Compass Rule or Least Squares
C. Hobbs
I think with computers common-place nowadays its maybe just as easy to do a
least-squares adjustment on observations, since in my opinion it provide the
most rigorous anaylsis and distribution of errors...compass rule only make a
survey mathemetically close by theorizing that the magnitude of the error is
proportional to the magitude of the traverse leg and in my opinion does not
always reflect reality...
Daniel A. Renton
What do most surveyors prefer. Probably the compass rule because it is
easy to understand. I think a lot of surveyors go by the KISS theory,
(KEEP IT SIMPLE STUPID). I myself like to use a least squares
adjustment. I can see what is going on in the traverse. If you want to
balance a traverse right, statistically speaking, then the least squares
method it the way to go. Please send input on this topic.
Daniel A. Renton
PLS EIT
Tyler Parsons
In article <35152199...@pacifier.com>, dre...@pacifier.com
says...
Least Squares is fine if you know what it's doing, and it does it
properly.
There is at least one major CAD/Cogo product on the market that
implements LS in a strange way. They advertise a LS adjustment option,
but when you really read the sparse intructions, you find out that what
they are really doing is BALANCING THE ANGLES first, then doing some
type of LS adjustment. It appears that you can only do a single traverse
loop at a time, it is not really clear. You get no error elipses, or any
sort of information about the adjustment output. It really sounds like
they are doing a Crandall's rule adjustment.
I have this program, and I don't trust it to do LS adjustments. I use
the compass rule adjustment, which assumes distance and angular
measurements are roughly equivalent, a valid assumption with Total
Station. I am already using one program to collect and transfer data,
and another to do my drafting. I'll have to get another to do my
adjustments. From what I've heard, StarNet is the way to go.
Tyler Parsons, PLS
Corvallis, OR
remove xyzzy- to e-mail.
Kent McMillan
In article <6f39s7$8p8$1...@bashir.peak.org>, tpar...@peak-xyzzy.com (Tyler Parsons) says:
>
<snipped discussion of CAD program with ersatz least squares adjustment>
> From what I've heard, StarNet is the way to go.
>
>Tyler Parsons, PLS
I would not hesitate to recommend Star*Net to any land surveyor for
everyday use. I began using a very early release of it about 10
years ago and have been along for the ride as its entrepreneur/developers
turned it into a really nice and practical tool for land surveying.
I do occasionally use the Compass Rule (Bowditch's Rule), but not for
anything other than:
- the occasional compass traverse run along riparian boundaries, and
- adjusting small mathematical misclosures in record metes and bounds
descriptions (Star*Net is the tool of choice for analyzing significant
misclosures and discrepancies even in record descriptions).
The main reasons why I prefer least squares (I do not mean the early
20th century invention known as Crandall's Rule) are:
- the results derived by least squares adjustment can be significantly
better (more accurate) than may be gotten by the Compass Rule,
- Star*Net allows the surveyor to examine the uncertainties in the
final survey results (accuracy of derived coordinates, accuracy of
inversed bearings and distances between derived coordinates) in a
way that is very useful as a quality control tool,
- least squares enables a surveyor to use a variety of redundant
measurements that cannot be used in a Compass Rule adjustment. One of
my own favorite tricks is to measure angles to intersected objects like
distant antennas or warning lights on water towers. The directions
taken to an object from different traverse stations should all converge
to the same point. Star*Net can use that fact as an additional
condition of the adjustment, and in the process derive an estimate
of the coordinates of the intersected thing that is good enough to use
as a backsight for work anywhere on the project. If the coordinates of
the distant object are already known, then that fact may be used in the
adjustment to considerably improve the accuracy of the traverse.
The best thing about Star*Net, though, is that it is very easy to learn
to use and to teach others to use.
