Least Squares Adjustment Training
Posted by Chief on April 30, 2019 at 8:15 pmDoes anyone know of any training being offered for this? We use Leica Infinity for most traverse(compass rule) and would like to learn how to correctly process the network adjustments and anything else that is necessary for a well made network. Star net is a option if there is a good trainer for that also.
MemberApril 30, 2019 at 10:46 pmhttps://www.microsurvey.com/training/microsurveycenter/ There facility is in Kelowna, British Columbia, Canada.
Not sure what background you have but there are lots of on-line resources on the subject of the least squares adjustment of survey data. Of course it is important to have enough redundancy in your work to allow use of the method. In other words, if you observe a traverse originating at one point to another point both with known values but only include the minimal number of observations required, not much can be done.
If you are developing a network with redundancy and more than the minimal number of observations and ties, least squares provides the best solution. The least squares process criteria is that the sum of the squares of the residuals is at a minimum in the best solution.
Lots of terminology to learn. I have not worked with terrestrial observations in years and in some ways it is more complicated than dealing with GPS observations. Take a look at the publications on the NGS site: https://www.ngs.noaa.gov/PUBS_LIB/pub_index.html Some of the older publications deal with terrestrial observations and are written to be understandable by many surveyors. I will argue that publications like: https://www.ngs.noaa.gov/FGCS/tech_pub/1984-stds-specs-geodetic-control-networks.htm still have a lot of good information.
On the issue of designing a network or project to meet a specified standards, many efficiencies are possible when planning your project. See: https://www.degruyter.com/view/j/jogs.2014.4.issue-1/jogs-2014-0005/jogs-2014-0005.xml and his book. Many software packages include simulation capability allowing you to run pseudo observations to plan your project.
Many aspects to the issue of adjusting survey data; methods other than least squares.
Dr Ghilani??s textbook, Adjustment Computations, is a good start. There are even adjustment texts available free online. Even some advanced texts like: https://www.academia.edu/661712/Adjustment_theory BTW, CF Gauss?? classic text on the subject is available see: https://books.google.com/books?id=0NdsqjwLPdAC&pg=PA139&dq=cf+gauss+adjustment+of+observations&hl=en&sa=X&ved=0ahUKEwj2uoS-8_jhAhXDN30KHRkuBe8Q6AEIMTAC#v=onepage&q=cf%20gauss%20adjustment%20of%20observations&f=false
Perhaps someone familiar with the Microsurvey training will pipe in with their opinion. I took a few courses on the subject many years ago. Lots of old texts are available from used book dealers at low prices.
MemberApril 30, 2019 at 10:57 pmBefore you get down the StarNet road very far I’d look into the availability of data converters for your Leica raw data. That could be a snag. If so, Leica has software (they used to call it LGO, but I understand that has been replaced) to do LS adjustment.
Call up Microsurvey and ask. I gather that it is not a gigantic operation – probably about on par with the average survey office. Leica , on the other hand, is. Good luck with that ….
MemberMay 1, 2019 at 12:30 amInfinity is their new software and it’s not bad. We have had some difficulty processing traverse with it. We have a good sized group of LSs and know what we want and how to do it on paper if necessary. It’s the execution with the software. Some of our guys have had Infinity training provided by Leica and they were fairly underwhelmed. Our facility has a great training setup so we would probably want someone to come to us. Not afraid to pay for good training.
MemberMay 1, 2019 at 2:23 pmThere are college level courses on least squares adjustment. The course I took used Wolf and Ghilani’s text “ADJUSTMENT COMPUTATIONS, Statistics and Least Squares in Surveying and GIS”.
Since Chuck Ghilani retired from teaching, PSU-Wilkes-Barre is currently not offering on line surveying courses. I just spoke with the program coordinator and they do have a new instructor on staff, but the on-line will probably not restart until Fall 2020.
Paul in PA
MemberMay 1, 2019 at 3:14 pmIt’s going to be tough finding dealer-trainers who specifically focus on least squares adjustments.
If you have people who already understand least squares adjustments in general, Infinity adjustment reports do give all the necessary statistical tests to analyze networks; it is just a matter of finding them and knowing what is important and what is not. If the user doesn’t understand all the terminology and significance of the various results then it’s tough to get an idea of what it all means.
Infinity, like a lot of Leica software, does the job but in a fairly opaque and user-unfriendly way. Despite the updated interface, to me Infinity still has the feel of LGO, i.e. a statistics program rather than a geospatial processing program. But it does give plenty of information on the adjustment, and things that most surveyors are looking for (like absolute and relative error ellipses, standardized residuals, chi-square results) are all there.
