[Proj] Google Earth Accuracy

Andrew Williams AWilliams at rapidmap.com
Tue Dec 2 17:55:53 EST 2008

This is a fascinating thread. Yet another issue arises as the Earth is not a static frame.
We resell a GPS unit here in Australia that outputs ITRF (International Terrestrial Reference Frame) coordinates. To look at the values they appear no different to WGS84 lat\long that any other GPS in the world outputs.
That is until you compare the location of a "fixed" point.

ITRF is a spaced based reference frame. It was originally designed for space based exercises such as docking a shuttle with a space station that needed an independent positioning mechanism.

The system has high repeatability but apparently low accuracy when compared to "accepted" coordinates.

On the There is an 8 parameter transformation to bring ITRF into co-incidence with WGS 84. The 7 are the standard 7 parameters we all know. The 8th is time. This system uses a differential correction signal based on ITRF in 1994 fixed in 2000. From then on a set of corrections has been calculated that takes into account continental drift.

In Australia continental drift equals 70mm a year North. So in 2000 the ITRF coordinate for a point was equal to WGS84. In 2008 the ITRF coordinate differs from a WGS84 coordinate for the same point by 560mm.

When you know this it's relatively easy to create code based on the 8 parameter transform and convert ITRF of the day to WGS84, however it does catch some people unaware.

-----Original Message-----
From: proj-bounces at lists.maptools.org [mailto:proj-bounces at lists.maptools.org] On Behalf Of Irwin Scollar
Sent: Tuesday, 2 December 2008 6:28 AM
To: proj at lists.maptools.org
Subject: [Proj] Google Earth Accuracy

MarKus Neteler wrote:

>How about precision in mountainous regions such as the Southern Alps?
>PROJ4/OGR reprojected mountain peak points from geonames.org as
>well as known points from our province are off for > 100m when
>visualized in GE which they match our local orthophotos. Is the positional
>error related to the height of the point (here, satellite orthorectification)?

In GE go to View, Sidebar and turn it on if it's not already
on.  There, go to Layers, Primary Database, and uncheck the Terrain
option checkbox.  That gives an orthophoto-like view even in the
Alps.  You can check this by turning the GE Grid off and on in the
View menu or pressing Ctrl + L.

It's far from perfect, but it's good enough for my application of
mapping archaeological sites visible as crop or soil markings in the
GE images over modest areas.  Cliff dwellings are fortunately rare
exceptions. :-D

A more important concern is to increase the apparent resolution of GE
imagery which is normally limited to the number of pixels of the
display screen to approach that of a cadastral map of a small
area.  Using the GE lat/lons of image corners which can be found via
KML API routines, one can mosaic multiple zoomed detail images if the
GE raw data is originally from a single good orthophoto or a mosaic
made from several of these or even from a single Digital Globe strip
in some parts of the world like the Nile valley.  I have had good
results with this in moderately hilly terrain up to a factor of as
much as 6 in some places on my 1600x1200 display.

Irwin Scollar

Proj mailing list
Proj at lists.maptools.org

More information about the Proj mailing list