.com] On Behalf Of Sam Vekemans
Sent: March 12, 2009 10:35 AM
To: simon at mungewell.org
Cc: Calgary Area Trail Maps; Talk-CA OpenStreetMap; Jean-Sébastien; Dale Atkin; Richard Degelder 

Subject: Re: ACQTECH vs. Valdate vs. OSM User imput for CanVec HD

Hi Simon,

So for the canvec data, it was recorded back in 1979 when the earth was flat .. lol :-)
it also says that the technique is "vector data" . ... it also says the planimeteric accuracy is "19" that would be 'meters' ... so your statement of '15 meters' off makes the canvec data also 'right' in it's own sence.
If your GPS had a better accuracy, then best guess is all you can do.

And John,
Thanks, would u be able to upload all your GPX tracks to OSM? 
If not, i can upload that for you, on your request.

Using JOSM, i just download all the local area GPS Traces, and i only see 1 track (Simon's)

And everyone,

Well, i "think" when it says "GPS" as the source, would be from a 'trimble differencial unit" (but not sure) .. and so, when canvec says it's a "7 Field completion" then we could says it's as accurate as any mapper account.

And so, looking at the contour lines on the Cyclemap, it looks pritty good.

Anyway,
I fixed the script so it reads as a "river" thats because it's a "Watercourse, non isolated" for the "Watercourse, isolated" thats when i list it as a "stream" and for all the "other" and "unknown" I list it as a "stream" so then they can be manually upgraded.
So for some of them that are labeled... like "York Creek" it's hard to say what it is.

I could change it so that "Watercourse, non isolated" is listed as a "stream" would that be better?

Thanks,
Sam

P.S.
I now need to go over the whole set again and fix up the wiki charts. Hopefully in the next few days it will be better. ... but of course, im always looking for errors.  anyway i published the rules.txt file and added the 'date-time stamp' from notepad so you can see how old it is.
http://docs.google.com/Doc?id=d2d8mrd_179gnhs54cw

On Thu, Mar 12, 2009 at 8:29 AM, <simon at mungewell.org> wrote:

> Hi, i have an area, which i found interesting.
> The York Creek (river) ish..

Are you perhaps refering to this spot?
http://www.openstreetmap.org/?lat=49.58793&lon=-114.5186&zoom=17&layers=B000FTF
http://www.flickr.com/photos/24244464@N03/2777260804/in/set-72157606820699424/

Where I sat with my doggie, having lunch on a fine afternoon last summer.

Looking at the sources available to me (OSM, OSM-GPX, CatMP and Canvec),
it appears that the Canvec waterway (coded 1470171) is about 15m too far
west.

That said, consumer GPS is not going to be very accurate. Look how the
(only!) GPX track is very noisy here. The only real way is to get multiple
GPS tracks going through the area an visually average/discount them to get
an approximately true location.

One thing that we have on our side when looking at waterways, is that
water is lazy and will take the easy route. So if you look at this area
with a contour layer you might be able to confirm which of the waterways
is in the wrong place.

Cheers,
Simon.

_______________________________________________
Talk-ca mailing list
Talk-ca at openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-ca

_________________________________________________________________
Experience all of the new features, and Reconnect with your life.
http://go.microsoft.com/?linkid=9650730
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20090313/3e0420ea/attachment.html>