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

[Rick Collins]

At 06:06 PM 10/9/2007, you wrote:
>Rick Collins wrote:
> > So how does this mean that GPS signals are more accurate?  The last
> > time I checked, GPS calculations were done by explicitly calculating
> > both the position and the speed of multiple satellites in a very high
> > orbit, which means a very, very high speed compared to an aircraft or
> > low earth orbit.
>
>Indeed. But this particular calculation _is_ in fact something that
>someone has spent billions of dollars on, if you will permit me to use
>your argument, as the US Government uses it to pass Tomahawks through
>windows.
>
>I know we don't necessarily get their accuracy, but my point is that the
>original calculation is going to be very, very good.

Yes, the calculations can be superb!  But as with any processing, 
garbage in, garbage out.  If your signals are not perfect, you don't 
get the accuracy claimed.  BTW, the military uses a completely 
separate system at this point.  Foot soldiers and the like still use 
the original system, but advanced weapons use newer, improved 
signals, not to mention processing; so there is very little comparison.


> >> My GPS on the other hand, is an OpenStreetMap recommended model, and I
> >> was walking up and down the centre of the roads (not cycling or driving)
> >> with a clear view of the sky.
> >
> > I have no idea what is meant by "OpenStreetMap recommended
> > model".  Are you trying to say that your model of GPS receiver is
> > more accurate and consistent than others?
>
>No, I mean that the openstreetmap.org wiki recommends it on their
>hardware reviews page, and has even set up a special deal so they get
>some money when you buy one. It's a NaviGPS (BGT-11 model).

So how does that affect the accuracy of the unit?  The data sheet 
claims 25 meter CEP.  That is actually not very good compared to 
some, but I don't know that any of the "specs" relate to real world 
conditions.


> > How was that
> > determined?  What model is it?  How about checking your route by
> > comparing to some other data rather than just making an assumption
> > about which is more accurate and which is less accurate?  If you
> > don't want to use the USGS images, why not try measuring the same
> > route on multiple days?
>
>What are these USGS images you speak of? I haven't seen them as an
>option or an add-in for JOSM. Are they as high-res as the Yahoo images
>in the area in question?

I have posted twice about them.  I don't know much about JOSM 
really.  I am not finding it to be a very facile tool.  If JOSM is to 
be used for this sort of work, it might be a good idea to add in 
capability of viewing the USGS data.  Of course this is only useful 
within the US.

USGS is a department of the US government (US Geodetic Survey, 
IIRC).  They produce maps and such.  They have made their topo maps 
and aerial images available to companies and in turn Terraserver 
(Microsoft) provides them in consumer usable image form.  A program I 
us to view these maps and images allows you to overlay GPX tracks and 
graphically edit the GPX data.  This is not a perfect editor either, 
but it is very good for viewing and not bad for moving points around 
(culling the data points is not so facile).  Once you have a culled 
data set, USAPM is not a bad way to edit the file for accuracy 
(assuming that you trust the accuracy of the maps and images).  As 
others have said, it is not unlikely that all of these data sets have 
some inaccuracies.  This will provide another reference for you to 
use before you decide to keep or move your data.  The few runs I have 
made show very good alignment to the USGS data.  You can download 
USAPM at http://jdmcox.com/.  The program will download the image 
data as needed when you refresh a screen.  There is a companion 
program to download arbitrary regions.  This is not an open source 
program, but he does provide the c source.  He does not provide the 
other source files such as the .h files or the .rc files.  So you can 
view what is happening, but you can't do anything to fix problems.

The USGS data comes in a variety of resolutions.  The topo maps reach 
4 meter per pixel (don't ask me what that is in terms of x:1).  The 
aerial data is available in 1 meter per pixel or for areas around the 
major cities 0.25 meter per pixel in color.