Sub-Meter Accuracy

Don't get me wrong here. It would be too good if reliable submeter is
possible using only WAAS. But with an internal, rather simple antenna,
and L1 only?

I just don't think that the spacial resolution of the WAAS is capable of
doing that. I would expect submeter as a very optimistic figure for very
short baseline DGPS.

But we'll see. Making and applying WAAS corrections is complicated and
may still be improved significantly as the people involved get more
experienced.

This report looks very promising:

http://www.satloc.com/site_Feature_DiffSource_WAAS.htm


- Carsten

 

Statistical data is the final arbiter!

 

Although this has nothing to do with the submeter accuracy attained with
some GPS receivers using WAAS as a source of differential corrections, it
is interesting to see some statistical data of some of the same receivers
collecting data in working conditions that include obstruction (such as
forested areas).

http://www.fs.fed.us/database/gps/gpsusfs.htm
Trimble Geo XT
Trimble Pocket GPS
Garmin GPS MAP 76
Garmin eTrex
Garmin GPS III with CSI Differential Correction (real time)
Magellan Meridian Platinum

And older data
Garmin and Magellan Recreational GPS receivers
Trimble GeoExplorer 3
Trimble Pathfinder Pro XR
Rockwell PLGR 96

It should be noted that this is not a comparison of all of
these GPS receivers under identical conditions. Technology
changes, the GPS as a whole improves, and test conditions are
not uniform!

 

Carsten,

It appears that the report you referenced above may be almost 3 years old
now, so I would imagine that many improvements to WAAS have been made since
then. Even so, the document states that WAAS is "at least as accurate" as
the signal supplied by other commercial providers of differential
corrections. I found this to be the case in my testing with the GeoXT as
well. I post-processed (code only) many of the points using the nearest
CORS station and found the errors to be significantly larger than the real
time WAAS corrections. In theory, post-processing with a given differential
source should be more accurate than real time corrections, yet the CORS
corrections were much less accurate. I soon quit going to the trouble of
post-processing and have used WAAS only since then.

The GeoXT's internal antenna has a ground plane set up plus the unit uses
the Everest Multipath rejection software, which is not cheap or "simple".
As I said in an earlier post in this thread, Trimble actually only claims a
submeter RMS figure for this unit. Choosing a worst case scenario of 0.99
meters as an RMS value, the 2DRMS value would balloon up to 1.71 meters and
still be within their specs for the unit. I have not observed this much
error, however, at least under open canopy conditions. You can pretty well
bet that Trimble would not advertise submeter performance for this unit,
unless it was capable of delivering it. They wouldn't stay in business long
by doing things like that! Why don't you try to borrow or rent a GeoXT and
do some testing for yourself? I think you will be amazed with its accuracy,
as I was. I was a doubter too, before trying it.

 

I want to further point out that there a least two knowledgeable posters
from this newsgroup who find the USFS testing "lacking" and I have to
agree, but on the other hand, the tests to give some indication of
expected performance in the field, where the variables are many.

-Sam

 

I, too, find it strange that two GEO XT reports show very different
results. And some of the other reports, though marked as 'preliminary',
should better not be published online before they have been
corrected/completed.

- Carsten

 

Great article Carsten.

Frank

 

Well, if you are willing to spend 30 minituts of your time, the new Magellan
SproTrak Color can do it: http://www.gpsinformation.net/mgoldreview/stc-2-10.jpg

 

Are others who are using professional grade WAAS receivers getting
horizontal accuracies in the 0.6 meter (95%) range without averaging?

Since SA was eliminated more than 3 years ago, I haven't paid much
attention to code differential as the ~4.5 meter (95%) accuracy of
autonomous GPS is good enough for some of my work and when I
need higher accuracy I just go straight to cm-level carrier phase
differential. The 0.6 meter figure is just a couple of decimeters
better than any other I've seen published so it surprised me a bit.
If others could share some numbers, be it 0.5 meters or 1.5 meters,
along with an approximate location within the WAAS service area
I'm sure readers would appreciate it.