Best regards,
Kent McMillan, RPLS
Austin TX
RSCHAUT
I have always felt the Crandall Rule to be the most accurate. When you have
short distance sights, the angular error is less reliable than the distance and
the opposite is true for long distance sights when most distances are less than
a quarter mile. Also, when you make separate angle corrections, you have the
option of holding two or more bearings the same as a record when the distance
corrections are within the probable error of your work.
Least squares makes no evaluation for unique site conditions which the
instrumentman will be aware of and which can be included in a Crandall Rule
adjustment.
rschaut.
Jim Frame
In article <6f39s7$8p8$1...@bashir.peak.org>,
tpar...@peak-xyzzy.com (Tyler Parsons) wrote:
>I use
>the compass rule adjustment, which assumes distance and angular
>measurements are roughly equivalent, a valid assumption with Total
>Station.
This depends upon the gun and its use. For a typical modern total
station (3" angles, 3mm + 3ppm distance) things are about equal for
shots in the neighborhood of 750'. Below that the angles are better
than distances (though instrument and target centering become more
critical); above that and the distances begin to outstrip the angles for
accuracy. By the time you get to 2,500' legs, distances are twice as
accurate as angles.
>From what I've heard, StarNet is the way to go.
Is Star*Net available in a Windows version yet? The DOS version I saw
last was a bit clunky on the interface side, but very good in terms of
flexibility and reporting.
---------------------------------------------------------------------
Jim Frame jhf...@dcn.davis.ca.us (530) 756-8584 756-8201 (FAX)
Frame Surveying & Mapping 609 A Street Davis, CA 95616
-----------------------< Davis Community Network >-------------------
Doug Bruce
RSCHAUT writes:
The weighted least squares technique DOES allow certain bearings (or
distances) to be constrained to particular values, if you feel that is
appropriate. It also DOES allow the weighting of certain observations to be
of greater or lesser value in the adjustment, if they have been reported to be
so by your field crew. I think StarNet has implemented both of these options.
(Perhaps a user of the current version should comment - Kent?)
- Doug Bruce
Ed Vincent
RSCHAUT wrote:
> I have always felt the Crandall Rule to be the most accurate. When you have
> short distance sights, the angular error is less reliable than the distance and
> the opposite is true for long distance sights when most distances are less than
> a quarter mile. Also, when you make separate angle corrections, you have the
> option of holding two or more bearings the same as a record when the distance
> corrections are within the probable error of your work.
> Least squares makes no evaluation for unique site conditions which the
> instrumentman will be aware of and which can be included in a Crandall Rule
> adjustment.
>
> rschaut.
Crandall Rule is scary. If I remember my Davis Foote correctly it was developed to
balance stadia traverses with the given that bearings are absolute and distance
gives. If you have a traverse North up a road for a couple of miles and return and
your closure is good in northing and you have 0.05' in easting it will change the
distances in your traverse by feet (create a four leg traverse with these
conditions and see). With our equipment now days I doubt that distances need to be
changed very much.
Ed Vincent, P.L.S.
Phx, AZ
Mark C. Williams
>What do most surveyors prefer. Probably the compass rule because it is
>easy to understand. I think a lot of surveyors go by the KISS theory,
>(KEEP IT SIMPLE STUPID). I myself like to use a least squares
>adjustment. I can see what is going on in the traverse. If you want to
>balance a traverse right, statistically speaking, then the least squares
>method it the way to go. Please send input on this topic.
>
>Daniel A. Renton
>PLS EIT
>
When I adjust a traverse I like to first make sure it closes within a
reasonable amount. If it doesn't then I need to do some checking in
my notes or run it again. When we do a traverse I like to turn angles
to a water tower or something as we go so that it is easier to
issolate errors. If an error is found then I will repeat the portion
of the traverse that is in error BEFORE going back to the office if at
all possible.
To actually adjust out the error, I look at the direction that it
needs to be adjusted and round the seconds to the nearest 5 or 10" and
nearest 0.05 or 0.1' weighting the rounding so that the proper
percentage goes the correct way. I also take into account
difficulties involved in the performing of the traverse. Which legs
had bad heatwaves, variations in wind speed and direction, which crops
were being irrigated, etc. It is amazing, but if you can generally
remember factors that would cause your angles to be either big or
small, as well as distances, simply round in this direction and you
might be surprised.
