Traverse Misclosure QuestionsPosted by sur04 on September 18, 2022 at 5:40 pm
I have some questions about misclosure on a closed loop traverse. I recently completed a closed loop traverse around a property. The traverse consisited of 35 stations and it was roughly 12000 feet in length. The linear misclosure in the end was around 1 foot and I would like to get it better than that. As a background, I set three control stations with RTK and started traversing. Also, I located most of the existing property corners with the RTK unit and decided to do this because most of the property corners were in the deep woods and I wanted to check the consistency of the GPS measurments against the total station points. While traversing, I doubled all angles and at some stations I doubled the angles twice. I haven’t worked with least squares and I wanted to try it in an attempt to look for a blunder to see if I can locate an error. I do have survnet and and wanted to try and work with it. Could someone give me some pointers while trying to do this.
- 13 Replies
- MemberSeptember 18, 2022 at 6:25 pm
Consult your equipment manufacturer’s manual(s) and identify the specs for the performance of your gear with respect to your method (RTK vs Static, Total Station, etc.) for accuracy/precision/error estimates. (Did you do any static observations?)
Key those values into your adjustment software.
Set realistic “set up errors” (centering/vertical ~ 0.005′ – 0.01′ (or whatever value is related to how well you keep your gear adjusted, and how careful you were during field work))
There may be a setting for using the weighted mean of observations, or individual sideshots. (if the individual measurements are tight, then the weighted mean is tight, it shouldn’t matter which one you pick.)
You should have an option for 2d or 3d; pick accordingly. (If you did measure ups, i recommend 3d, and the terrestrial will get you through the woods pretty well, while using RTK as a “check” for said woods points.)
Perform a “free adjustment” (no points fixed) to test the quality of your network; OR Fix the GNSS base from the RTK and let the adjustment do the rest.
If your error estimates are reasonable and field work is tight, the adjustment should come back with a reference factor close to one. If it’s higher than one, you have larger errors than you told it. If it’s lower than one, you overestimated your errors.
*edit* There are a lot more things to consider, but if you start here, the rest kinda falls into line if your fieldwork is fairly sound. I’m sure the others will fill in the gaps.
- MemberSeptember 18, 2022 at 6:42 pm
Did you have any short legs in the traverse? A little error there can leverage into a lot by the end of the loop..
- MemberSeptember 18, 2022 at 7:19 pmPosted by: @sur04
I set three control stations with RTK and started traversing
There are some folks who have much more expertise in processing LS data. I’m sure one of them will post soon to help with specific directions.
I do wonder about the starting method. 1/12,000 is pretty low for modern equipment and the doubled angle procedure you followed (of course the terrain can heavily impact that).
Did you set three RTK points and hold those coordinate values as absolute fixed values, set the TS up and set zero between two RTK points, then traverse around the site, and check back into those RTK coordinates?
or did you hold 1 RTK coordinate and the azimuth to a second RTK point (storing that backsight coordinate based on your TS observed distance along the RTK defined azimuth) to start your TS traverse?
