Re: GPS

["Duncan Fletcher" <dfletche@gte.net>]

Accuracy depends on several factors. Visible satellite geometry is the
primary factor. A minimum of 4 satellites is needed for an accurate fix. A
fix can be had with only 3 satellites, but the receiver then uses an assumed
altitude (general the last altitude it calculated when it had a view of 4
satellites) and this can severely degrade accuracy. Now if 4 satellites are
visible and spaced appropriately, accuracy will be quite good. But if two or
more of the 4 satellites are bunched up to one side, accuracy falls off.
With good satellite geometry, my maximum error is less than 150 feet (this
was measured against the road system on a USGS topo map after dowloading and
overlaying the track points on the map).

For those that have not used one for over a year, the government turned off
selective availability a year ago (May 2, 2000). This was the deliberately
introduced error factor by the military that affected civilian GPS's and
introduced substantially greater errors.

Differential correction will improve accuracy, but requires a second GPS set
up at a known location with receivers that have the capability of
communicating with each other. With high quality receivers and additional
post processing software (which applies all kinds of corrections),
accuracies of less than 1 mm are possible - this is what has been recently
used to detect the plate movements before and after the recent earthquake in
Washington.

Short switch backs and other short direction changes will be ignored and
introduce errors. This is related to how often the GPS records a track
point. This can be changed at least in some receivers. For example, the
Garmin 12 is set to Auto by default - track points are recorded based on
distance from previous track point (and I think, but am not sure, related to
the display scale set). This can be altered to any preset time interval to
the nearest second.

There is much good information on GPS and related subjects at
http://joe.mehaffey.com/.