-
Static GPS control and on the ground traverse
Greetings all! First let me say if this is in the wrong forum I apologize, this is my first post here.
Background: I got back in to survey last fall after a near decade hiatus. The field stuff came back quick, re-learning CAD and C3D having only used LDD with Carlson Surv-CAD extension a decade prior took a bit, but I’m up and running. I do primarily base plan topo to support our design staff. A lot of bridge work, roadway projects, site plans for developers, etc, and I get a handful of small residential work in the door each month.
My current issue: We won a large (about 4.5 miles of topo including the side streets) contract to create a base plan for a section of road in a neighboring town. This job will eventually become a state funded project (hope to get the design contract as well) and as such needs to be to their standards. The state has provided us with 6 state plan coordinate control points along the project using static GPS. There are 2 at the beginning, 2 at the end, and 2 in the middle. I am using a tp12, AllegroMX with SurvCE, and Leica circle prisms and new trimax legs. Gun was calibrated at the end of last season, I take good care of it, and run check and adjust pretty regularly.
So I started at the north end of the job occupying the first 2 points (300’+/- apart, checked to under 0.01 vert and horiz) and began laying out my main traverse. My intent was to traverse the length of the road, tie into their GPS control, balance the error, and then all do all of my topo.
So, when I arrived at the middle set of GPS points and checked my coordinates versus theirs, I found myself about 1.33′ off horizontally, and 0.05′ vertically. When i finally got to the other end of the project, I am now 3.65 off H and 0.13′ off V.
This was pretty disappointing as I am turning quad angles, and the most error I ever saw when occupying a newly set point was 0.009 and that was for vertical, nothing more than 0.004 for horizontal.
The thing is, as the crow flies, I’m really not that bad. The calc from first to last GPS point is 13298.7899′, and my actual measurement is 13298.6387 (-0.1512). First point to mid point calc is 6128.8909, on the ground measured 6128.8229 (-.0680). So it would seem that the majority of my error is in the angles turned. When I draw a line from the first to last gps point, and a line from the first and last points of my traverse, I am about 0.0059 (d.mm.ss) to the right of where I should be according to their gps control. Holding either end point and rotating through the other end puts me about 0.45 out in the middle, which won’t work if they use my coordinates for layout years down the road when this thing will get built.
I had 59 legs on the traverse, and I was off 59 seconds to the right, so I went in and reduced each angle in my traverse by a second in the .fbk file, which got me about half way to where I needed to be. I again edited the .fbk by reducing each angle by another second, that leaves me off 0.07′ at the far end of the run, but still out 0.52 in the middle.
I also checked my direct reverse report and tried editing my original raw .fbk by the standard deviation numbers (highest SD was 4 seconds on a few legs >300ft) for each turned angle, and again I’m only about halfway back to where I should have landed by their coordinates. Not sure if I should do this again (2 standard deviations) and see how that ends up, or if I’m just faking it to make it. At least in this case I am distributing known error along the appropriate legs? Idk. I just don’t want to spend all summer knocking this topo out and end up with a junk survey…
The best result I have had is breaking the .fbk into 2. The first runs from the 2 notherly points to the middle, and then on the second I basically started over, occupying the known gps coordinates and running down to the end. Rotating those 2 separate runs by standard hold one end and rotate through the other puts me under 0.10 off max for any one point, but doing this I’m also just twisting the traverse to fit the coordinates, not balancing the error through each leg?
So, is there a way to hold multiple known points along an open traverse with equal weight and distribute the error throughout the traverse? I’ve tried using least squares in c3d and carlson x-port, and that gets me dead nuts on either end, but I’m still out of whack in the middle. I can’t seem to find any method in either that lets me specify known points I’d like to close on and then point it to where I’ve calculated those points to be based on my traverse.
If I had to do it again I would probably start at one end, work to the middle, then go to the other end, work back to the middle, and just have 2 separate traverses. At least then I’m just trying to hold 3 points for each and not 6 for one long as traverse.