I have often used this method by noting in the fieldbook which
direction and possible magnitudes of deviation were probable. Once I
used this method 'on the fly' while doing a 6 mile closed traverse
with nasty heat waves and closed it to about 0.1' with a 10"
instrument.
To answer the question though... I don't use any standard adjusting
methods.
Mark Williams
Russ
Mark,
I feel that your adjustment approach, i.e. considering the field
conditions existing at the time each leg was measured, is preferable to
a more mechanical adjustment, be it least squares, compass rule, or
other.
Just one LS's opinion; what do you think, Jim?
Tyler Parsons
In article <351594D4...@swlink.net>, sr...@swlink.net says...
Davis, Foote, Anderson, and Mikhail's Surveying Theory and Practice, 6th
Edition, page 322-323, had a good evaluation of the different method of
adjustments, including the assumptions that are valid for the various
rules. Transit violates most angle/distance measurements, unless they are
made parallel to the grid system. Compass assumes angles and distances
are roughly equivalent in precision. Crandall is more rigorous but
requires more computations, and is not pursued further in the book.
I want to be able to check the adjustment procedure before applying it. I
have found significant errors in some of the programs available for
surveyors, and it's my rear end, not their's. Ever read the disclaimers
on the $3000+ software packages?
Tyler Parsons, PLS
Corvallis, OR
remove xyzzy for e-mail replies.
Jim Frame
Somebody said:
> Least squares makes no evaluation for unique site conditions which
> the instrumentman will be aware of and which can be included in a
> Crandall Rule adjustment.
A good least squares package (Star*Net, for example) allows any
individual measurement value to be weighted appropriately.
> With our equipment now days I doubt that distances need to be
> changed very much.
In general, I agree, and a competent least squares adjustment will
reflect this when it's true.
> I want to be able to check the adjustment procedure before applying
> it.
Absolutely. Adjustment software that doesn't provide adequate
(read: copious) statistical reporting is a liability. This is
especially true for the least squares method, as a measurement or data
entry blunder will not necessarily preclude a solution, though it
usually produces one with unacceptably (and sometimes absurdly) large
residuals. Review of residuals is a critical step in evaluating the
adequacy of any network adjustment solution.
Kent McMillan
In article <bruce.44....@osu.edu>, bruc...@osu.edu (Doug Bruce) says:
>
>RSCHAUT writes:
>
>> I have always felt the Crandall Rule to be the most accurate. When you have
>>short distance sights, the angular error is less reliable than the distance and
>>the opposite is true for long distance sights when most distances are less than
>>a quarter mile. Also, when you make separate angle corrections, you have the
>>option of holding two or more bearings the same as a record when the distance
>>corrections are within the probable error of your work.
>>instrumentman will be aware of and which can be included in a Crandall Rule
>>adjustment.
>
>distances) to be constrained to particular values, if you feel that is
>appropriate. It also DOES allow the weighting of certain observations to be
>of greater or lesser value in the adjustment, if they have been reported to be
>so by your field crew. I think StarNet has implemented both of these options.
> (Perhaps a user of the current version should comment - Kent?)
>
>- Doug Bruce
>
The real power of rigorous least squares is its flexibility. I would urge
any surveyor who has formed an opinion about least squares without having
actually adjusted any of his/her work with it to get the demo version of
Star*Net to try it out.
For the case that Mr. Rschaut describes, Star*Net does allow the surveyor to
hold the bearings between any points positioned by the survey at fixed
values. For example, in a large network you may have observed azimuths
along different lines. Plainly you would expect the adjustment to take
account of that condition and not change the bearing of a line of known
high accuracy by an unreasonable amount. I cannot myself imagine why
one would want to arbitrarily fix the directions of lines unless they
were the result of survey measurements, but Star*Net does also allow such
arbitrary manipulation of the data. If the manipulation is unreasonable,
i.e. excessively inconsistent with the survey measurements, then the
adjustment output will tell you that, in effect, through the statistical
testing of the adjustment.
