GPS quality
Milby
Glen Thurow made the base files available to me along with a screen
shot of LSGMM for that regional. One thing was the version of LS he's
running only allows for 22,530,000 terms and the region requires
31,773,223 terms to generate error estimates so the suggestion to turn
error estimates on for the final run won't work. Ray has a version
that will handle 32,000,000 terms he compiled for the USFS Surveyor
from Michigan that attended the NMSU training session and had maxed
out LS. However.....
In looking at the ADJ file, over 2,000 of the 3,161 control points
were held fixed with reliabilities of zero. This is a problem
especially in light of the fact that the survey data was no where near
GPS quality. Here are the key stats from the ADJ file:
NORTHING CONTROL RMS = 19.858 SNOOP RMS = .6
MAX. NORTHING RESIDUAL AT 100467TT14NR13W OF 187.992
EASTING CONTROL RMS = 18.544 SNOOP RMS = .6
MAX. EASTING RESIDUAL AT 340700TT17NR21W OF 127.990
DISTANCE RMS ERROR = 13.318 SNOOP RMS = .7
MAX. DISTANCE RESIDUAL 900100TT18NR24W - 900110TT18NR24W OF
396.777
BEARING RESIDUAL RMS = 0-14-16 (D-M-S) SNOOP RMS = .6
MAX. BEARING RESIDUAL 200600TT18NR26W -140600TT18NR26W = 8-43- 0 (D-
M-S)
STANDARD ERROR OF UNIT WEIGHT IS 1.133
WITH **** DEGREES OF FREEDOM
CHI SQUARED TEST ON ANALYSIS
.748 < 1.133 < 1.208
(LOW) (HIGH)
PASSES AT THE 5 % SIGNIFICANCE LEVEL
A STD. ERR. OF UNIT WGT. BETWEEN 0.5 AND 2.5 IS
CONSIDERED SUITABLE FOR MOST APPLICATIONS
I had a discussion with Glen about no such thing as perfect data but
his dilemna is that is what the Missoula county folks are asking for -
the same coordinates out as they give for GPS. This, of course, is
not a new discussion and it seems impossible to convince folks that
GPS is not foolproof.
On the other hand, if GPS control moves within it's error estimate,
you can't say that you improved the data by changing it a foot or two
so I have no problem with folks wanting the same values out as they
put in. Given the complexity of combining so much data from
independent sources, it's impossible to say if we've improved or
degraded the control coordinates. But then, we haven't changed the
values in the CON file have we?
In a similar piece of information to consider, when NGS adjusted the
HPGN network in Wyoming, sations were adjusted up to 6 centimeters.
Stations continuously observed for 2 1/2 days and post processed using
satellite tracking data from NASA. I checked several other state
adjustments and this is well within the norm. I wonder what NGS
publishes. What goes in, or what comes out?
Bottom line is holding that much control as perfect turns least
squares adjustment and analysis from extremely difficult to impossible
on several levels. My suggestion to Glen was hold GPS at 1-3 feet and
then write a simple program to replace the lat/long in the final LX
file with the values in the CON file.
Milbert
Mil
*BE CAREFUL TO BACK UP THE OLD VERSION* as the executable in the ZIP
is named exactly the same.
I tested this version on several individual townships and tried
running the Missoula data set through last night but the pitiful
laptop of mine chugged away for 9 hours overnight before I killed it.
David Grainger
regions. I have had to do this when I was working with large areas.
Hmm. I wonder if it would run in SM/MM? I recently imported a large
region in NILS, I worked on several years ago in GMM and had to break
up into 2 smaller ones. I will let you know if it will run there. It
has 165 tps. in it, but I did not break it up because of GMM LSGMM
limits. I broke it up because of WinGMM limits when trying to preview
a region. By the way this area has numerous Amended Protraction
Diagrams, with 1596 fixed control points (.001 error estimates), out
of 2140, included. These points were isolated from the rest of the
measurement network by having no RAW data ties to them, so as to avoid
them having an affect on actual surveys. I agree with Milbert about
the use of "fixed" control in other cases, especially when it is GPS.
The only purpose of the LSA is not to only be able to provide
coordinates, but the quality of them as well.
Mil
closures in the RAW file. A perfectly executed geodetic (GPS) survey
should misclose something (varies by latitude) when converted to
planar bearings and distances due to convergency right? Here in
Wyoming, we see around 1.5 ft. E-W misclosures on the good stuff.
I'm a bit confused. In talking with Dennis and others, as well as my
own experiences, I've always understood that running anything over 25
townships through SM/MM least squares is a dicey proposition. Yet you
talk about running 165 townships as if this is a common thing. Is the
perceived limitations a network thing? I would be very interested in
seeing this region run through SM/MM LSA. You'd think holding all
that control fixed, along with the error estimates needed to
accomplish that on the surveys, there shouldn't be alot of differences
in the coordinates.