So yeah, sorry for the novel, but I need help!. I’ve googled every term i can think of and come up with nothing of this nature. The PLS I’m working under is great for property line rotation questions I have, but he’s been out of the field for a long time, so he can’t really help me with this one. I’d rather not reach out to the state and admit that I’m not exactly sure what I’m doing, but if I have to I will swallow my pride and do so.
Thanks for reading and for any help you can offer!
LP
-
37 Replies
-
Starting with an azimuth on one end pair of points is sure to leave you off at the other end, and I wouldn’t adjust each and every angle to make up for that. Holding end points and figuring out what to do in the middle would be much better.
A mis-fit of 0.45 in 6128.8 is 1 : 13620 and probably within the distortion of a state plane coordinate system, depending on where you are relative to the center of the zone and which orientation your road has relative to the TM or Lambert. If that isn’t good enough, then the work should be done with a low distortion projection customized to the project.
. -
A properly weighted Least Squares Adjustment would probably be best suited for adjustments. Bill93 is correct too, that the SPC has its own distortion values in the projection.
-
Bill above is right.
Another thing comes to mind. Scale factors. Have you any experience with that? Do you know how their gps was done?
Nate
-
Bill above is right.
Another thing comes to mind. Scale factors. Have you any experience with that? Do you know how their gps was done?
Nate
-
Bill above is right.
Another thing comes to mind. Scale factors. Have you any experience with that? Do you know how their gps was done?
Nate
-
Bill above is right.
Another thing comes to mind. Scale factors. Have you any experience with that? Do you know how their gps was done?
Nate
-
I’m a StarNet aficionado. I suspect that your biggest problem right now is trying to adjust that in C3d and not something much better suited to the job. We just had a thread late last week about the trials of adjusting using C3d.
SurvNet would probably be better than C3d, if you have that and are familiar with it.
Remember that your GPS points are not perfect, and that a few hundreths in a 300′ backsight to begin would extrapolate out to several tenths of error in 2 miles of traverse.
We don’t know your knowledge level of working in state plane, on the grid. If raw ground distances are matching the GPS distances – then maybe you don’t have grid coordinates for the GPS points? Maybe you have scaled to ground coordinates. IDK. But could be the problem is just a settings thing in the adjustment.
59 angles is a lot. I’d prefer to see maybe 10 or 12 at most between points that had been GPS’d. Then the GPS vectors and terrestrial measurements simultaneously adjusted, rather than using coordinates. But you have to work with what you have.
Post your fbk files and GPS points. I bet somebody will find the time to run an adjustment for you. Don’t forget to let us know which SP zone you are working in.
-
Are you running a traverse and closing it back to where you started? Running the other side of the road has angles too tight, but if you are able to brace it, it will help alot.
The large error set could either come from errors in your traverse or from the gps point. Have you ensure that grid to ground conversions have been done for your GPS Points?
This type of situation is where least square solutions shine. Unfortunately i have no experience in using civil3d to do adjustment. However if you use startnet, move3 or even columbus, it will help greatly and are easy to use. I regularly use starnet.
-
First thing I would do is to check on the previous survey’s GPS points – start, middle, end. See if your processed results conform with their coordinates. If you have similar numbers then :
1. I would break the traverse lines into may 2 or 3 segments each starting/ending with a GPS point.
2. Adjust each segment and use the adjusted points as starting points for the next line segment and so on until you reach the final GPS point.
3. If your coordinates from 1 don’t agree then better let you client know asap that something is amiss.
-
The first thing I always do is occupy all of the control and determine their positions using my equipment. Then compare my results to what was given to me and go from there.
Thee next thing is to never show distances to the ten-thousandth of a foot. I get nervous saying any distance is accurate to the hundredth of a foot, or, for that matter any bearing is accurate to the second.
Good luck!
Jack Chiles
-
A 300′ backsite to start a 6000′ traverse is always going to be problematic. You mentioned that you wished you ran the traverse in a different direction, well you did just that. If you closed the angles at both ends then you can reverse the direction of the traverses. This should allow you to check a closure from the middle to the start and from the end to the middle.
I would not run a traverse past the two middle points except for informational data. I would hold the given control.
