DPRG List  

[DPRG] how is GPS covariance estimation done?

Subject: [DPRG] how is GPS covariance estimation done?
From: Ed Okerson ed at okerson.com
Date: Sat May 19 22:41:01 CDT 2007

> The interesting thing is that two GPS units (same model, built by the
> same manufacturer) will wonder the same, and have the same differences
> if they are not too far apart, (couple of miles) because they see the
> same atmospheric errors in the waves and calculate the equations the

This is a common myth.  The fact is that two identical receivers can
choose different satellites if they are as little as half a wave length
apart.  And since the L1 wave length is only 20 centimeters, that is quite
short of your estimate of a couple of miles.

> same. If one is fixed at a known location, then if the GPS says it is 10
> feet north of where you know it is, then you know its partner is also
> reported 10 feet north of where it really is. This is what Differential
> GPS is all about, and it is another way to improve your GPS accuracy.

This is also incomplete.  Differential GPS works not by calculating a
correction based on the error in the final position, but by measuring the
error in the signal from each satellite.  There is no way to take the
output from a normal commercial GPS receiver and post process it to make
it more accurate.  The reason is because the output from them is in the
form of NMEA sentences that do not contain details about which satellites
were being used to calculate the position, therefore you don't know which
of them contained the error that caused the position fix to be wrong.  The
signal path of each satellite is vastly different, and each is changing
constantly as it moves around the globe, and as the receiver moves. 
Differential receivers take data from survey grade receivers on the error
of every satellite in their view, and calculate corrections on each
satellite, then use that corrected data to generate a position fix.

If you want more details about this do a google search on "poor man's
dgps" there are many fine articles that explain the gory details.

Ed Okerson

More information about the DPRG mailing list