Some Statistics
Note!  This does not mean the waypoints are accurate to 8 meters or so.  This is not a test of accuracy.  A few marks at each location are not enough for high accuracy.  These numbers also combine points where the satellite view is good, so relatively good accuracy and repeatability is possible, with points where the satellite view is lousy, so relatively poor accuracy and repeatability is the norm.
For anyone interested in a real life sample of waypoints...
I ran some of my waypoints through a spread sheet to get some statistics.  These are waypoints marked while hiking in the Colorado Rockies after Selective Availability was turned off.  Note that this is not a test of GPSR's.  My intent is to get good data for navigation including an idea of the error possible at a given location.

I only included waypoints of three or more marks at one location to eliminate those where a 12XL and a Venture were used at the same time.  These numbers might be slightly better than actually occurred because I occasionally delete or modify numbers that are clearly bad in the interest of getting the best location value.  However I also keep some "bad" numbers so a user can get an idea of how far off the GPSR can be.  That having been said...  I don't remember any marks that were really worse than indicated here and the percentage of waypoints I deleted or modified is very small - probably not enough to change the numbers.

One bad mark I remember was done on top of a ridge at 14,000 feet.  There was a great view of the sky with plenty of satellites.  The Venture placed my waypoint mark 30 meters away from the trackpoints it recorded (it recorded three trackpoints where I stopped to mark the waypoints).  The track recorded by the 12XL matched the track recorded by the Venture and the waypoint recorded by the 12XL was where I expected it to be.  I think the Venture needs some smarter firmware (better filtering) to prevent recording a waypoint that is clearly wrong.  This is the only time I have gotten an error this drastic with a good satellite constellation.

An anomaly: On one occasion I saw the 12XL move in one jump about a 1/4 mile away while I was standing still.  After a few minutes "it returned" to my current location.  No marks were made at that time.

The results..
There are 6700 marks with 1385 at the same location which is an average of 4.84 marks per location.
There are 1403 marks that included the elevation with 573 at the same location for an average of 2.45 per location.
Item UTM Easting UTM Northing
Largest spread: 57 meters (187 feet). 73 meters (240 feet).
Average spread: 7.6 meters (25 feet). 9.2 meters (30 feet).
Standard deviation: 6.3 meters (21 feet). 7.6 meters (25 feet).
Number of times spread greater than 20 meters: 64 times (4.6% of 1385). 97 times (7% of 1385).

Note: The spread is the max. value minus the min. value for the point.

The largest spread is 168 meters (552 feet).
The average spread is 15.5 meters (51 feet).
The standard deviation is 18 meters (59 feet).
Return to the Index (Home) Page
Copyright © 2001, 2002 Dan Anderson. All rights reserved.