Total Station check on Triumph LS control
- Thread starter Adam
- Start date
ctompkins
New Member
You should really check with the other forum to make sure it passes the all knowing minds of the GPS experts on that site....Not sure if they have ever used RTK because after all it is really not all that reliable nor worthy of passing the StarNet Least Squares chi test....
FWIW...I find it amazing the results we have been getting.
FWIW...I find it amazing the results we have been getting.
I don't have many checks anymore with the total station. I just don't get it out enough to see how the LS compares. It just stays in the truck.
ctompkins
New Member
I get that too...Having now tested it against reliable monuments measured with a total station the error between them is so nominal it is almost mind blowing.
Good on you for testing it thoroughly. I would recommend any new user do that until it becomes absolutely routine. It is amazing.
Jim Frame
Well-Known Member
I most often use RTK to establish site control for boundary and topo. My SOP is to get 3-minute observations, with repeat observations on a different day. I then run through the control with a total station, and roll all the data into a single adjustment. The RTK vector residuals are usually less than 0.04' horizontal, and less than 0.1' vertical, as long as I'm in the open. The RTK vertical isn't tight enough for design-level hardscape topo around here (flatlands), and I couldn't get reliable RTK on most of the topo anyway due to buildings and trees, but it's great for establishing position, scale and orientation.
I agree about vertical, Jim.
It's a flat versus hilly thing. I stake sewer, drainage, curb, all the time with mine. I use it for hardscape topo as well. The only thing I don't use it for is pinning a foundation. I do use it for staking the excavation of the footers just not the pins in the footers.
Depends on the accuracy requirements of the project. If a tenth meets the needs then it's fine. I wouldn't go certifying hundredths without numerous observations.
Darren Clemons
Well-Known Member
I agree with Adam here. I've used GPS for years for sewer, drainage, etc. The knock has always been you can only get within "a tenth", but I myself have always seen much better than that, especially with this Javad system.
The last control I set for a construction company on a subdivision site we designed was five iron pins - 4 around the outside edges and one in the center. I took 600 epochs (2 minutes) of data on each point. My next visit to the site the foreman who'd localized to my control (with a Trimble system) specifically told me he'd never tied into control this tight. His residuals were 0.006 h and 0.016 v!! He even joked and asked, "can you guys follow us around and set ALL our control"? Now, obviously this was a bit of an anomaly, no way anything can typically repeat those kinds of numbers, but my "standard" error, in the open, in the vertical with this Javad is about 0.04' - at most. Now, add in any type of obstructions and it will definitely increase to a tenth quite quickly.
As far as horizontal, there's not even a question. I, like Shawn don't even get my total station out anymore. I, for one, would actually reverse the "check". Use the LS to "check" your total station points. The LS is much, much more accurate. It is quite amazing,like Ct said above, of the "doubters" in the survey profession who still truly don't believe in RTK. We actually ran into that very scenario last week. We were out on a job (wide open flatland by the way) and a local competitor's guys came trudging up the road with nothing but a metal detector and shovel, doing the "old time" type of recon even before they got the gun out.
They looked interestingly at our rover and asked what it was. We told them and they said "you mean you guys use GPS on boundary surveys"?
I honestly had to hold back from busting out laughing. I just replied, "yeah, we use it quite a bit".
So glad I don't ever have to go back to the "old days".
Wouldn't trade our LS's for ANY other equipment out there!!
Last edited:
Back a couple of years ago I did some accuracy tests with the Triumph-LS at various distances. I looked at 2.55 miles and 9.5 miles. The GNSS engines have seen some changes since then, so I can't say for certain that my accuracies from then would still apply, but I suspect that they would. I found that with 4 minute observations, 240 epochs, (this predated beast mode) the horizontal accuracy was 5mm+0.7ppm and the vertical accuracy was 10mm+1ppm at 68%. Outliers were observed at roughly 3x the standard deviation (which makes sense statistically). So it's not impossible to have a tenth of a foot vertically (about a 1% chance) in good conditions with 4 minutes of observation. You are much more likely to be within 0.03' vertically. Horizontally, you may see an occasional 0.05' in a position. Repeat observations are a good idea. Javad has an excellent average tool to average multiple observations on the same point to improve precision and minimize those occasional extreme errors.
