On Thu, Aug 15, 2019 at 10:17:00AM +0200, MD wrote:
> I wonder what people use to find errors. For example im I have too
> loops L1 and L2 and L1 is “good” (0.1% error) while L2 is “bad” (4%
> error) i can assume that the error is in the stations/shots which are
> part of L2 but not L1. Is there an automated approach for that?

I would avoid looking at the percentage error - what's reasonable
varies depending on the number of legs in the loop, and also how
long the legs are, so you can't simply compare the percentage errors
for two different loops.

What's more useful is what the observed error is compared to what you'd
expect the error to be based on the grade(s) of the survey(s).  The
grades say things like "compass readings within one degree" and from
that you can work out what you'd expect the error for each leg to be
(in 3 dimensions) and sum those to get the expected error for each traverse.

Benedikt's suggestion to view coloured by error in aven is where I
would start (right click on the colour key is the easiest way to change
what to colour by).  This shows you each traverse coloured by its error
measured in standard deviations - assuming the grades are appropriate
then 99.7% of readings should be within 3 standard deviations:

https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule

The colour scale is fixed such that blue is "good" and anything else
is suspect.  You probably want to look at the worst first.

Once you have a suspect traverse identified, you can look it up in the
.err file which is produced alongside the .3d file when you process the
data.  Here's an example entry:

76.brave.18 - 76.brave.17 - 76.brave.16 - 76.brave.15 - 76.brave.14 - 
76.brave.13 - 76.brave.12 - 76.brave.11 - 76.brave.10
Original length  32.35m (  8 legs), moved   1.26m ( 0.16m/leg). Error   3.90%
4.920289
H: 5.888476 V: 0.864600

Here 4.920289 is the number of sds (which is what aven colours by).  But
below that the H and V are the equivalent numbers looking only at the
horizontal and vertical errors.  So here the vertical error is very good,
and the horizontal error lousy.  That suggests the error is probably either
in a compass reading or the tape reading in a low-inclination leg.  (If
H was fine and V bad, you'd be looking for a clino error or a tape error
in a steep leg or plumb.)

So you can now look at each of the legs in the list on the first line and
look at their compass readings and the tape readings for any low-inclination
legs.  For manually entered data, compare with the original notes.

It's still somewhat manual, but at least you can narrow down where to look
significantly.

Cheers,
    Olly
_______________________________________________
Therion mailing list
Therion@speleo.sk
https://mailman.speleo.sk/listinfo/therion

Reply via email to