Affordable GPS systems

[Gary Fewless]

There are several basic points to be made regarding use of gps
receivers, if you are new to this arena.

1. The accuracy cited by some manufacturers and many retailers of gps
units relates to the internal precision of the instrument (often 5-30
meters).  What they fail to tell you is that the signal provided by the
satellite network intentionally introduces an error of up to 100 meters
(for military security reasons). Therefore basic gps receivers do not
provide readings which can be depended upon to be less than plus or minus
100 meters.  The true location of your site is within a circle of radius
100 meters around the location specified by your gps.  The location COULD
be closer to the truth, but you cannot tell from the basic reading.
Actually, the algorithm for the error is designed to produce 95% of
readings within 100 meters, so a few readings may even exceed that
distance, and of course many will be less.  Inexpensive units will NOT
produce better accuracy, without add-on equipment, although the ads will
try to imply this.

It is even worse if you try to relocate this site again using gps,
since, for example, the first reading may be off up to 100 meters to the
east and the second reading at some future date could be off up to 100
meters to the west.  This level of accuracy is fine for large scale
purposes, but is less useful for locating specific study sites, individual
plants or populations, etc.  The fact that any given reading COULD be more
accurate or even that most readings will be more accurate is of no
consolation if you don't know how accurate any particular reading is.  In
the midwest (USA), better accuracy can be obtained with a good topo map
and a little observation.

2. The errors may be random, but they are NOT random on any short time
scale.  I have seen several experiments conducted on the error and I have
collected data on my own instrument (Trimble Scoutmaster) with the same
result. If you take readings at 5 minute intervals you can see that the
points "wander" about the theoretical circle of 100 meter radius,
forming a curling, twisting line.  The data is NOT random within a
period of even several hours, and the accuracy of your readings cannot be
improved by averaging readings on the short term.  Indeed it is quite
conceivable that an average could be more inaccurate than your initial
reading. It would probably be more successful over the period of one or
more days (some models provide for this option), but that is usually not
an option.

Within the wandering pattern of this almost constantly changing
error lies the explanation for several of the replies to this post.
As the data wanders about the area it periodically jumps or moves rapidly
to some new location within the circle of probability and then begins its
slower movements again.  This might explain the response which mentioned
turning the unit off to reset it when it drifted off.  It also relates to
the constantly changing altitudes mentioned by several people.  That is
not a function of poor equipment, but rather of the changing error factor
applied to the signal.  If you hold the unit still and observe for 10
minutes or so you will see the lat/long vary as well. I have read several
reports on GPS which suggest that the potential error in altitude is
nearer to 200 meters.  I have clearly observed errors from known locations
on usgs topos which significantly exceeded 100 meters in error for the
altitude.

The good news is that you can achieve better accuracy by using
differential correction signals, provided by ground based equipment which
precisely "knows" its own location and can calculate the error introduced
by the satellite signals.  You can either buy an add-on unit to receive
this signal in real-time or you can store the data point in your receiver
and "post process" the data when you get back to the lab, if your gps also
saves the time of the event.  Horizontal accuracy of about 5-10 meters is
promised by the manufacturers and I have seen U.S Forest Service data
which suggests that such accuracy is possible.  I don't know what the
vertical accuracy is, but I suspect it is always going to be less than the
horizontal accuracy.

Real time differential units are basically specific to the individual
receiver and are not cheap but most are in the hundreds of dollars
category.  Then you must get the correction data which means you must
either 1. buy a differentital beacon which you must set up and
maintain (expensive - thousands of $$), 2. Pay for a service which
supplies the signal (you will have to buy their differential receiver
too),  3. get the data from some other source, some of which are free.  In
the Great Lakes area and near other major waterways the U.S. Coast Gaurd
provides differential signals for navigation for free (but you must buy
from a third party the appropriate receiver to get the CG signal).  There
is a lot of info on the web explaining all the aspects of gps and a quick
search will turn up some good general explanations.

If you choose the post processing route, you must find a source for the
correction data (and it should be near your area of use) and you will need
software to handle the data if you have numerous sites to correct.  The
web site listed below is rich with information regarding data sources.

http://www.cgrer.uiowa.edu/servers/servers_geodata.html#gps

If you can live with the 100 meter accuracy, there is no reason to spend
more than $500 for a gps unit.  Several models by Garmin are in the
$200-300 range and are suitable for basic work. They can be purchased from
sporting magazines and sportsmans equipment stores.  Magellan and Trimble
are two well known manufacturers with a broad range of models spanning the
sub $1000 to "big bucks" range.  Several examples are listed in the
"Forestry Supplies, Inc." catalog.  I have used a Trimble Scoutmaster
since 1995 and find it to be very sturdy and functional -- it has almost
certainly been superceded by newer models by now.

If you are serious about the more expensive equipment you may want to
check out the publication "GPS World", which is very expensive and hard
to find, but it has a site on the web (I've lost the url temporarily) as
well as a paper edition. If you have the big bucks then you can achieve
much higher accuracies, down to the 1 or 2 millimeter range, but most
biologists don't have the means or the need for that level of accuracy.

You should also consider where you will be using the gps unit.  All gps
receivers have some difficulty receiving the satellite signals if used
under dense (particularly high moisture content) canopy, or in steep
ravines, canyons etc., and remote areas are not likely to have
differential information available.


g fewless
University of Wisconsin-Green Bay
fewlessg at uwgb.edu