What exactly is in the .err file?

[Olly Betts]

On Mon, Jul 28, 2003 at 12:08:54PM -0400, Philip Balister wrote:
> I think I understand these lines, but I could use somone to explain what
> they mean so I can be sure I undertand what is going on.
> 
> It seems like the second entry is not a complete loop, just a portion of one.

In a dataset with more than one loop, you get reports for traverses of
legs between junctions and fixed points (side branches are removed
first, so some junctions don't end a traverse).

These traverses won't generally close back on themselves, and it
wouldn't make sense to forcibly merge statistics for these into closed
loops.  We want to use these statistics to identity bad surveys, so the
smaller each bit is, the better.

> newberries.712 - newberries.713
> Original length   7.59m (  1 shots), moved   0.05m ( 0.05m/shot). Error   0.61%

This is a single leg, 7.59m as measured.  After closure, the ends moved 0.05m
away from their relative positions from this leg, and 0.05 is .61% of
7.59.

> 0.535941

This is the ratio of the movement to what the statistical model predicts.
It shouldn't be too large - < 1 is nothing to worry about, > 3 is rather
suspect.

> H: 0.468307 V: 0.650443

Similar ratios, but for horizontal and vertical components.  Again,
< 1 good, > 3 bad.  If these are different in size, that suggests a
blunder type.  Small V, large H suggests a misread compass for example.
While large V small H could be a reversed sign on a clino.

Also a bad error in H or V can be masked by a good value for the other
so the overall reading looks ok.

There's also a more sophisticated gross error detector in the code, but
it's disabled as it needs access to the unclosed data too to work well.

> coon.20 - coon.19 - coon.18 - coon.17 - coon.16 - coon.15 - coon.13
> Original length  14.66m (  6 shots), moved   0.39m ( 0.06m/shot). Error   2.65%
> 2.067961

This is similar, except you seem to have failed to paste the H and V
values.  The error ratio is a little high - it's probably worth checking
for transcription errors from typing in the data, especially in H and V
differ in size.

Cheers,
    Olly