[OSM-newbies] Trust GPS or Yahoo aerial imagery?

[Rick Collins]

At 11:47 AM 10/9/2007, you wrote:
>On 10/9/07, Rick Collins <gnuarm.2006 at arius.com> wrote:
> > At 09:01 AM 10/9/2007, you wrote:
> > >Richard Fairhurst wrote:
> > > > To answer the original question, though, trust your GPS over the
> > > > imagery. The imagery may not always be rectified, whereas as long as
> > > > your GPS has produced a good-quality track - i.e. no "concrete
> > > > canyon" distortions - it should always be accurate.
> > >
> > >OK. So if that's the answer (and it's sort of what I expected) then the
> > >follow-up question is: do we have any procedure for fixing the
> > >rectification, or letting Yahoo know their images are out?
> >
> > Maybe you guys should take a step back and look at what you are
> > saying.  You have one person with a $200-$400 GPS receiver recording
> > a track which then is hand edited to add to an open source map.  This
> > track disagrees with aerial imagery that companies have spent
> > millions or even billions of $$$ to collect, process and
> > provide.  Now you are thinking of telling a distributor of that data,
> > based on your single track of collected data, that they need to
> > "correct" their data.
>
>Yes, actually. It's well known that the imagery is not precisely
>orthorectified, but it's usually acceptably close. We've seen this
>happen in many areas, backed up by statistically significant numbers
>of GPS tracks from different receivers over huge amounts of time. The
>most common thing is to be a few tens of meters off in a particular
>direction e.g. the images have the junctions 15m NE of their true
>location.

But "statistically significant numbers of GPS tracks" is not what 
this discussion is about.  The OP said he had a single track using a 
single GPS receiver which was compared to a single source of 
imagery.  He was asking which needed to be corrected.  The right 
answer is neither until you verify either the imagery or the GPS 
data.  Anything else is being careless.


>Remember they haven't spent "billions" of dollars on each image tile,
>so it's expected some of them will be a bit squiffy.

"a bit squiffy" is not much of a technical term.  Yes, there may be 
some errors, but don't automatically assume the imagery is wrong just 
because you have a difference between your track and the 
imagery.  Actually, I expect that billions have been spent on the 
various sources of imagery over the years.  It's not like that is a 
lot of money, you know... ;^)


> > Do you really think they are going to take you seriously?  Perhaps
> > you are thinking of the Yahoo imagery as if it were a collaborative
> > effort like OSM?
>
>We've got a lot of experience, collectively, at this game nowadays,
>and we know what we're talking about.

Like I said, do you really think Yahoo is going to take you seriously 
if you show them one track that disagrees with their imagery?  Or 
more importantly, will Yahoo care???  I think the answer to both 
questions is "NO".