Survey Problem?

[Olly Betts]

[I'm replying to the list as this is likely of general interest]

In message <001a01c08fcb$3b4cdd00$f4199fd4@oemcomputer>, "Raymond Duffy" writes
:
>I've included a survey file which if you process it you should find that in
>the error file station MC033 has two loops which go from it, however, there
>is only one loop on the survey but the line crosses over itself in plan
>view.  How can it make two error calculation on the same data????????

Simplifying a little, the situation is essentially this:

  X <-- Fixed point
  |
  A-B
 /  |
E-D-C
  |
  F-G
  |/
  H

And Ray is asking why errors are reported for A-E-D and A-B-C-D separately
rather than for the loop A-B-C-D-E-A (splitting at A is arbitrary - the
error statistics would be the same for B-C-D-E-A-B, etc).

Note is that Survex reports errors for traverses, not for loops.  In this
survey, the traverses are:

X-A (*)
A-B-C-D
A-E-D 
D-F (*)
F-G-H

Because there's only one fixed point involved, the traverses marked (*)
which are across "articulation points" (i.e. those traverses which would
cause the survey to fall into two pieces if removed) will have no closure
error (but they are still reported in the .err file).

So A-B-C-D-E-A loop is split because of the traverses attached at A and D.
Note that these traverses don't affect how the error is distributed around
the loop, so it is arguably better if such loops aren't split in the error
stats, but that's how it works.  For the moment at least - it's just occured
to me how I might be able to fix this fairly easily.

Cheers,
Olly