There aren’t that many folks training who really understand least squares beyond “red text means failed test, push this button to make it go away”. On the flip side, I used to do certified adjustment training (not on Leica) and I had a very tough time explaining all the facets of least squares to trainees who were unfamiliar with basic stats, let alone network analysis and error trapping. It takes understanding the theory first (whether through university courses or self-taught) and then a fair amount of time getting experience with real data.
I second the recommendation for Ghilani’s Adjustment Computations – hands down the most comprehensive book for geospatial professionals doing LS adjustments. The NGS library is a gold mine too. I would also add to that a book on introductory statistics.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman
MemberMay 1, 2019 at 3:52 pmFor an overview, the training provided by StarNet’s documentation and tutorials is a great introduction. It does not get deeply into who and how the software works, but it does provide a good “best practices” starting point. There is a demo version available, so it would essentially be no cost.
As MicroSurvey/Starnet is part of the Hexagon family, it seems likely that it would work with Leica.
-All thoughts my own, except my typos and when I am wrong.
MemberMay 1, 2019 at 4:55 pmThanks for all the info guys. We have a decent understanding of the statistical part of it and some of us have taken classes on it, but??s it??s more the application and use of the programs that we have that we need training with. You can understand theory all day long, but if you don??t know what button to click or settings to check, then you are just spinning your wheels.
MemberMay 1, 2019 at 5:39 pmA least squares program is a nbd. If you understand compass rule, especially if you have done it by hand the least squares program will be easy to move over too.
I got starnet back in the early 1990’s if I remember correctly, it wasn’t a big deal to learn at all. The math behind it is a different issue, but for the program, basically you imput the data and pick the parameters it kicks out reports and adjusts the values.
I haven’t used star-net for quite a while but the programs I do use will create a report that is very impressive, sometimes it’s tens of pages long for a handful of points, you can’t beat that. I look it over, put it in the job file.
If you want to learn the math, that is one thing, but learning the programs should be easy for you. (it’s not that hard)
The thing about least squares vs. compass rule it that it’s way more versatile, making cross connections, allowing you to weight data properly, analysis of your traverses.
MemberMay 1, 2019 at 6:01 pm> analysis of your traverses.
The best thing about LS is that it is not limited to simple traverses as most other adjustment methods are. Redundant measurements and cross-connections are simple to include and can significantly improve the results. Giving you estimated accuracies (error ellipses) for the final points is a bonus.
MemberMay 1, 2019 at 7:59 pmOhio State has published the notes on their classic Adjustment Computations classes as given by Burkhard Schaffrin here:
I looked at the first chapter and it looks pretty advanced to me. I took a lower level course given by a visiting professor (Haim Papo), but I don’t remember what textbook we used. I just remember squeaking through tests and then having the light dawn the following week. Rinse, repeat. Prof. Papo told me that he was sorry that he had to give me a B (or maybe a B-) as it was obvious I was working hard. I was thrilled, because I was sure that I was going to get a C…or worse.
I’m just one of those evil GIS people. Bwah-hah-hah! Seriously, I do coordinate systems and transformations at Esri.
MemberMay 1, 2019 at 8:13 pmDr Ghilani has greatly contributed to the surveying community and continues to do so. I recollect he wrote a series of articles for the XYHght magazine. See: https://www.xyht.com/series/where-theory-meets-practice/page/4/
I found the magazine’s search function lacking when attempting to find all his articles on the subject of least squares but it appears with a little searching it is possible to find them all.
Good reading from a good source. Short enough to convey the information in short easy-to-digest articles.
I do not know whether he would consider it but he might be willing to provide on-site or host training. He delivered a four or five day “for-for-credit course” covering his Adjustment Computations book for a group of us in the late 90’s. I haven’t checked the PSU-WB site to see whether the software he developed is still available. He is in emeritus status now. Perhaps enjoying retirement too much to consider the idea of delivering a short course.
He is an excellent communicator. While trained in the Hirvonen/Mikhail era, I appreciated his book and the course.
BTW, I left behind most of my geodesy library when I retired and feeling nostalgic decided to see if I could obtain another copy of Mikhail’s Observations and Least Squares. Amazon links to a number of sources wanting >$100 for poor condition versions with good quality going for over $300. Sheesh…
MemberMay 1, 2019 at 10:23 pmThanks for the link. It is some great stuff. I will read it with great interest. But is not likely that I will understand much of the math. It’s great that you do. Because, luckily, it is not necessary for users of Least Squares adjustment software to understand all this math as long as the programmers do. It is only necessary to visualize the concepts.
