This diagram shows a network that I recently surveyed to establish horizontal and vertical control points at a site for building monitoring purposes. To give a sense of the relatively small scale of the network: the distance from CP 100 to CP 102 is only a little over 1000 ft.
The object of the exercise was to determine the horizontal and vertical coordinates of the control points with uncertainties smaller than 0.0.007 ft. at 68% confidence in relation to a fundamental control point (Control Point No. 100, a deep driven rod) in the vicinity of the site.
In the survey design, the NAD83 coordinates of Point No. 100 were treated as fixed and the other control points were surveyed in relation to that fundamental point in a way that was designed to keep the uncertainties of the points positioned from CP 100 to less than +/- 0.007 ft. (standard error).
If you examine the network connections, you'll see that most of the control points were surveyed from at least two network points and that there is a fair amount of redundancy that can also be used to test the additive constant of the EDMI.
This is done by adding some small increment such as 0.003 ft. to all of the distances measured in the course of the network survey or subtracting a similar quantity. Then the effect of systematically "correcting" the measured distances by, say, +0.003 ft. is assesssed by the analysis of variance of the residuals in a least squares adjustment of the network. The best fitting additive constant is that which also produces the smallest weighted sum of squares of the residuals.
You are correct, Kent, in that (hardly) anyone ever measures long distances with an edm anymore. But, shooting a long distance is still a good way to accurately get the scale part of the calibration. But, it probably doesn't need to be 1500 meters!
Oops-this should have gone under Kent's previous post!
> You are correct, Kent, in that (hardly) anyone ever measures long distances with an edm anymore. But, shooting a long distance is still a good way to accurately get the scale part of the calibration. But, it probably doesn't need to be 1500 meters!
If you're routinely combining GPS vectors with conventional measurements, you should have a means to check the scale of the EDMI ranges also. For example, in that network above, the vectors 100-101 and 100-102 were determined by static GPS solutions, more than two sessions on different dayts for each vector.
Interesting project. I do miss messing with Star*Net, a great tool
Is GPS used to constrain the network or just to index to lat/lon?
If so, how long is the shortest vector?
Does Star*Net allow computation of scale and constant factors for EDM measurements? The only commercial package I am aware of that does that is GeoLab.
Peter Lazio
> Does Star*Net allow computation of scale and constant factors for EDM measurements? The only commercial package I am aware of that does that is GeoLab.
No, for a typical calibration, you'd have to generate several input files with different trial corrections applied to all the distances. Then you just plot the change in the standard error of unit weight of the distance residuals to find the correction that minimizes it.
It would be nice if the process were automated, but considering how infrequently one needs to verify the additive constant, it isn't prohibitively laborious to do it manually.
> Is GPS used to constrain the network or just to index to lat/lon?
> If so, how long is the shortest vector?
Yes, the repeated GPS vectors 100-101 and 100-102 were adjusted with the conventional measurements, although for the purposes of determining the additive constant of the EDM, you'd probably want to adjust just the conventional measurements using just one azimuth from the GPS vectors to orient the network but not fix its scale.
The length of the shorter of the two GPS vectors, 100-101, was approximately 719 ft.
> ..., for a typical calibration, you'd have to generate several input files with different trial corrections applied to all the distances. Then you just plot the change in the standard error of unit weight of the distance residuals to find the correction that minimizes it.
>
How would you differentiate between the scale portion and the constant portion of the EDM calibration? I understand for short lines the the constant portion would be the dominant correction so perhaps you just ignore the scale factor?
Peter Lazio
> How would you differentiate between the scale portion and the constant portion of the EDM calibration? I understand for short lines the the constant portion would be the dominant correction so perhaps you just ignore the scale factor?
Well, the assumption that I'd make would be that the scale error is mostly due to drift in the quartz oscillator that provides the reference frequency for the instrument. Under that assumption, the scale constant of the EDMI would be reasonably expected to be the same for all lines and so wouldn't figure into the determination of the additive constant.
That is, the scale error will most likely be on the order of <20ppm and the additive constant <<1m. So in the worst case, the additive constant might be in error by 0.000020 x 1m, and in most normal cases considerably less.