Jerry L. Wahl
based on the record being mean true bearings. The misclosure you see on a
cadastral plat, that is sometimes called convergency, (and I renamed to
'apparent misclosure due to convergency'.) Because the basis of bearing in
the PLSS is supposed to be true bearings everywhere and meridians do
converge, a typical 'perfect lower-48' township will appear to misclose by
half a chain more or less if you plug those bearings and distances into a
regular cogo program or calculator. However if you put them into a
geodetic traverse program that understands how to interpret mean true
bearings it will close fine.
So there is nothing in the geodesy of these programs that would limit the
size over which they will work except administrative problems in dealing
with SPC zones. In NAD83 you could compute your whole state using any SPC
zone without any appreciable error because the SPC capabilities in these
programs also deal properly and rigorously with the grid scale and rotation
factors and even the second term factors.
Although there is a mix of SPC computations and geodetic computations there
is minimal errors involved on the computational aspects. One test I
performed for the original GMM CMB was to run data using zones from
adjacent states (under NAD27) and demonstrate that you got the same
results. NAD27 SPC formulas contain some small degree of imperfection for
mercator zones, the NAD83 formulas are very accurate so it might be
possible to adjust a project in Wyoming using an Oklahoma zone for example
and the lats and longs would be fine. The SPC's would look pretty skewed
due to the larger mapping angle, but all the comps would be fine.
There are a few small problem areas. One is how elevation and variation of
elevation is handled. The other affects some proportions for non cardinal
long lines in some programs in CMM. Elevation issue one could see if you
use project elevations and have the same (let us say perfect) common
boundary data but each tp is being run at a different elevation. The
coordinates cannot match exactly. There are a few little nuances like
that where some more fancy elevation handling might have been implemented,
but I am getting off track. Regions are more likely to have a significant
variation of elevation so there will be some small bias in computations.
Since most (or a lot) of the PLSS data we use now is not highly accurate
19th century data, most of these nuances are not noticable amid the general
noise level of accuracy of the inherent data.
Control that is held fixed, if it is good stuff like GPS, isn't too far
from the truth. Holding a fixed boundary that is not in fact fixed by GPS
is lying to the adjustment. The biggest problem I recall seeing is that
due to large errors and blunders in the PLSS, even fixed control, (or
control weighted properly at say 0.1 foot) will still move. The LS tells
you something is wrong, but there are limits to how far you can go to junk
sids or deweight distances, etc. in order to try to make it work so that
control doesn't move. After a point, you are just wasting time because you
cannot necessarily tell where the 5 chain busts are without really making
decisions that should probably involve more control or field
work. Blunders cause problems with the adjustments. Control will move
even if it isn't supposed to. The LS tells you something stinks but you
just cannot do that much about it.
I have always argued that if someone wants that control back out the way it
came in, then write the input back to the output and let the measurements
into and out of that point take the heat. You can only do so much with
adjustments of blundered data. GMM and LS is still useful because it does
allow you to put different weights on measurements based upon your educated
knowledge and experience, and it does it;s best to give you the best
answer. It also tells you when the results probably are not that
good. A 1500 foot error ellipse is pretty telling, as is a 10 foot ellipse.
I won't go into the regional stuff more at this time. It is too bad all
the concepts were not completed, and possibly not even known to current
users, but the tools are there to make a virtually seamless PLSS
computation over a state (or more) but it requires procedures and processes
that would be pretty labor and concentration intensive.
The thread started with a little problem computing reliabilities, I don't
know if anyone has discovered the solution to that problem. I ran the
group of 9 tps as one kerplunk for the central township and it seemed to
work fine. The situation of all the data, the hardware and the storage,
etc. must be playing a role that would be hard to diagnose without being at
that machine with that data.
- jlw
Kurt Luebke
I hope I have a simple question for somebody. My WinGMM blew up on me a few weeks ago and I have installed, uninstalled and re-installed WinGMM many times trying to get it to run correctly. Whenever I go to the options to adjust the defaults I get the following message:
I had the software running properly before my system blew up and cannot seem to get deep enough into my registry to kill something that is hurting me.
I have located the tabctl32.ocx in my system32 folder and have given all permissions, but every time I try to adjust my settings I error out.
I am running Windows 7 Ultimate 32 bit operating system with a Core 2 Duo processor.
Thanks,
Kurt
Kurt Luebke, PLS, CFedS
Professional Land Surveyor
Certified Federal Surveyor
DJ&A, P.C.
Engineers, Planners and Surveyors
3203 Russell Street
Missoula, MT 59801
PH: 406/721-4320
C: 406/544-0646
F: 406/549-6371
Kelly Bellis
Try Windows XP Mode in the Virtual PC. It's the only way I found to run WinCMM, a slightly different animal, on my Win7 64-bit machine.