- MemberSeptember 18, 2022 at 7:56 pm
My 1:12000 what is the angular accuracy of your equipment and what was the angular misclosure per set up and total? Where your starting rtk points intervisable or spread out throughout the traverse. If I know I am going to run through rtk control I try and space out those points as they do not need to be on one end or the other but spaced throughout the network when especially doing an least squares adjustment. Did you scale those plane coordinates to ground or leave them at grid state plane etc. if left on grid did you set the data collector up for your conventional traverse on grid as well. If not I suspect if your angular closure and all your doubled angles are good then you have a distance issue like grid vs ground at that distance. You might even see that on certain lines that are running close to straight the error ellipse all look identical in those situations and start to change as you make 45 degree or 90 degree turns on the traverse. Remember know matter how small the difference between grid and ground is at say a 300 to 600 ft line it will be compounded like a systematic error over the lenth total of your traverse. Hint why i stated after an unconstrained adjustment you might see the ellipse looking i ldentical over points in somewhat of straight lines. Also like stated above any short legs will increase as you move around. The control from rtk spread out throughout will often mitigate this because you have rtk and conventional which helps increase your redundancy. I have no idea about survnet but if that is made by carlson they have a great power point on this topic of something along the lines of ??why least squares vs compass etc. ?? I will try and look it up tomorrow and post the link. I am out checking cows today. Alan Dragoo also has a great power point that helps with showing how setting control thoughout vs starting and ending pairs etc. I will have to look that one up for you also. I just changed companies so my cheat sheets are all boxed up at the moment. I have not unpacked yet at new office. If survnet will allow you to use rtk and conventional all at once and such that can really help identify blunders on an unconstrained adjustment. Check the tao histogram if available. Only after you have done what the first post mentioned and done your best to identify the blunders and systematic errors prior too adjustment. Like angular misclosure distance issues like grid vs ground . Oh if you did a say grid to ground check the elevation of where that was vs elevation around the site. I just saw a traverse done and 200 ft of elevation change on site but elevation used was on one end and each quadrant had minimum of 100 ft in one end 200 on other from a quick check. Resulting in more than ppm needed to have a good average elevation factor. So distance were still affected. A small amount but more than needed with a little more care of getting the average elevation for better combined factor.
- MemberSeptember 18, 2022 at 9:43 pm
One way to find an angular error is to recalculate the traverse by changing your starting point to include every set-up. The starting point that gives little or no closure error will have the angle error. Tedious? Yes, but better than redoing the entire traverse in the field. Do not adjust the angles prior to doing the calculations.
- MemberSeptember 18, 2022 at 11:23 pm
@dave-lindell nice little trick for sure. another way is to take the deltas of N and E take arc tanjent to see what side that bearing points towards never really exact but usually has you withen the 3 points it could be at. Next is to spin a angle in cad off the mis closure and look at where it points. But first I would check the angular mis closure total vs allowable per set up based on instrument specs and care procedures. My guess is either short bs to fs or its all systematic distance errors. Could be the rtk that could be cked by assuming 1000 5000 and running latitude departures. And such to see if grid bs ground. I have ran into more surveyors lately that are thinking that least squares is some majical way to fix bad data. This is not good. It just helps in today??s world of intergrated measurements tools like rtk static levels total station all on same project. It does help identify errors quicker sometimes but it??s not a fix for bad data. I always say we can??t fix bad data only try and identify it. We have to also remember as we integrate rtk position it has an error ellipse at both the occupy and bs station if we set intervisable pairs. So if running a loop go back to basics and check the old formula N-2 x 180 for angles and sqrt of angular mis closure time set ups. This will point to angle or distance issues grid va ground prism constant ppm etc can then be cked. You have a nice quick way of identifying for sure.
- MemberSeptember 18, 2022 at 11:31 pm
Yes, as you say the perpendicular bisector of the misclosure course points to the point with an angular error, or one near it.
And a least squares adjustment only gives the most probable values for the data given.
- MemberSeptember 19, 2022 at 2:24 pm
Others have already pointed this out, but I would suspect short backsights/long foresights and steep terrain as the likely culprits if no blunders are suspected or detected.
You said you have RTK observations as well – I would be looking to use those in my least squares analysis as long as the solutions were fixed and reasonable.
I’m sure others may disagree, but least-squares adjustments – especially if you suspect there is a blunder in your dataset – are not exactly something you just “get pointers on” for the first time around.
