I'm trying to run a least squares adjustment on a job and I can't get the data to pass the Chi-Square test.
It's a new one second instrument and the network has 18 degrees of freedom. The data converges in 2 iterations.
I am curious as to what kind of preprocessing and standard errors others have set up. I feel like it's something unreasonable in those tabs that's causing it to fail.
Thanks for any input.
Sent from my iPhone using Tapatalk
Did you put in centering errors? That will make a good network fail chi square if they are zero.
Are you failing the Chi-Square above or below the stated range?
If it is below then your data is better than your default standard errors, if it is above your default standard errors are to tight for your data.
What is your Standard Error Unit Weight?
If your default standard errors are correct for your data the Standard Error Unit Weight should be approximately 1.0
If you want to email me the rw5 file and your default Standard Errors I can take a look at it. My email is in my profile
Just because you have a new one second instrument that doesn't mean that your standard error for angles is going to be 1 second. If you are using standard issue eyeballs, tribrachs, tripods, and glass then 3" or more may be reasonable.
Dan Dunn, post: 395211, member: 911 wrote: Are you failing the Chi-Square above or below the stated range?
If it is below then your data is better than your default standard errors, if it is above your default standard errors are to tight for your data.What is your Standard Error Unit Weight?
If your default standard errors are correct for your data the Standard Error Unit Weight should be approximately 1.0If you want to email me the rw5 file and your default Standard Errors I can take a look at it. My email is in my profile
Dan,
I don't see your email in your profile. Thank you for your willingness to help.
Mark,
That is exactly what I was thinking to, so I set it to 2.5" and It improved it. I do use Tri-Max tripods and the highest specced Leica 360 prism (on a rod though). Thanks.
Does SurvNet break out the adjustment error values by measurement type, i.e. can you tell whether or not one type (e.g., horizontal angles, zenith angles, distances, GPS vectors) is inflating the the SEUW)?
Jason Graves, post: 395217, member: 9531 wrote: Dan,
I don't see your email in your profile.
Click on Start a Conversation in Dan's profile. The board will forward your message to Dan.
I like Jim Frame's comment. Look to see if you can identify the error source. If the error is very large then look for a mislabeled point. Be sure you have reasonable errors for both target and instrument centering and for pointing. I think I would start with 1.41" rather than 2.5"