This was of course in optimum conditions and with long occupations. I know RTK can provide really good coordinates and for most topographic work that I do, these accuracies are acceptable, but you have to be cautious when you add the strain of imperfect job sites with obstructions to the sky.
This was of course in optimum conditions and with long occupations. I know RTK can provide really good coordinates and for most topographic work that I do, these accuracies are acceptable, but you have to be cautious when you add the strain of imperfect job sites with obstructions to the sky.
Jim Frame
Well-Known Member
This begins to become true once you get out beyond a quarter mile or so between points, but for closer-in work a good total station is going to be significantly more accurate than RTK unless you repeat the RTK observation multiple times.
Below is a simplified plot of a network from a small project I did last year, with distances between points annotated. All of the points shown are open except for point 5, which has a large building about 10 feet behind it. The RTK base -- a GPS-only RTN station -- is about 1/4 mile away, and all points were observed twice on different days for 3 minutes per observation. The adjustments to the RTK vector components are mostly around 0.03' H and 0.04' V, with some (both H and V) closer to 0.1'. The total station distance residuals are all less than 0.01' except for one that shaded slightly over 0.01'.
Darren Clemons
Well-Known Member
Very true Jim. My statement, as a fact, isn't completely true of course. It's all about what we're referring to and where we're working.
I, for the most part, was referring to medium to large boundary surveys of 100 to 600 acres with major elevation changes and almost all in the woods. Very rough terrain and, if ran with a total station, several close together stations. With those types of traverses, the LS and RTK (good reliable and verified RTK) will almost always provide better overall results than a gun, no matter how tight the traverse is.
As Adam also said in an above post, I also do some building staking and will set corners for digging footers with the LS, but will not set the final corners in the footers with GPS. With close together points of 10' to 50', as you say, the 0.04' or so is just not good enough. Only way to do this "dead nuts" as the contractor's say, is set two, then start back sighting and turning 90 degree angles all way around the structure.
We also don't have to deal with the very small amounts of slope you describe where a half a tenth would be nowhere near good enough.
I have performed hundreds of topo's with GPS, but they are mostly in open areas over hilly terrain where future subdivisions will be designed, drawn at a contour interval of 2' where a half to a tenth is more than good enough.
Nate The Surveyor
Well-Known Member
It's just important to know the accuracy requirements of the job. I worked at O'hare airport. + - .03' was good enough.... BUT going up 0.03', and down 0.03' every 50 feet was TOTALLY unacceptable. There was a contractor who READ his contract, to mean that he could go up and down... NO!
they wound up hiring us to carefully survey out all the dips, and hills, and design a STRAIGHT line, and paint marks for GRINDING it all down, to the lowest valleys. This worked... But, in flat country, you have to be sure what specs mean....
they wound up hiring us to carefully survey out all the dips, and hills, and design a STRAIGHT line, and paint marks for GRINDING it all down, to the lowest valleys. This worked... But, in flat country, you have to be sure what specs mean....
Jim, that vertical seems ok. But I can't imagine horizontal approaching a 0.10' and averaging around 0.03'. I would expect 0.06' to be the very high side of horizontal (in the open) and typically around 0.02'. I know you are meticulous about poles and levels, so I won't even ask about that. I do wonder if you have tilt correction turned on in the LS, though.
Tilt correction, when properly calibrated provides accuracies as good as a bubble. However the tilt correction must be calibrated and the "level offset" must be applied. If not, the tilt correction can actually make positions worse. One other thing regarding tilt correction, the tilt sensors will add noise to the scatter plot during collection, but the noise is random and means out over time, but the cluster will look bigger. I can generally tell from the scatter plot when tilt is on and when it is off because of this.