Subject: Re: Single-receiver Accuracy
From: Sam Wormley
Date: 12 Nov 1996 22:52:31 GMT
Accuracy Statistics
Trimble's Acu-Lock on ScoutMaster
MIT Lincoln Laboratory's GLONASS group provided daily scatter plots of
GPS Position Estimates. See http://satnav.atc.ll.mit.edu/. It is obvious
that one's true position can be estimated using the average of positional
data collected over a long period... in this case 24 hours. I set out to
obtain an estimate of the repeatability of obtaining "true" position by
averaging over 24 hours.
If one obtains a sample of the population of computed positions and
computes the mean value for each sample, the sample mean values will be
normally distributed around the population mean, whether or not the
original population was normally distributed.
Using the Acu-Lock feature on Trimble's ScoutMaster GPS receiver the
following sample means (in meters) were collected. Each mean was derived
from 38000 samples automatically collected over a 24-hour period.
The original data are reproduced here so that any reader may apply their
own statistical analysis to the data. My conclusion is that one has a
probability of .95 of being inside a circle of 5 meters (16.4 ft) radius
when using the Acu-Lock feature on Trimble's ScoutMaster GPS over a
24-hour period.
DATA -------Mean(meters)------
Easting Northing Alt Date Time (UT)
#,001,ACU.1.96.5.19 ,15,444912,T,4653491,00311,19-May-96,20:21:33,Acu-Lock,$
#,002,ACU.2.96.5.20 ,15,444917,T,4653488,00309,20-May-96,20:50:14,Acu-Lock,$
#,003,ACU.3.96.5.21 ,15,444915,T,4653492,00306,21-May-96,23:40:23,Acu-Lock,$
#,004,ACU.4.96.5.23 ,15,444909,T,4653493,00299,23-May-96,01:37:26,Acu-Lock,$
#,005,ACU.5.96.5.24 ,15,444919,T,4653490,00324,24-May-96,02:15:48,Acu-Lock,$
#,006,ACU.6.96.5.25 ,15,444917,T,4653493,00308,25-May-96,08:03:39,Acu-Lock,$
#,007,ACU.7.96.5.27 ,15,444915,T,4653493,00310,26-May-96,00:38:37,Acu-Lock,$
#,008,ACU.8.96.5.29 ,15,444916,T,4653491,00313,29-May-96,03:59:21,Acu-Lock,$
#,009,ACU.9.96.5.30 ,15,444917,T,4653489,00310,30-May-96,12:17:14,Acu-Lock,$
No of Sample Means 9
Range Min 444909 4653488 299
Max 444919 4653493 324
Diff 10 5 25
Population Mean 444915.3 4653491.1 309.7
SEM 3.12 1.83 22.6
95% Conf Interval 6.12 3.59 44.4
Scatter Plot
|---------- 10 meters ---------|
. . .
|
. |
|
. . |
| 5 meters
. |
|
. |
|
. |
"A measurement is worth a thousand expert opinions."
__________________________________________________
Sam Wormley - http://www.cnde.iastate.edu/gps.html
Subject: Re: GPS vs Traditional Surveys
From: jmurphy2@mindspring.com (Jim Murphy)
Date: Wed, 13 Nov 1996 02:41:22 GMT
arun gera posted the following:
>I am trying to understand the relative strengths of GPS and optical
>surveying equipment? Is GPS taking the market away from traditional
>surveying equipment?
I see GPS as an augment or adjunct to traditional equipment.
Maybe in portions of the country that are wide open to the sky,
there will come a day when GPS will be the equipment of choice.
Or maybe we will see much, much stronger transmitters on the
replacement birds that can burn through leaves. That won't do
much for buildings, though.
> Do all surveyors use GPS (what percent do)?
My best guess in my state (NC) is that perhaps 5% of surveyors
have GPS capability, primarily the larger firms. It's a long
ways from being cost effective for a one or two crew operation.
> AT what price point and performance levels will GPS overtake the optical
>surveying market?
>
When it costs the same thing as a total station and will work in
the deep woods reliably in real time without the Rube Goldberg
array of radio modems and antennae and other such potentially
problem prone components. In other words, maybe ten years, maybe
a hundred, maybe never. A better bet may be a system working
with the numerous FM sub-carriers transmitted. We heard of a
system at Duke University that purportedly would determine a 3-D
position to a millimeter within a millisecond. Cut through the
trees and woods and through buildings and everything. It was
part of a graduate level MBA marketing program, and I don't know
if the system actually worked (they swore it did), but that will
be the magic black box we have dreamed of for years. The
technology and physics appear, at least to me, to be workable,
although I doubt the millimeter stuff.
On a slightly different note, how many users of GPS have found
uses that they never considered before? One of our biggest
benefits is the ability to run open traverses with the terrestial
gear, tying each end to a GPS shot. That is a huge timesaver and
we probably use that feature as much as control extension. Just
today we ran 2500 feet of sewer profile and section, starting on
a GPS position and tying in to another. We made no effort to tie
to a plane coordinate system, we just use the 3-D vector for a
redundancy check on the ground work.
Also, who can justify the additional cost and complexity for RTK?
Maybe if you're doing construction staking at the Grand Canyon or
in the desert it can become cost effective. What's the real
benefit of real time for any collection efforts? And please
contrast the benefit of real time against the cost and complexity
of the current crop of equipment.