There are a lot of articles and tutorials out there on the process (I really like this xyHt series), but it’s far too much information to convey in a forum post. I would recommend running through some of those tutorials and reading the articles, giving the LSA a shot and then coming back with any questions or concerns.“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman
- MemberSeptember 20, 2022 at 5:26 am
I want to thank everyone on the suggestions. I will explain a little more what I did. When I started the traverse, I had set three points using RTK GPS. All of theses points were located in a wide open field and observed twice for three minutes. I took one shot, reset the sattellites, and then rotated the rod and took a second three minute shot and averaged the coordinates. After that, I occupied the middle of the three points I set. My backsight distance check was 0.03. I felt that was acceptable. Holding that azimuth, I turned into the third point I set as a check and the turned in point checked at 0.03 also. After that, continued traversing setting new stations. While traversing, I did measure up at every station I set and carried elevations. As I traversed, I did have one paticularly short set up and I made a note of it in my field book. It was roughly roughly 125′ in length with a vertical drop of around 22′. When I set the next station from that point, it was a distance of roughly 180′. I was trying to keep my traverse lengths balanced as possible to avoid errors. At some of the stations I set, there were some higher vertical distance differences. One station had a 33′ vertical rise over roughly 320′, another was 25′ of rise over 408′. Another had 53′ of rise over 540′. The largest one I had was 75′ of vertical drop over 571′. Some of these points are located in an open field and I went back through with the GPS and located those ones to get a comparison between the points. This traverse is long east and west with very little north and south. While checking into some of the points towards the middle I was roughly 0.2′ off of my turned into coordinate. Also, I was checking into monuments that I located with the GPS and I was roughly 0.1′ to 0.2′ off of the coordinates I had. After traversing out of a second section of woods and back into a field, the coordinates were roughly 0.4′ to 0.6′ different from the GPS coordinates. As I turned south the coordiantes at one point was roughly 0.7′ different than the turned in point. As I traversed back to the POB, I traversed through my GPS base location and I was 0.07′ off from that coordinate, and the next few stations I checked into were roughly 0.4′ off of the turned in coordinates. At that point, the traverse was nearly complete. At the last two stations before the POB the error started to pick back up again until I ended up with 1′ turned in misclosure at the end. As I stated in the earlier post, all angles were doubled. So, as some have stated maybe the higher vertical distances at some observations could be causing the problem or the short set up could be it. Before starting this project, I collimated the total station and checked the rods and they are plumb. One again everyone thanks fo the input.
- MemberSeptember 20, 2022 at 1:53 pm
Start the traverse on the short line foresight. Then end the calculation on the short line back sight.
This will isolate the figure and eliminate error caused by accumulating a small error with the short line in the traverse.
Also you can run both directions from the original starting point to the short line, this will usually show the error is isolated to that line.
LSA programs will allow you to do this quickly, they really aren’t that difficult to use. But if you don’t want to pay the cost for the LSA program the one you’re using can isolate errors too, just more time-consuming. If you run lots of long traverses in deep woods you really need a LSA program of some kind simply for the speed and catalog abilities.
- MemberSeptember 20, 2022 at 2:10 pm
Those GPS vs. traverse deltas are really hard to analyze, not in the least because RTK will have some random float in it. At first it looks like they are increasing slowly, as one would expect for a traverse survey, but then when you turned back the errors got smaller (especially through that GPS base point, which is strange) except then they jumped up all of a sudden at the end.
I would either suspect that something went wonky at the end, there is a systematic or compensating error (but still doesn’t quite fit that theory), or the steep inclines just took their toll.
What equipment are you using? Specifically, does your instrument have a level compensator and has it been calibrated recently (either field or in the service center)?“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman
- MemberSeptember 20, 2022 at 2:38 pm
I??d love to see a plot of this traverse. That would really help guide this discussion
- MemberSeptember 20, 2022 at 8:47 pm
Also as you state the errors maybe break them up to North and East deltas around the site. I assume that at some point the greater error flipped from either north to east or east to north based on your writing even though they started to come back down for a bit and then towards the end jumped back up. As you wrapped around. You say project was mostly east to west. Was this in a transverse mercator state projection or lambert. And were you traversing on grid in data collector or ground vs rtk grid or ground. Steep angles for trunnion axis if your equipment has a procedure to account for that. So as the scope moves vertically away from the horizontal plane axis . But as we all know steep angles can bite us in the rump at times.
Log in to reply.