Activity Feed › Discussion Forums › Strictly Surveying › Control networks – how to know which stations have moved?
-
Control networks – how to know which stations have moved?
Posted by fobos8 on June 24, 2019 at 6:59 amHi guys
I’ve just been surveying for two years now and would really like some advice on a topic. I’ve done a few of construction jobs when I’ve had to set out from an existing local control network that was typically put in by a surveyor 2/3 years previously. First thing I do is traverse it and check if any of the stations “appear out”. So of course you set over a known point, and backsight another known point. So far I’ve had good agreement with my co-ordinates and those of the previous survey but I’m sure down the line I’ll get some jobs where some nails apear to be out by some distance.
What is the best procedure for checking existing networks and deducing which nails may be out or have moved. Of course I have to have two fixed points to start a traverse and when comparing your co-ordinates with the previous co-ordinates you may apear to have good agreement on these but these may be the ones in error.
What does everyone else do when faced with this scenario?
Cheers, Andrew
fobos8 replied 5 years, 3 months ago 8 Members · 13 Replies -
13 Replies
-
The very next question anyone should be asking is why are you traversing?
Its the worst possible method to establish or check control. Only very exceptional cases when nothing else is available should one traverse.
Back onto your question of checking for physical mark movement.
The quickest, easiest method would be to run your traverse (if you cant do a braced network) and run it through Star*net or similar.
In the field your not holding anything fixed or constrained your simply taking distance and angle measurements. End of story. The magic happens in an LS program.
To isolate movement you’ll need to run numerous adjustments each time holding 2 points fixed until your coordinate changes (delta initial issued control versus adjusted control) are as low as possible across the entire network.
-
Redundancy
To check your control, you must use that control and stake to existing known points and locations that were set from that control for comparison to the required accuracies of the project.
It is always good procedure to not take the first point or set of points to be sufficient to continue a project.
Verify what you find
-
In reality, the control provided should be treated with a grain of salt. In doing your verifications you are essentially starting from scratch just with values set to check into. If you find one that does not agree then in your process you should have enough information to disregard or re-establish that point. This goes for horizontal and vertical control. If possible, see if you can find out the method used to set the previous control, that way if thare is error you may have a better idea of what you’ve been handed. Also, knowing how it was set and running through it with another method would be great redundency if it all checked well.
Of course if you find gross errors in the existing network then raise the red flag and double check your work. If you are good, keep going.
-
If you have least squares capability, measure the original control with lots of redundant cross connections. Enter the original measurements if you have them available or the original coordinates if that’s what you have, with reasonable standard errors. Hold one point’s coordinates and one bearing or other coordinate. See what fits and doesnt fit.
If no LS program, and the original control was 101 thru …, run your measurements on those points as 201… etc. And look at the distances 101 to 201, 102 to 202, etc. Try holding different fixed and backsight points to see what works best.
. -
“The very next question anyone should be asking is why are you traversing?
Its the worst possible method to establish or check control. Only very exceptional cases when nothing else is available should one traverse.”
This is a perfectly ridiculous statement.
As with most things, it depends on the situation. A couple things that would influence how I approach it would include:
- How many stations are we talking about? Two or three control points don’t constitute a network
- Do I have GNSS available? Am I in a GNSS friendly environment?
- How was the control done initially? If it was done with GNSS the points may not be intervisible, and if you don’t have GNSS then you’re going to be traversing.
What matters is that you understand your equipment, its precision, and its limitations, and that you use the best possible procedures to get the best possible results. Typically, at relatively short distances, a total station measurement is going to be much more accurate than a GNSS measurement, but only if all error sources are identified and mitigated to the extent possible. If one’s idea of traversing is to take a single face one measurement to one’s backsight and foresight, then yes, the results might be less than wonderful.
On large sites with extensive control, a GNSS site calibration will give you a very good idea of how everything fits together in a short amount of time – but you have to understand how site calibrations work and that there are rules that you have to follow in order to get valid results.
The bottom line is that on every job I look at, I’m evaluating what I’m given, what I need to accomplish, what tools I have at my disposal, and only then formulating a game plan.
-
At the risk of being scorned as an ignorant Luddite, I would also posit that Least Squares is not the be-all and end-all of survey measurement. I’m going to be able to verify control and identify whether one station, multiple stations, or the entire thing is bad by making observations in the field, regardless of the measurement tool.
-
Columbus has a very interesting feature where you can enter the given control and various measurements between them and it will exhaustively test all the combinations and tell you the given control coordinates which, when held fixed, most accurately fit the test measurements.
-
Posted by: Lee D
At the risk of being scorned as an ignorant Luddite, I would also posit that Least Squares is not the be-all and end-all of survey measurement. I’m going to be able to verify control and identify whether one station, multiple stations, or the entire thing is bad by making observations in the field, regardless of the measurement tool.
Please share how you do it
-
Please note that I said determine if points are bad, not re-establish them. If I’m given coordinates on control points then I know what the angles and distances between them should be… with a 1″ total station with a 1mm+1.5ppm EDM it’s not hard to figure out if they work or not. I always shoot in cross ties if I can see them, not just my BS and FS. Of course, what do you consider bad? If a point appears to be off by 0.03′ am I going to re-observe and re-adjust the entire thing over that?
As I said above, given good geometry a site calibration will tell me if points fit their coordinates or not. Again, I’m not splitting millimeters here… a valid site calibration will provide meaningful enough residuals to let you know if the overall network has any problems. If so, then you have to decide what the next step is.
Every job is different; there is no one size fits all answer to these questions.
-
This is a job least squares does so well that I can hardly imagine trying it without. With it I’d set up in the middle of the site, ideally, and shoot all the control I could see. Or maybe RTK the control, if the site conditions allow it. Or traverse about if that is what it took. Then I’d have the LS best fit my observations onto the supplied control – giving the control a reasonable standard error. The LS adjustment report would tell me which points fell within acceptable tolerances and which didn’t.
Even if you don’t have access to an LS package in the office you may have something almost as good in your dc. Resection. Resect a position for a free station using shots to the control. When using several points in a resection you will get a report on the residuals and outliers will be apparent. But only on the dc screen, and once you accept the results the documentation is iffy.
Lacking all that, I’d still do the field part the same and then fiddle with rotation & translation of one set of coordinates (the field ties) to the other (the record coordinates) until I was satisfied that I had the maximum amount of agreement between the two possible.
-
Indeed, there was a time before Least Squares was a common tool available to surveyors. We got on somehow, didn’t we?
-
I do a lot of resections, usually to establish the instrument from GPS but on some sites I’ll shoot in things like radio towers and building corners to use for resections. On a topo and especially in a facility like an electrical substation It saves a lot of time and effort.
Log in to reply.