GPS Error Measurement

What?

An empirical estimation of GPS precision (in the sense of giving measures consistent with themselves).

How?

I placed my Holux GR-213 (SirfSTAR-III GPS mouse) on a fixed position, and captured its NMEA output during two intervals separated by some months:

  • 2005/11/20 22:55 to 2005/11/24 14:56;
  • 2006/02/19 01:21 to 2006/02/27 14:41;

    • There's a total of 955468 measurements in the sum of both intervals.

    I then wrote a program to process that NMEA data and produce a gnuplot script that will in turn produce a graphic chart of the GPS position variations (i.e, "errors") found on the collected data, using colors to diferentiate the measurements by their reported HDOP (Horizontal Dispersion of Precision).

    Graph:

    Conclusions:

  • Beware of GPS measurements with HDOP > 2.0 (especially > 4); you could easily be off a few hundred meters... :-/
  • Measurements with HDOP <=2.0 (especially <= 1.0) tend to be very precise, in the range of less than 50 (20) meters.

    Last modified: Mon Feb 27 20:32:16 EST 2006