Activity Feed › Discussion Forums › Strictly Surveying › Static GPS control and on the ground traverse
Static GPS control and on the ground traverse
sireath replied 4 years, 9 months ago 14 Members · 37 Replies
The PLS is not able.
So? Go get one that is.
“…primarily base plan topo to support our design staff. A lot of bridge work,..”
is the E&O insurance up to date?
If this contract was won low-bid, if I had bid on it and lost? Yeah, this thread would be entered into the record.
Hey all, I greatly appreciate all of the feedback so far, you’re all presenting great points and suggestions. From everything I’ve read, digested, and done to this point, I’ll add some more information and maybe some files to see if someone more familiar with these scenarios might be able to help.
General: Projection is SPC83-Massachusetts (Mainland)
This isn’t a highway job, it is a residential street with limited lines of sight. I set my points in an effort to have longish legs but also set points that would be functional in getting my topo done, so yes many are at corners of side streets. I likely could have skipped some setups if my sole intent was to run the traverse in as few legs as possible, but due to H&V curves the distances would have been limited.
I was given a pdf (attached) with what are listed as GRID coordinates for the 6 static GPS points. I started off by creating a crd file with the GRID coordinates for points 2313 and 2312 in it, calling them 100 and 101 respectively. These points were leveled through by the state from a benchmark, so I thought for vertical control that starting here would be the best bet. Those points checked to 0.006 for horizontal and 0.099 for vertical, so I said to myself, it’s under a hundredth, lets roll.
Not sure if this is a coincidence but when I multiply my measured distance (299.577) by the scale factor for the occupied point (0.999981762444169) I get the calculated distance. Should I be using a TS scale of 0.99998 in my dc for the entire thing? Should I have scaled each point by its scale factor before I started? I’ve tried doing that and it makes things worse, so I don’t think that is the case, but now I’m concerned that I’m totally out of whack due to scale factor and I’m going to have to re-turn this traverse.
Again I greatly appreciate each of you who have taken the time to respond. I’m feeling out of my depth there, and I want to get out of the office and start collecting. I wish someone in my office was familiar with this stuff, but sadly I am a one man crew working in an Engineering heavy company.
Attached: pdf with gps coordinates
.rw5 for the traverse
.dwg with points provided by the state, and my traverse broken in 2 as I mentioned in the OP.
The 2 .fbk files i created from my original raw file to make the traverse fit at it’s current rotation.
Best – Lucas
Edit: couldn’t add the drawing?
Moe that’s correct, i didn’t put out all of the points i will need to get it done, but i was trying to place them in useful positions @bill93
Yeah i found the carlson x-port gave me a better fit as far as just doing a simple least squares adjustment, but again that’s only asking me for one closure point, so the middle is off. c3d wasn’t much help at all. It sounds like I need better software in order to hold multiple known locations for the adjustment. I’ll check out starnet, thanks!
That is how we do it, you will have one every 300′ or so. Of course that all changes with GPS, I would occupy all the ones that GPS can occupy using the existing 6 control points as fixed assuming they check. We would level through the main control and run little traverses to the ones that need it a traverse run. Usually there will be a couple of hundredths of error for the side traverses and those may or may not get adjusted. The state DOT has printed a book explaining the procedures required for control, I figure your state DOT would also have one, you might give them a call.
Suggestion – I would not hold fixed any of the GPS points. I’d give them all a reasonable standard error, like 0.02′
For a divided highway I’d have control down both sides of the road at 500-700 foot intervals. The goal being to keep any topo shots from the control points at less that 500′, with overlap, and never having to shoot through both directions of traffic. 13000′, a double set of mons, at that interval, would come close to 59 stations.
A least-squares adjustment solves for a solution where it adjusts all of the measurements the “least”. It is a rigorous fit of your measurements and the estimated standard errors for each measurement to control, although even your control can be weighted for its estimated accuracy. However, most will hold fixed the control values for the adjustments. Weighting involves setting reasonable estimates of your measurement inaccuracies (GPS or conventional), centering errors, HI measurement errors, sighting errors (if using optical instruments). Its been around a long time, and is generally the most reliable and accurate adjustment. The more redundant measurements you have, like cross ties, or repeat measurements, the more the adjustment shines in the ability to determine the correct geometry of your points. There is a lot of information on this board, or on the interweb that can give you more information. Easy to use and relatively in-expensive programs like StarNet are available for everyone, and some have a free trial use. Most all of the GPS processing suites have LSA included in the package.
