I find that accuracy a bit hard to believe also. The William J. Hughes
Technical
Center - part of the FAA - produces performance reports on WAAS. They use
fixed reference stations and analyze data on a quarterly basis. In the last
report:
http://www.nstb.tc.faa.gov/reports/waaspan5.pdf (3376.8K)
The best 95% horizontal accuracy was from Columbus OH at 0.734 meters.
The worst was at Elko, NV at 2.132 meters. A year earlier the best horizontal
accuracy was at Chicago at 0.884 meters (95%). What you are claiming -
and I don't totally dismiss it - is better accuracy with your ~$4000 GeoExplorer
than any of the more expensive reference stations in the above referenced
report. It does border on "unbelievable".
I spent about 20 minutes or so wading thru the acronynms of this report,
trying to make some sense out their experimental method for actually
collecting the data they analyzed. In referring to the data collection
method, the report said "This was accomplished by utilizing the GPS/WAAS
position solution tool to compute a MOPS-weighted least squares user
navigation solution ... ". It almost sounds like it was a simulated result,
rather than an actual collection of data by a GPS receiver? If the data
were collected at the WAAS reference stations, where the correction
calculations are performed, shouldn't the error be zero there? I'm afraid I
never made any sense out of that, so maybe you could enlighten me. I would
also throw out that the specs that DOD has put out over the years for GPS
accuracy have always been extremely conservative, so maybe some of that is
going on here? Perhaps a CYA approach, if you will. (Even us civilians can
use acronymns sometimes.)
I noticed that the Elko, NV 95% vertical accuracy was also the worst of all
stations, so it makes you wonder if something else was going on there. Once
again, I didn't collect a tremendously large amount of data, since the
object of my testing was only to determine whether or not the unit was
capable of submeter performance with WAAS corrections. Even so, after
looking at about 200 data points and seeing that the greatest error recorded
was only 0.78 meters, it seems safe to say that the unit passed the test.
Actually, Trimble only claims a submeter RMS value in their specs, not
2DRMS, as my results indicate. Looking at a worse case scenario of 0.99
meters as an RMS value, then the corresponding 2DRMS value would be 1.71
meters. This would still satisfy their claims for the unit, but I found it
to be much more accurate than that. We have a saying when doing wetland
delineations, "It is what it is!" There's nothing quite like going out and
doing an experiment for yourself, in the environment where you will be using
the GPS. This is what I observed with the methods I used. Maybe if I catch
up on my work load, I will go out to some different NGS benchmarks and do
some more testing.
One way I could see this type of accuracy being achieved is if the receiver
"auto-averaged" the position - perhaps without telling the user. Did the
receiver sit on the tripod for a while before you recorded a position?
The receiver was not set to average the position and does not "auto-average"
without telling you. These were all single point captures.
If this receiver can provide 0.61 meter (95%) accuracy while moving -
in open terrain - then I'm certainly surprised and would like to know more.
It could be checked by recording carrier phase info while wandering around
an open field and then comparing the WAAS corrected realtime output
to the carrier phase post-processed position info.
I never attempted any tests while moving, so I can't comment on that. The
unit does have the ability to do post-processing with carrier phase, with
the proper software. According to the specs, the GeoXT is capable of
achieving a 30 cm RMS accuracy (52 cm 2DRMS) by doing that. I don't have
the software to attempt this, so I have never tested this figure.