From your data you don’t seem to have any scale value issues, .01′ per 1000′ is 10ppm. You should be able to adjust for that. I’m guessing that your points are in a generally straight line. Having longer legs would be good also, if there is a way to jump points I would do it.
-
Thanks Bill. I expected some error, but I also, probably foolishly, expected to hit a little closer their coordinates. I hadn’t really thought about the 1:xx number, I think I just balked at a delta of 0.45. I’ll consult with the state surveyor using this least squares adjustment and see if they find that satisfactory for the purpose of this project. Thanks again!
-
Thanks @NorthernSurveyor
I’ve not used least squares before, but using Carlson X-port i came up with the error I described above. But what makes for a “properly weighted” least squares adjustment? X-port just asks me what the closing point is, so I put in the last number of my traverse and entered the coordinates from the states static gps survey, and it calculated new coordinates for the traverse.
Is there another utility you would recommend that I can use that would allow me to weigh all 6 gps points equally or to varying degrees?
Thanks
-
Hey Nate. I don’t have experience using scale factors or GPS surveys in general. When I left the old company we had done some static setups, but I was just a field hand babysitting the setup while it cooked.
The state did a static gps survey, with a combined ground to grid scale factor in the area of 0.99998xxxxx for the 11 points that were adjusted. Adjustment type is plane +height, constraint, 95% confidence level. Since I have no experience in this area I’m not entirely sure what to do with this info. I did look at some videos about to grid to ground conversions in c3d, but I would presume that the state is giving me ground coordinates already?
Thanks for the reply!
-
Lots of good posts and good advice so far.
I would think the easiest and most expedient way to fix this is to add more static control at regular intervals between the existing GPS control. You already have good GPS control points with published coordinates on them at both ends and the middle of the project.
Occupy two of their GPS control points with static receivers, and then run fast static sessions on the recently set traverse control, say every half mile or so. Process baselines and then run an adjustment with all observations, static and total station, in the mix holding their GPS points fixed.
(Or alternatively, adjust the GPS network and then just re-run the traverse adjustment holding the GPS values fixed.)
That should tighten up the network significantly, without having to re-run a 59-station traverse.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman -
I’m curious why it took 59 traverse legs to go 13298.8 feet, or an average of 225 feet per leg, along a road where I’d expect sights to be pretty clear. If you are limiting the lengths in order to get elevation at the same time, I would suggest you also do some long sights to help constrain the horizontal. Any least squares program should be able to handle that, and having a smaller number of random errors on the long legs may tighten things up.
. -
I would think he is setting up points at or near block corners, just a guess, there are often limitations inherent with control points along a highway survey. It’s necessary to see down intersecting streets and with trees, traffic, poles, buildings, the points usually end up at street intersections. Still I would try to jump over as many points as I could and lengthen out my main traverse lines then fill in secondary points with smaller traverses.
-
You have no experience using state plane (grid factors, elevation factors, 1:xxxx precision, least squares adjustment) and are working a state (tax payer) funded project. Did I read that correctly?
When did you pass your PLS? Is there a PLS in your office?
-
In the OP he stated that he was working under a PLS who is not able to help him with this.
-
I don’t fault the OP for trying to figure this out, but I do share Larry’s concern. I am curious as to how the firm was able to win the project when the PLS in responsible charge appears to be practicing outside of his/her area of competence.
Control network design, observation and adjustment as well as least squares analysis QC/QA is something that every license should be able to do in their sleep. It’s not a specialized area of surveying. One cannot competently perform ALTA/NSPS surveys without a solid understanding of least squares.
This is especially true for state projects. When I worked in AK and did DOT work we had to submit proposals detailing how we would establish control. Which published points we planned on holding, how many intermediate points we would set, as well as procedures for observing and adjusting. Part of the deliverable was a reduction narrative discussing which control we held and why, any discrepancies found and what the remedy was. As well as all the field books (signed and stamped by the crew chief, always a PLS), raw data, processing reports, and adjustment reports…if you threw out a published control point or had a blunder, you had better have an explanation as to why.
All of these things should be SOP for larger projects like this. Which, again, makes me wonder how firms can win projects like this with no experience in what is arguably the most critical piece of the job.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman
Log in to reply.