I'm preparing to adjust my control network in Star*net, and am pondering the mechanism Star*net uses when adjusting 3D networks.
Let's say there's a measure up problem at one point in a simple triangular traverse. Star*net might have the option of adjusting the elevation to make it work, or push the point further out to make it work.
How does Star*net "weight" elevation adjustments vs. horizontal adjustments if a solution can be found either way?
If one is less confident of elevation data than the horizontal data (angles and distances), is there any advantage to just doing a 2D adjustment (either first or only)?
If one is less confident of elevation data than the horizontal data (angles and distances), is there any advantage to just doing a 2D adjustment (either first or only)?
We run into this all the time, where we only want an XY adjustment since we either intend to level through the points because trig elevations would be unsatisfactory, or we simply don't need the elevations adjusted. We do, however, include measure-ups and zenith angles in our data collection process "just in case", and we put the .3D tag in Star*Net so it doesn't throw the 'Unrecognized Data' error. We found it was much more hassle for us trying to do a strict 2D data collection via Carlson, so we stuck with the 3D to 2D thing. Is there an advantage? Maybe, if only for headache reduction on our behalf.
rfc, post: 370646, member: 8882 wrote: How does Star*net "weight" elevation adjustments vs. horizontal adjustments if a solution can be found either way?
Then answer, of course, is that Star*Net weights observations using the standard errors of those observations that the user thinks to be a realistic representation of their uncertainties. Naturally, on typical lines of a conventional survey, even errors in zenith angles ten times the standard error of the observation introduce relatively small errors in the horizontal component of the distance.
The best long-term solution is to use proper techniques with such consistency that either no blunders are present in the work or are very rare. That is accomplished through choice of equipment and methods.
For example, if HI measure-ups are a problem. Measure them at the beginning of the setup and at the end, recording the measurement in a paper field book to supplement the data collector's record.
If target heights are a problem, use the same prisms on the same poles and adopt procedures to check them. Also record the HT values in a paper field book to supplement the DC record.
If there are problems related to the actual measurement of zenith angles, then adopt procedures and follow them without deviation until the problem disappears.
Surveying field work is a combination of rote and novel elements. Start with the rote.
Kent McMillan, post: 370654, member: 3 wrote: For example, if HI measure-ups are a problem. Measure them at the beginning of the setup and at the end, recording the measurement in a paper field book to supplement the data collector's record.
If target heights are a problem, use the same prisms on the same poles and adopt procedures to check them. Also record the HT values in a paper field book to supplement the DC record.
If there are problems related to the actual measurement of zenith angles, then adopt procedures and follow them without deviation until the problem disappears.
Surveying field work is a combination of rote and novel elements. Start with the rote.
All excellent advice. I've known for some time that measure-up procedures are my Achilles heel. For now, I'll do the adjustment in 3D, but loosen up the vertical uncertainties to what I believe them to realistically be, so the "Vertical tail doesn't wag the Horizontal dog".
Then, since leveling has long been on my list of things to learn, I thought I'd start at the highest point of the network, and (using an extendable 5 meter level rod I have), use the TS as a transit (adjusting V datum to 0, and doing D and R of course), and do a level run(s) downwards to the lowest point. When the time comes, I can probably figure out a way to introduce that data back into Star*net, for a "elevation only" adjustment.
