[Proj] Google Earth Accuracy

[Irwin Scollar]

In defence of Google coordinates:

>Cliff Mugnier wrote:

>Ah, there's a BIG difference between the true coordinate system relations of
>geodesy used by national governments and one cooked up by an ignoramus at
>Google Maps that did not know what they were doing ... I guess there's a lot
>of that going around, too.
>
>I suppose even twits help contribute to keep knowledgeable consultants in
>business.

>Noel Zinn wrote:

>Not that Frank is responsible for the geodesy and cartography in Google Maps
>(or their abuse therein), but the phrase "the resulting lat/long coordinates
>are intended to be treated as WGS84 after that" so troubles me that I am
>sympathetic to Cliff's sentiments.  So, let's quantify the offense with an
>experiment that anyone can duplicate, perhaps in Proj4 (I work in Matlab).

I think that an appropriate way to examine the errors in Google Earth 
is to use points with known WGS 84 coordinates that can also be seen 
clearly in Google Earth.

In the UK, you can obtain free complete lists of latitudes and 
longitudes for more than 6000 white trigonometric point pillars 
placed by the Ordnance Survey between 1936 and 1962 from:

http://www.haroldstreet.org.uk/waypoints.php?hill=trigpoints

The lists include coordinates in the British National Grid and their 
values converted to WGS84 using  official Ordnance Survey software.

Unfortunately, many of these points are no longer maintained and many 
are not visible in Google
Earth. However, checks made on a random sample of 12 visible trig 
points in England and Wales showed most to be positioned by GE within 
about 2.0m which is adequate for any work on 1:10,000 scale maps such 
as the digital versions issued by the Ordnance Survey. However, some 
were off by as much as 12 meters. Similar lists of trigonometric 
points are available in some other countries, but usually require 
payment for accessing them.

The UK data probably gives a good idea of general GE accuracy under 
the most favourable conditions. Much of the GE data in the southern 
UK is obtained from vertical aerial cover rather than from satellite 
imagery.  In other parts of the world where GE does not use data of 
this quality, errors are higher, but are usually acceptable if the 
area examined is less than a 2 km square and only one image rather 
than a mosaic of multiple images is the area of interest.

A report on this test in PDF format is available on request.

I also compared the results obtained by a collaborator who used a 
hand-held Garmin GPS receiver (+/- 3 meter accuracy)  at some clearly 
visible GE points in a few places in Egypt and the Sinai desert with 
Digital Globe imagery and found that the errors were acceptable 
compared with the Egyptian Old Datum maps at 1:25000 after datum 
conversion using the old NIMA 3 parameter constants for datum 
transformation between the Helmert 1906 and WGS84 ellipsoids.

Irwin Scollar