Say you're working on a corridor construction (just over a mile) project with old control that you have no idea where it came from or how it was produced, just NEZ, and the statement that it's a "local coordinate system". For all I know, the project I'm looking at could have been traversed and adjusted with compass rule, it could have been set with GPS, I have no idea.
Those who came before me (in trimble speak) calibrated to 4 points in the highway corridor. They had more ties, but whittled it down until they had small residuals. There is control beyond the current construction area, but the dude in charge said you only want to calibrate in the construction area or else "it gets noisy".
So of course now we're laying out and the project has a significant scale factor. Our total station measurements between control points indicate the control should be S.F. 1.00. I generally don't have a problem with scale factors, but not when the existing control indicates there shouldn't be one.
It's my position that low residuals don't mean good control, and that site calibrations provide a false peace of mind. What would have been the alternatives? Would greater distance in the control used to establish the projection generally be superior just using points in the immediate area? Would a projection or "calibration" that yielded a S.F. close to 1 be superior to one with low residuals?
"Its a road, not a watch....." is what my old boss used to say. I would try doing a 2 point calibration and doing checks to the other control. don't be so concerned about scale factor, it is what it is. Its only a mile long project, and since you said total station measurments are coming out pretty close, I would consider using my robot to do the staking. Vertical is gonna be more critical than horizontal, so definitely you are gonna want to use the total station for the paving hubs.
Doug
I have seen Trimble calibrations show small residuals and in fact the site be "tilted" by .3' or more when checked with conventional instruments or differential leveling. A "Tilted" site occurs when you have error in one of the control points calibrated to and the software creates your tilted plane, or as Trimble calls it your own site specific geoid model. Thats fine in theory, but you had best really scrutinize any provided control because the residuals don't always throw up the red flag for you.
Be VERY cautious about that inclined plane, especially along an alignment and not a "box" surrounding the site.
If you're ok with the distances checking, use the two point calibration for horizontal (likely one at each end of the project) and one point calibration for your vertical (something in the middle). Use all that other data as checks.
Better yet, run your own control. You'll sleep better.
It does sound like whoever ran the control used a total station and your basic TenThousandLand coordinates system, hence the 1.0 scale factor. All the distances are ground and it's very simple. Then someone else started using their RTK rig and "calibrated" to it. So you've got something that was good to hundredths being warped to something that might have been good to tenths, if the stars were aligned.
In a way, TenThousandLand is like the world's tiniest Low Distortion Projection (LDP).
It's exactly right for the tiny area around its origin and distorted further away (as the inventors of the PLSS discovered!)
I agree with the other posters: if you were given this control system in a gift wrapped box, be suspicious of the elevations til you check them. 4 points for elevation can tilt the whole model (as can 2 or 3 points, best hold 1 or use 5 or more).
Low residuals don't mean anything without redundancy. By getting rid of ties they got rid of the redundancy that shows what is actually going on. The 4 point model fits itself nicely, hence the low residuals, but does that overall model fit the physical world? is it gonna drain right?
You ask what the alternatives are or might have been. I'm not a construction or roads guy so other posters will probably have better answers. I would shoot my own control in state plane to keep the GPS and the total station on the same page and avoid "calibrations", and fluff it up to ground if someone else needs that later.
Your last question: Would a projection or "calibration" that yielded a S.F. close to 1 be superior to one with low residuals?
Apples and oranges, seems like. If you are getting a scale factor of 1.0 to match the previous ground-based total station work, that makes sense. If you are getting noticeable residuals doing this, the uncertainty is probably coming from the RTK shots, not the control you are matching. It might even be that those 4 good points were the ones the previous RTK krew shot with good DOP and the others were not so great, & so didn't fit the control as well.
>
I would shoot my own control in state plane to keep the GPS and the total station on the same page and avoid "calibrations"
Amen, That will tell you which points work good together. THEN you could calibrate, using those good points, if you still need to.
If this is a public works project or a state highway project, the agency should have the field notes and be able to tell you exactly what was done. Have you made a phone call to the agency yet?
JRL
Low residuals are good....and the calculated scale factor should be very close to 1.000000000
You can also set the scale factor to be 1.0 in the Trimble software.
I agree on the vertical...on a corrider (and in most other cases) it is better to use one point vertically to calibrate to and it and the geoid model to check other benchmarks.