Would anyone who's used/is familiar with both, care to comment on how TraversePC's Least Squares "module", stands up to what appears to be the gold standard?
The Traverse LSQ is very easy to use and to the point . It works for Boundary Surveys very well . Does not allow GPS vectors . Everything is entered as a control point . This is the program I would use to teach college students surveying . It not really setup for Engineering Topo or construction staking . Big jobs are a pain in the arse.
Starnet what can you say except its the gold standard but thats all it does.
> The Traverse LSQ is very easy to use and to the point . It works for Boundary Surveys very well . Does not allow GPS vectors . Everything is entered as a control point . This is the program I would use to teach college students surveying . It not really setup for Engineering Topo or construction staking . Big jobs are a pain in the arse.
>
> Starnet what can you say except its the gold standard but thats all it does.
Thank you. Very helpful. In my case, I'm not doing Engineering Topo or Construction staking, and I don't think my jobs are big by any standard.
Is it a reasonable to assume that if the same assumptions about standard errors are put into each program, the results should be identical?
Maybe the same, maybe not. We had a couple threads a while back that explored different assumptions they might make regarding the error estimates.
Star*Net takes your standard errors and if those turn out to be acceptable by the Chi-Squared test, then it reports the results based on that assumption.
Someone else posted about a commercial program that took the approach, outlined in Wolf & Ghilani, where they assume they are estimating the scale of the error of fit from your measured data, not from your standard errors, and only use your standard error values to proportion their estimate of the error among the sources. This results in more pessimistic statistics, especially so if the redundancy is low.
[msg=281864]Old thread[/msg] discussing an encounter with this difference in SurvNet.
Early [msg=75990]flailing around[/msg] with the topic.
The nugget in those old threads is Dennis Milbert's post about 80% of the way down [msg=242017]this thread[/msg]
> Is it a reasonable to assume that if the same assumptions about standard errors are put into each program, the results should be identical?
Very close, but not identical. Least squares is an iterative calculation. That and there is no one "right way" to do it from beginning to end. Different programmers will follow different paths to get to the end. The final answers from different packages should be expected to end up with very slightly different answers. Those differences should be miniscule, but they will be present.
> Maybe the same, maybe not. We had a couple threads a while back that explored different assumptions they might make regarding the error estimates.
Thanks! Those threads are something. Don't know why I didn't come across them in my search on the subject (of TraversePC vs. Starnet). Definitely some good reading.
I think the gist of what I'm gathering (in this thread) is that given that my principle purpose of acquiring LSA software is education, it may not matter which I choose (given the other factors involved, like budget). I can learn either way.
> > Maybe the same, maybe not. We had a couple threads a while back that explored different assumptions they might make regarding the error estimates.
>
> Thanks! Those threads are something. Don't know why I didn't come across them in my search on the subject (of TraversePC vs. Starnet). Definitely some good reading.
>
> I think the gist of what I'm gathering (in this thread) is that given that my principle purpose of acquiring LSA software is education, it may not matter which I choose (given the other factors involved, like budget). I can learn either way.
The best question to ask is whether TraversePC will accept the same combinations of measurements and observations that Star*Net will or whether it is limited to a basic loop or line traverse adjustment. Part of the power of Star*Net is that it allows the user to add just about anything that can be measured - angles across a network, distances, elevation difference - to an adjustment.
Carlson's SurvNet is a pretty good program. You can purchase it as a standalone program, and you can use conventional as well as GPS information.
That CAD Girl has it available here:
SurvNET
It is a part of my CAD software, Carlson Survey 2008. It is pretty affordable.
Good luck!