MemberMay 12, 2019 at 6:51 pmI would look into StarNet software and videos unless you’re looking at how to form matrices based on observation equations. If it is the latter, I’d look for a college level class
MemberSeptember 20, 2019 at 4:20 pmThe LSA program that runs inside LGO is called Move3 and works very well in LGO, but not so well inside Infinity as far as I have been able to make sense of it as late as a couple years ago when I tried it. May be they sorted all that out by now? Or, you could just use Move3 itself. Move3 will import nearly any manufacture’s file formats directly.
MemberSeptember 21, 2019 at 2:24 pmLarry Phipps (RIP) used to teach LSA with Carlson’s Survnet software. I learned a lot from him. I don’t know if anyone filled his shoes or not. Larry’s company was Land Surveyors Workshops, if I remember correctly.
Freedom is just another word for nothing left to lose
MemberSeptember 21, 2019 at 3:56 pmI do not dispute the fact that Dr. Ghilani has contributed to the surveying community, but I will tell you as a recent student who used his Elementary Surveying ($200) book last year, I was very disappointed. Why? Because there were so many mistakes in the answers in the back of the book. I actually wrote him and Pearson Publishing; he did respond to a couple of my emails. If you’re a student and trying to learn, when the answers in the back of the book are wrong, besides wasting a lot of time, you get confused. The publishers change the numbers in new editions, as a way to prevent cheating and force new sales. That’s okay with me. They’re a for-profit enterprise. But to have wrong answers, it’s really bad. They could just a pay a couple of PhD students a few dollars more, and have them double-check the numbers. I go to a small school, there is ONE instructor, NO TAs, so you’re really on your own for learning unless you want to constantly bug the one instructor, which I didn’t want to do. I would end up looking at previous editions online that had the answers, figuring those out as a way to understand I was solving the problem right, and that the new edition was wrong.
I do not know Leica, but as someone who works in a technical industry where we offer training, you may want to consider going back to Leica, and asking for a higher quality training. I am sure there are people within Leica that can deliver that. Or alternatively, see if you can send your LSs to their office to get better training. Your question seems less to do with the concepts and math of least squares, and more of what is this piece of software doing behind the scenes. So they may be in the best position to help with that. Go back to Leica, explain your experience level, what you want to learn, and ask for a quote.
MemberSeptember 21, 2019 at 6:04 pmThe following is available on line for less than $30 as an e-book, that is less than I paid for the 1997 addition 20 years ago. My edition came with a 3.5″ DOS disk of an accompanying adjustment computations program. As I recall later editions came with a CD. I recall doing adjustments with that, that gave results within a 0.001′ of full blown StarNet solutions. The documentation for the program is the last 30 pages of the text.
A text book is to teach you the right methods. Having the exact right answers is not always that necessary.
Paul in PA
Title: Adjustment Computations : Spatial Data Analysis Author(s): Charles D. Ghilani Edition: 6 Year: 2018 ISBN-13: 9781119385981 (978-1-119-38598-1); 9781119390312 (978-1-119-39031-2) ISBN-10: 1119385989 (1-119-38598-9); 1119390311 (1-119-39031-1) Adjustment Computations 6th 6E Ghilani
E-Book: Digital version only, No Access Card/ Code.
Format: Searchable PDF, can print physical copy.
Duration: No expiry date, Use forever.
Delivery: Instant Online Storage Download.
MemberSeptember 21, 2019 at 6:38 pmI appreciate your aggravation with paying so much for a text that has not been throughly checked. I do not know whether there is an errata web page.
While familiar with the Elementary Surveying text (I was one of the many listed in the acknowledgements for the 10th (?) edition), my posting addressed his Adjustment Computations text that I found to be clear and easy to follow.
Now that this thread has re-emerged, I checked the PSU-WB surveying program site and found that the software developed by Dr Ghilani was still available. See: https://wilkesbarre.psu.edu/academics/surveying/free-resources
In closing, why the hesitance to approach the instructor? When I taught I was always frustrated that more students did not take advantage of my office hours. Most people have different learning styles. More people than you suspect are having the same problem.
MemberSeptember 21, 2019 at 7:19 pmYou can make an account and go over the archived course here, very much worthwhile.
Log in to reply.