--
You received this message because you are subscribed to the Google Groups "Measurement Management BBS" group.
To post to this group, send email to win...@googlegroups.com.
To unsubscribe from this group, send email to wingmm+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/wingmm?hl=en.
V. Kelly Bellis, PLS 2099
Horizon Surveying
Company, Inc.
17 Union Street
Ellsworth, ME 04605
207.667.6912
Kurt B. Wurm
Kurt, Kelly:
Can you run regserver from the command line?
Try this: from the windows start button, click RUN.
In the popup box, type REGSVR32 [path]\tabctl32.ocx
Let me know Kurt.
FWIW, I’m having fits with a Microsoft .NET 4.0 webbrowser thingy that is preventing my immediate distribution of a new application: QuadFinder Pro. … anyone care to try it out?
Kurt Luebke
Kurt & Kelly,
I had some issues with my office network server permissions and this was the last straw. I just finished rebuilding my computer and now I am running with no problems.
Thanks for the response and Happy Thanksgiving to all.
Kelly Bellis
The command prompt is no longer available from the start button... and the start button is no longer labeled as such unless you hover your mouse over the round Window logo in its stead; however, under Accessories, the command prompt is still there :) - anyway... here's what I got:
kw...@nmsu.edu
kw...@nmsu.edu
----- Original Message -----
From: kw...@nmsu.edu
Date: Saturday, November 27, 2010 8:05 am
Subject: Re: RE: WinGMM install
To: win...@googlegroups.com
> Kelly, please ensure that the appropriate path was entered for
> the tab control Active X file. Remeber that the non-dos
> compliant foldernames MUST be properly addressed:
>
> Example:
>
> Regsvr32 c:\Users\V. Kel~1\
>
> This can be a very tedious path to type out .... (for me, its as
> difficult as texting on a numberpad)
>
>
>
>
>
> ----- Original Message -----
> From: Kelly Bellis
> Date: Wednesday, November 24, 2010 8:41 am
> Subject: RE: WinGMM install
> To: win...@googlegroups.com
>
>
>
> The command prompt is no longer available from the start
> button... and the start button is no longer labeled as such
> unless you hover your mouse over the round Window logo in its
> stead; however, under Accessories, the command prompt is still
> there :) - anyway... here's what I got:
>
>
>
>
>
>
>
> Kurt, Kelly:
>
> Can you run regserver from the command line?
>
> Try this: from the windows start button, click RUN.
>
> In the popup box, type REGSVR32 [path]\tabctl32.ocx
>
> Let me know Kurt.
>
> FWIW, I’m having fits with a Microsoft .NET 4.0 webbrowser
> thingy that is preventing my immediate distribution of a new
> application: QuadFinder Pro. … anyone care to try it out?
>
>
>
> From: win...@googlegroups.com [ mailto:win...@googlegroups.com]
> On Behalf Of Kelly Bellis
> Sent: Tuesday, November 23, 2010 10:12 AM
> To: win...@googlegroups.com; Win...@googlegroups.com
> Subject: Re: WinGMM install
>
>
> Try Windows XP Mode in the Virtual PC. It's the only way I found
> to run WinCMM, a slightly different animal, on my Win7 64-bit machine.
>
>
> At 11:41 AM 11/23/2010, Kurt Luebke wrote:
>
> I hope I have a simple question for somebody. My WinGMM blew up
> on me a few weeks ago and I have installed, uninstalled and re-
> installed WinGMM many times trying to get it to run correctly.
> Whenever I go to the options to adjust the defaults I get the
> following message:
>
>
>
> V. Kelly Bellis, PLS 2099
> Horizon Surveying Company, Inc.
> 17 Union Street
> Ellsworth, ME 04605
> 207.667.6912
> --
> You received this message because you are subscribed to the
> Google Groups "Measurement Management BBS" group.
> To post to this group, send email to win...@googlegroups.com.
> To unsubscribe from this group, send email to
>
> --
> You received this message because you are subscribed to the
> Google Groups "Measurement Management BBS" group.
> To post to this group, send email to win...@googlegroups.com.
> To unsubscribe from this group, send email to
>
>
> V. Kelly Bellis, PLS 2099
> Horizon Surveying Company, Inc.
> 17 Union Street
> Ellsworth, ME 04605
> 207.667.6912
>
> --
> You received this message because you are subscribed to the
> Google Groups "Measurement Management BBS" group.
> To post to this group, send email to win...@googlegroups.com.
> To unsubscribe from this group, send email to
>
> --
> You received this message because you are subscribed to the
> Google Groups "Measurement Management BBS" group.
> To post to this group, send email to win...@googlegroups.com.
> To unsubscribe from this group, send email to
> group at http://groups.google.com/group/wingmm?hl=en.
>
>
