Bill93, post: 428025, member: 87 wrote: Why is that? I would think a good combination of angle and distance measurements would work fine.
I would certainly yield to you, Bill. It just seems that there's very little room for error in measuring those small acute angles. A little work with the law of cosines and some assumed lengths of sides and measures of angles shows that the figure is more robust than I thought it would be.
[USER=12714]@Steward Souten[/USER] seems to be asking more about the statistics than about the measuring. XYHT has a series written by Dr. Ghilani on error statistics. Here's a link to some of that material. Looking around that web site for other of Dr. Ghilani's articles may be helpful.
http://www.xyht.com/?s=ghilani+statistics&submit=Search
At the risk of sounding crude you are starting surveying without having the rudiments of a surveying education.
What you have is a problem in understanding Statistics. Surveying is the task of measuring the earth and to do it with the professional precision required nothing is measured only once. Take a distance measurement with an EDM and you are getting a statistically derived answer, even if you only pushed the button once. Take an angle measurement with an electronic total station and you are getting an answer based on the instrument reading at least two angles on a circle. Any measurement has error which can be minimized by taking multiple readings. You can do this personally or you can allow an instrument to do it for you, but it is crucial that you understand the process and what it takes to improve the process if necessary.
You can start by taking a Statistics course at a local community college, but they are typically only 3 credits and give you enough information to handle statistics in a non technical field. Technically schools (engineering/surveying) teach 4 credit Statistics course that can answer your full list of questions. Were you in a typical surveying curriculum, at some point the instructor has to begin by a least reviewing advanced Statistics and in some case actually teaching it before the surveying education can advance.
Statistics is a course best learned by doing, so it is difficult to just pick up a book or some advice on an internet site and be good to go. The first task in learning statistics is to measure a distance with a measuring device that is much shorter than that which is required, and then do the math to prove you have a good answer. Measuring is not counting which deals in whole units that almost everyone can agree on. Measuring is dealing with parts of units, which renders a variety of opinions.
Setting up and instrument over a point, then setting up a target over a point and then pushing a button does not account for all the errors that occur in such a simply described process.
You describe not understanding "the terms sample and population" which are the simple statistical descriptions of that part of Statistics dealing with whole quantity concepts which must be understood before getting into the partial quantities.
Paul in PA
MightyMoe, post: 428021, member: 700 wrote: I know surveyors who longhand calculated compass rule adjustments, dmd areas, average end area volumes, geodetic coordinates, solar observations, and on and on (heck I used to do all those, some of them almost everyday),,,,,,,,,
but I've never heard of anyone who sits down and does least squares,,,,,,,,,
It's a black box operation.
I saw it done once back in the late 80s.
By a PLS/CE that I worked with in Houma La.
We were having a hypothetical discussion about adjustments of surveys and record titles that would be part of a cadastral layer in a GIS map.
He proceeded to solve a test case problem with me at lunch one day while eating his tuna salad.
He used pencil and legal pad and an old funky TI.
He was LSU grad too. Very much a Cajun.
Back to original programming.
The compass rule was developed to be solved by manual calculations. It was at the times considered a least squares solution.
A simple modified compass rule can come within thousandths of matching LS computer results, because I have actually done it. The reasoning is not to adjust it all by lengths. Assuming some angular closure error, if 50-75% (i.e.67%) of the angular error is spread out to all angles before the length adjustment you get amazing results. Assuming a 33" angular closure error and 6 traverse stations, either a 3" or 4" correction will get you almost matching final coordinates. It does not matter if the closure angular error is degrees, minutes or seconds.
Least squares implies the test of the residuals which you can do by hand, not necessarily creating a LS matrix by hand. But then again I have done the latter as a classroom not a workroom problem.
Paul n PA
Take a look for flagged zenith angles, probably bad
hi/ht. Fixing this , most likely will fix the adjustment so that it passes the chi test. A bad hi/ht is a blunder that must be removed before a valid adjustment can be achieved.
MightyMoe, post: 428021, member: 700 wrote: I know surveyors who longhand calculated compass rule adjustments, dmd areas, average end area volumes, geodetic coordinates, solar observations, and on and on (heck I used to do all those, some of them almost everyday),,,,,,,,,
but I've never heard of anyone who sits down and does least squares,,,,,,,,,
It's a black box operation.
Hey Moe! I take exception to including "solar observations" in the same category as doing least squares long hand. Two very different things. Heck, I use the state of the art computerized computations to reduce solar observations. Solar observations are no different than any other kind (including GPS); what's at issue here is understanding the errors and how you assess them.
The figure in question is a bad way to start to comprehend network adjustment. No redundancy, absolute minim input. And severe angles.
You'd be better off working with a braced quadrilateral; a box with diagonals for starters. Then experiment. It'll either be real clear real fast, or you'll make mistakes.
Paul in PA, post: 428070, member: 236 wrote: The compass rule was developed to be solved by manual calculations. It was at the times considered a least squares solution.
A simple modified compass rule can come within thousandths of matching LS computer results, because I have actually done it. The reasoning is not to adjust it all by lengths. Assuming some angular closure error, if 50-75% (i.e.67%) of the angular error is spread out to all angles before the length adjustment you get amazing results. Assuming a 33" angular closure error and 6 traverse stations, either a 3" or 4" correction will get you almost matching final coordinates. It does not matter if the closure angular error is degrees, minutes or seconds.
Least squares implies the test of the residuals which you can do by hand, not necessarily creating a LS matrix by hand. But then again I have done the latter as a classroom not a workroom problem.
Paul n PA
Most of my old traverses will reduce very close to identical using least squares and compass rule, of course adding cross ties and adjusting them as a network will change the results. Which is one nice feature about least squares.