by Roppelcaver » Oct 24, 2007 11:31 pm
"Weighting" of loops is quite easy in Walls, and is something we routinely do at Mammoth. We have surveys of varying age and quality, so this is quite necessary.
In Walls, you weight by assigning different variances -- a variance of 0 results in a fixed station, vector, traverse, or survey; and a large variance effectively "floats" the station, vector, traverse, or survey.
Walls has a useful help file to help you do this, so there is no need to repeat it here.
I would suggest that you may not want to "fix" (i.e., not adjust) the better surveys, but would want to weight them more than the other surveys -- I think the default variance might be 2; if so, you might set the "good" survey variance to 0.5. That might give you the desired effect.
Generally, setting variances is not something to do lightly. Walls will "weight" traverses with low error (as part of a loop system) over those with high error (presuming a reasonable network size), and also provides tools for blunder detection.
At Mammoth, older surveys did not have backsights, more recent surveys may not have had a compass course used. Current surveys are fully backsighted and use a compass course; the latter have the most weight applied by design, but the older surveys are still allowed to contribute to the closure of the network -- blunders can certainly still occur in the best surveys.
Again, I suggest you refer to User guide for Walls. I would be happy to answer questions if you find the documentation inadequate. The real expert is David McKenzie, but at Mammoth we use this feature often enough that we have learned how to do it fairly well -- how, when, and when not to use it.
Jim
Roppel Caver guy