“Least squares” by its very name implies that the sum of the squares of the residuals is minimized.
It does not adjust the measurements “the least”.
It provides the most “most probable” value for every given measurement.
Have ran your .rw5 file and GPS control in Starnet and managed to pass the iteration. Pardon my knowledge in US Feet as I primarily deal in metres. So if my instrument settings are slightly out, its because it is convert from my metres setting.
From the GPS control PDF, it seems as if the coordinates have been given in grid coordinates and scaled accordingly already.
I had managed to fix your stn 100, 101, 158, 157, 129 & 130.
Your stn 130 was supposed to be DOTGPS2310 not GPS2311 as recorded in your raw data, after that edit the adjustmnet passed.
Adjustment Statistical Summary
==============================
Iterations = 3
Number of Stations = 62
Number of Observations = 363
Number of Unknowns = 112
Number of Redundant Obs = 251
Observation Count Sum Squares Error
of StdRes Factor
Angles 121 146.910 1.325
Distances 242 87.533 0.723
Total 363 234.443 0.966
The Chi-Square Test at 5.00% Level Passed
Lower/Upper Bounds (0.913/1.087)Forgot to mention this is in 2D. Normally I would prefer to work in 2D first to cut out any potential errors first.
Have managed to run starnet in 3D and managed to pass as well. It showed that your zeniths were not as well executed as your horizontals but still managed to pass.
Adjustment Statistical Summary
==============================
Iterations = 3
Number of Stations = 62
Number of Observations = 605
Number of Unknowns = 168
Number of Redundant Obs = 437
Observation Count Sum Squares Error
of StdRes Factor
Angles 121 24.333 0.528
Distances 242 46.413 0.515
Zeniths 242 346.617 1.408
Total 605 417.363 0.977
The Chi-Square Test at 5.00% Level Passed
Lower/Upper Bounds (0.934/1.066)From a purely mathematical perspective, there are two pieces of information regarding the nature of the coordinates. In the original post, measured distances were less than calculated distances. With scale factors less than 1, the opposite should be true if the coordinates are grid coordinates.
In his most recent post, the OP said that multiplying his measured distance by a scale factor (not the right one, but very close) produced the calculated distance. It’s a short line and the scale factor is close to one, so the difference is a few thousandths, probably insignificant if the coordinates are ground coordinates.
From an empirical point of view, the coordinates seem to be some sort of ground coordinates rather than grid coordinates. The method used to calculate them may be the source of some of the observed differences.
Thanks sireath! I’m going to download starnet as that seems to be a good solution, multiple posts have recommended it. Also, I forgot to edit a foresight height in the rw5 that is wrong and I had changed in the fieldbook file. Hopefully that will tighten the vertical control up a bit as well. Thanks again, have a great weekend!
Your data isn’t bad at all. Sireath has held your GPS points fixed (he could have reasonably given the points a small standard error, and thus created an additional place for the imperfections to ripple), and set the standards errors very tight. Merely relaxing the measure up standard error slightly, to a still very reasonable +/-0.01′ would probably bring your zeniths error factor down very close to the 1.0 you seek. The adjustment report shows very tight data in all respects. The adjustment is on a Local (non-projected) datum – the adjustment would likely be still better if it was on the appropriate projection.
Sireath must have invested some serious time to convert your fbk files to the StarNet format. AFAIK there is no automated way to do that. There are converters to change your Leica formatted data into StarNet. But nothing for fbk to StarNet. Before you send Microsurvey any money make sure you have the converter issue nailed down.
I have a RW5 converter so it was no trouble converting the RW5 file to Starnet. It was no trouble and gives me more experience in doing least squares. I like to keep it tight with plenty of redundancies and this allows me to set minimal or even fixed points.
Log in to reply.