A more familiar (to me) case of selective weighting of measurements is that
described by Mark Williams, certain angles having greater uncertainties
(surveyor-statistics-speak: "higher standard errors"). Star*Net in particular
allows the poorer quality measurements to be treated as such, so that the
adjustment can make larger changes to those observations than would be
probable if the measurements had been of normal quality.
In short, least squares does allow the surveyor who longs to tinker with
observations the chance to do it. However, I have found the real power of
least squares to be that it gives high quality, realistic results without
the sort of endless cutting and fitting of survey data.
For example, I use a total station that measures angles with a standard
error of +/- 3 seconds. I occasionally use a one-second theodolite that
measures angles with a standard error of +/- 1.2". The EDM on the total
station measures distances with a standard error of about +/-3mm + 3 ppm.
I can center the instruments over a ground mark with an centering accuracy
of about 1mm (standard error). The target system I use has a centering/
plumbing accuracy of about 1mm under most conditions. When I run a
conventional adjustment of survey measurements made with that equipment,
those are the standard errors that I input into Star*Net. It's a fairly
impersonal process so that when if adjustment (as rarely happens) is kicked
back with a red flag, it means that something out of the ordinary has
happened that needs to be examined more closely. It does not mean its time
to tinker with the data.
CoPLS
>> Mark Williams wrote
>When I adjust a traverse I like to first make sure it closes within a
>reasonable amount. If it doesn't then I need to do some checking in
>my notes or run it again
I recall an 'oldtimer' telling me; concerning the balancing of field work; -
"I don't get it, If a traverese closes, then why even balance it" -
I also recall a situation early on in my surveying, when I was a rear chainman
(there's a term I haven't heard in a while), the Boss had a traverse that did
not close by a lot (about 1:500), He said "don't wory about it, we'll just
balance it"... and with that field work, we laid out a subdivsion and wrote
many descriptions.
Food for thought - And with that I have always been cautious when balancing any
work, and to answer the question; I stand by Mark Williams, he stated; "I don't
use any standard adjusting methods." Me Too.
Ron CoPLS
JGRACETTE
Speaking of a good least squares adjustment package. Has anyone ever heard of
Cadastral Measurement Management? What do you think of it?
JGR
Kent McMillan
It occurs to me that I should have mentioned that any surveyor
interested in Star*Net may contact its developer Ron Sawyer by
e-mail at:
The least expensive version of Star*Net is still quite powerful,
handling up to 200 traverse stations and an unlimited number of
sideshots.
Disclaimer: Ron Sawyer is not my brother-in-law and I don't get
anything out of offering this recommendation other than the
satisfaction of having done the right thing.
Russ
I have an old (vers. 4-1991) copy of Star*Net which I would like to
sell, if this is legal. Reason? Retired!
SKIPPOM
>Speaking of a good least squares adjustment package. Has anyone ever heard
>of
>Cadastral Measurement Management? What do you think of it?
Jerry Wahl, Are you out there?
The boss swears by it. It is not the easiest program to use, but does a lot of
neat stuff. The manual imput is kinda clunky, but you can download a TDS or
SMI data collector or you can download BLM collector called Arizona (?) for
HP-200.
CMM has a program called INREC in which you enter GLO record measurements off
of the township plat and it will proportion all of the corners it can from the
existing monuments. This is great for search possisions.
CMM is a DOS based program, and you may run into memory problems if you are
running Windows.
