[talk-au] Way errors in Quilpie Qld

[osm.talk-au at thorsten.engler.id.au]

The data in OSM right now does not get anywhere close to being aligned with
currently available Esri, Bing, or Maxar Premium.

I would assume that the lot boundaries recorded in DCDB are "exact" and any
discrepancy between them and the physical world come down to the margin of
errors the surveyors did when placing boundary pegs at some point.

Comparing DCDB against the different imagery layers:
Maxar Premium is significantly off: 8.06; -2.09
Bing is almost right: -0.15; 1.79
Esri is very close to: -0.30; -2.09

With these offsets, all 3 imagery layers are also in good agreement with the
(rather marginal) available GPX tracks.

Selecting a road, then all connected ways with it, and moving them to match
the correct imagery layer shows that, within an acceptable margin of error,
all roads are correctly defined relative to each other and are simply all
offset by the same value.

Other mapped objects, like buildings and such are not all on the same offset
and can't be corrected in a simple single step like the roads. They are also
all not squared and in many cases aren't even close to follow the actual
building outlines visible in imagery.

It is my understanding that we are not actually allowed to use DCDB for
mapping in Queensland currently because of the outstanding and probably
never to be granted CC waiver? I'm not sure if that does even prohibit the
minimal use of it to validate imagery layer alignment against it. In that
case the best course would be to collect more GPS data to see if it's
possible to derive a comparable offset for the imagery layers purely from
that.


-----Original Message-----
From: Bob Cameron <bob3bob3 at skymesh.com.au> 
Sent: Saturday, 2 October 2021 20:40
To: OSM Australian Talk List <talk-au at openstreetmap.org>
Subject: Re: [talk-au] Way errors in Quilpie Qld

Wow, what a changeset!

Bing was 2010-2012. The Maxar Premium seems the most recent.

Whatever the case with the buildings, the roads tend to be used as anchors
for many features. I guess the opinion I am looking for is that if a formal
source can't be found/used, will my GPS run tomorrow (plus the existing
traces) be acceptable? I can make a point driving a set centre distance but
errors will still happen. The current trace/way error is in the region of
8-20m.

Being Sunday too the dashcam view wont be too obscured by parked cars. 
Also worth a walk to see if any businesses are closed.

On 2/10/21 7:01 pm, Warin wrote:
>
>>
>
> Note the changeset comment in
>
https://www.openstreetmap.org/changeset/58251817#map=16/-26.6197/144.2679...
>
> "l know these places are in the right place now as l have lived here 
> all my life." He used bing as a source .. so I think that may mean 
> they are ok in relation to one another ..
>

_______________________________________________
Talk-au mailing list
Talk-au at openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-au