Gene is right about large GNSS networks with a small conventional tie off in a corner, it can make the statistics look funny. If the GNSS and conventional are roughly equal sizes then the issue doesn't appear as much.
Jim,
The scale factors that I use are much larger than yours, but I combine OPUS Projects session solutions over multiple years with post-processed local GPS vectors and EDM measurements.?ÿ I'm inclined to prefix "optimistic" with the term ginormously rather than notoriously.
The point I'm trying to make is that the chi-square test should not be the focus of one's analyses of an LSA.?ÿ I'm not saying that others should ignore it when it falls outside the bounds.?ÿ The mere fact that the chi-square test fails does not mean that the adjustment is poor or wrong.?ÿ I have always been a solo surveyor, so I have a fairly good idea of how to tweak things to get the chi-square test to pass (if I feel the need).?ÿ For my work in the mountains, correcting for the atmosphere and deflection of the vertical are more important than tweaking error estimates to get a "good fit" in Star*Net, Columbus or Move3.
A general question to Jim and Norman: If you don't scale the GPS, do the residuals significantly change?
No, but I find it helpful to know -- and be constantly reminded -- that the accuracy statistics produced by my GPS gear are consistently unrealistic.?ÿ I wouldn't want to start believing them.
I don't have quite the hat or mustache for this, but here goes..
One advantage of Starnet is that you can do proper geodetic computations. The project does not use one scale factor but rather computes the average csf of measured distances for terrestrial observations. All measurements are mathematically abused in a consistent fashion after reduction. This allows for a robust adjustment of mixed data sets with realistic statistical tests...
A general question to Jim and Norman:?ÿ If you don't scale the GPS, do the residuals significantly change?
Norman has gone home for the day so I'll answer for him....
Increasing the error factor on the GPS vectors reduces their weighting in the adjustment. So it does affect the residuals, at least in theory. But increasing by a factor of 1.5 is hardly noticable. It does change the standarized residuals, and therefore the chi-square. The residuals themselves change slightly with that kind of error factor but not significantly.?ÿ
TBC has a handy scalar button you can push after the first run...unit weight will be 1.00 or 0.99 after that 😉
The OP posted traverse data in the Starnet traverse format. It's not GNSS data.