I have never used Star-net so I can't compare the two, But It is definitely 'a
bargain at twice the price' (FREE) download yours today at www.cadastral.com
Don Pomeroy, PLS
Muried A. Pepling, LS,PP
You are doing the right thing, Ron is a real conscientious guy and
really does care about the practicing surveyor.
col...@echoweb.net
I have been using Star*net for about 5 years, and you couldn't get me to
switch for anything!!! I have also used Trimnet, TDS, & SoftDesk/DCA least
squares packages & none are as friendly as Star*net, & only Trimnet has the
informational output that compares. The very BEST thing about least squares
(IMHO) is that I no longer think "traverse", now I think "network" - all of
those extra measurements that used to be "check" shots can be included in
the adjustment. & no more "well if I adjust this loop 1st & then adjust that
(attached) loop, it works better than if I adjust that loop 1st & then this
loop". Whew, am I glad those days are gone! There is-was-used-to-be a
Star*net confrence on the Professional Surveyor BBS that was attended by
several very intelligent people (& more than a few ignorant soles). I have
not looked in quite a while, but it was very interesting & informative. Take
care, Bill Colisch PLS
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
CAD Technician
A very capable program and the price is right!
E Johns
JGRACETTE wrote in message
<199803230442...@ladder03.news.aol.com>...
>Speaking of a good least squares adjustment package. Has anyone ever heard
of
>Cadastral Measurement Management? What do you think of it?
>
>JGR
JGRACETTE
One of the features of Cadastral Measurement Management (CMM) allows the
operator to create a standard deviation file ("jobname".sd file). This file
contains the error estimates the surveyor believes to be reasonable considering
the field conditions.
If you run line through a swamp, you would not expect the error estimates to be
the same as if you were running line along a road. CMM allows you to apply
these error estimates to the measurements accordingly. If the entire job is
similar, then the operator may apply "default" error estimates which he feels
are appropriate.
Someone mentioned portions of CMM are a bit awkward. Aren't most programs when
you use them for the first time?
JGR
Michael G. Shackelford, PLS
Russ (retired):
That's good work if you can get it! <G> No, but seriously, how about
donating the program to one of these needy surveying students or schools.
You don't have to worry about if it is OK, and you might get a tax write-off
for it and it will make you feel good to give to a needy young surveying
student. If not, and you really need the money, I will buy it from you and
donate it myself.
Michael G. Shackelford, PLS
mcs...@mcs.net
Russ wrote in message <3516FD...@mcn.org>...
David Ferguson
If you take a simple traverse (no network of observations) with angles and
distances averaged at each setup, the results will be very nearly the same using
the compass rule or least squares. The real advantage to least squares is the
adjustment of control networks.
David Ferguson, PLS
df...@bellsouth.net
Shelby H. Griggs, PLS
That is true many times in a practical sense. I have had this argument
before, if the answer is the same, what difference does it make which
adjustment you use? The answer of course is if you have cross-ties, check
shots, and multiple interlocking traverse (or level) loops. This is where
the beauty of LS starts to come into play. In a four sided traverse around a
block for instance, the final answer will be very close to the same with
either compass or LS adjustments. In this situation, even the blunder
snooping abilities of LS are limited. Redundancy must be present with the
proper weighting for the real power of LS to show itself. Because of these
requirements for redundancy I was recently told that the Oregon DOT (ODOT)
has gone almost exclusively to control networks, rather than traverses.
Regards,
--
Shelby H. Griggs, PLS
OrbiTech, LLC
GPS Surveying, Consulting and Training
P.O. Box 5305
Bend, OR 97708-5305
Voice / FAX: (541) 383-2715
Mobile: (541) 912-8530
E-mail: orbi...@transport.com
David Ferguson wrote in message <35202430...@bellsouth.net>...
mar...@gte.net
On 23 Mar 1998 04:42:29 GMT, jgra...@aol.com (JGRACETTE) wrote:
>Speaking of a good least squares adjustment package. Has anyone ever heard of
>Cadastral Measurement Management? What do you think of it?
>
>JGR
Now your talking ____REAL_____ least squares adjustment. You won't be
sorry you tried it. It might take you a few days to learn but after
that..........look out.
Mark