In one of the references which Sam Wormley pointed out the
unaveraged WAAS accuracy of the GeoXT was in the 6 meter (95%)
range (in the open). I gotta think this guy did something wrong though.
http://www.fs.fed.us/database/gps/mtdc/geo_xt/ridley_ck_geoxt_rich_mccollough.pdf

As for getting significantly worse accuracy when postprocessing with
CORS data as compared to WAAS corrections, that shouldn't (in
theory) be the case unless the base station is hundreds of km distance.
My guess is that the CORS station you used had an Ashtech
receiver. In that case you will get worse results as the code
performance of the Ashtechs is worse than basically all other
higher end receivers. Try post processing with data from an AOA
or Trimble CORS station, even if it is a little farther away. You will
see better results by a factor of 2 or 3 (as compared to using an
Ashtech CORS, not necessarily WAAS).

The specs for the GeoExplorer may say 30cm RMS with carrier
phase post-processing, but any receiver designed to output good
carrier phase info is capable of ~ 1 cm rms accuracy. A little
worse when moving, a little better with redundant static measurements.
You just need the appropriate software, good receiving conditions,
and good data collection techniques. The software is expensive and
there are more limitations when working with carrier phase data, but
the accuracy you can get makes 0.6 meters (95%) seem quite
crude.

 

Actually the result was 0.8m for 95% and 0.6m for the "mean"... But I'd think
that for hand-helds, it would require averaging... In the case of Magellan,
that graphic is an average of an average...

Please don't get the idea I know anything about professional receivers! -One
thing I can say is that Magellan's spec for WAAS-corrected accuracy of 3m 95% of
the time is easily attained...

 

I believe you are referring to my experience with code post-processing here.
Actually the CORS station I used was closer than the nearest WAAS station.
I used the Trimble Pathfinder Express service to do the post-processing.
This service does not give you a choice of which CORS station you want to
use. It only uses the closest available station. I received corrections
from several different stations, however, during my testing, as some were
not available at various times. It didn't seem to make any difference.
They were all significantly less accurate.

I realize that post-processing should be more accurate, because of the
better coordination of time, but you seem to have the notion that WAAS is
inherently inferior to other DGPS sources. If so, why do you feel that this
is the case? I'm not sure it is.

The 30 cm RMS figure is just the spec that Trimble is quoting. As I said
before, I have heard that their specs are on the conservative side. The
GeoXT may be capable of 1 cm RMS accuracy with carrier phase, but I doubt
that Trimble wanted to advertise that, since their cm units go for much more
money.

 

Source?


TIA
BRS

 

Ah, black-box post-processing by a third party. I thought you were
doing it yourself with your own software. There are always so many
variables. I just thought I'd point out one common problem that might
be the cause. Sounds like there is a problem with their service.

I have no idea where you came up with this. Coordination of time
isn't really a factor since SA was discontinued. WAAS should be
inherently superior to other DGPS sources in the most general sense.
Of course there are many factors involved. If I happened to be
300km from the nearest WAAS base station and 300km from the
nearest single-point DGPS base station the corrections from
WAAS should be superior. If there was a single-point DGPS
base station at 50km distance I would choose it over the WAAS
corrections. These differences aren't going to be huge. This is
assuming everything else is the same - usually not the case.

That's exactly it. It's all about marketing and maximizing profit.
They get you to spend 4 grand for a good "submeter" receiver. Then
after a while you decide a bit more accuracy would be nice. They
tell you that you can use your existing receiver and just buy some
carrier phase processing software for ~$2000 (I have no idea what
the current price would be). You try it out and your happy because
you are usually getting better than 30cm, but because you aren't an
expert yet there are times when the accuracy is at the 5 cm level and
others when it's up around 30 cm. Now you want to reliably
obtain a couple cm accuracy. Well the specs for your receiver
don't claim 1 cm accuracy so it's time to upgrade both hardware
and software.

I haven't processed any data from a GeoXT so I don't know
for sure, but I have from a GeoExplorer III and 1 cm rms is
possible.

Back to the original question though. There must be people
reading this newsgroup who have done tests with "professional
grade" WAAS receivers. Any results on non-averaged data
would be appreciated.

 

See below

I have both WAAS (SporTrack Pro) and a DGPS/GPS (Garmin 12xl w Garmin DGPS
receiver) hardware. I am about 10 miles from a NDGPS beacon in Ft Edward
NY. I have not run definitive tests, and since many variables can influence
the perception of accuracy, I won't hazard a guess anymore which setup
produces the most accurate fixes. I think they are close. When WAAS first
became a viable system, I had thought that since it was a far more regional
system, it must inherently be more generalized and that the provided
corrections would not improve position fixes as much as a local DGPS beacon
signal. However, I have been surprised to find the accuracy specs for WAAS
versus DGPS suggest that WAAS is, in fact, more accurate.

 

Stupid question time here. Isn't the former L1 only and the latter L1/L2?
That makes all the difference in the world that I can't afford (yet, he
says).

 

Both are L1 only.

 

That is great news. I'm finally making some money mapping a few State
Parks trail systems, and though I've grown quite fond of my GPS16/PocketPC
combo, the best I can get from it kinematically, even in post processing, is
around 2 meters. Plus YOU Sam Wormley have posted so many interesting
Trimble articles that I have pored over that I have learned lust, though
thank God, not covetousness. (weird Christian joke there, I guess). I'm
guessing that the GeoExplorer III is quite a bit more than the GeoXT?
I now need to go back to Dennis Milbert's triple difference carrier phase
software, find a close order horizontal control marker nearby and find out
if I've been accidentally lying to people about how close (actually not
close) an L1 receiver can get. I realize there's worlds between a handheld,
a low/mid grade mapping/avionics receiver like mine (GPS 16. It just runs
circles around my GPS 12. You'd be shocked.), and a higher grade mapping
receiver like the Trimble, but things just keep on changing. Now if OPUS
would go ahead and open up to L1, it'd just save me lot's of trouble. Well,
all but the fact that OPUS also doesn't do kinematic.

Frank

P.S. Go to bed. You've got school tomorrow.

 

The data is online at http://www.ngs.noaa.gov/CORS/Data.html
for you to check out yourself. Use the maps and station descriptions
to find various combinations of receiver brands/models and baseline
lengths and do some tests. As I said in a previous post I haven't
bothered much with code DGPS for the past three years, but I did
a lot of testing before that. With my own equipment and CORS
data.

Just to refresh my memory the other day I downloaded a day
of data (Sept. 20) from MIL1, CALU, and LCDT. LCDT is
about 75km from both MIL1 and CALU. MIL1 is an Ashtech
receiver and the other two are Trimble. Making the LCDT
data kinematic and processing the CA code using MIL1 as
a basestation and CALU as a basestation shows much lower
CA code residuals for the Trimble/Trimble baseline then
the Ashtech/Trimble baseline.

I've done this type of test dozens of times in the past with
similar results. The CORS data is a great resource for doing
all kinds of testing be it short baselines, long baselines, CA
code only, L1 only, or L1 and L2.

You do have to be a little careful though as various
processing packages can behave differently. If you are
getting comparable CA code performance from Ashtech
CORS and, say, Trimble CORS I'd like to hear about it.

 

My Garmin Extrex Vista goes from 6 meters without WAAS down to 2 meters
with WAAS enable and running. This is Switzerland, two WAAS birds (maybe
three, cannot remember when the third was to become operational).

Ciao,

 

It's a slightly different thing. The sub-meter real-time WAAS accuracy was
achieved using software inside the receiver that reduced multipath error, a
prime cause of inaccuracy in the longer bandwidth L1 receivers.

Frank