[Therion] Wrong average calculation from compass/backcompass data

[Olly Betts]

On Thu, May 17, 2018 at 09:25:23AM +0200, Evaristo Quiroga wrote:
> I attach the original file with the fore/back sight (SCAGUA.th) and a
> manually averaged  (SCAGUA_avg.th) to compare if Survex and Therion work
> well. The error loop is similar between the two files (2,9m) but I have to
> check if the station are positioned at the same coordinates.

I feel I should point out that Survex's handling of backsights isn't
coming in to play here at all - Therion flattens away the backsights
so they don't appear in what it asks Survex to process.

Out of interest I did a crude conversion of the two attached files to
Survex format and processed them natively, and with scagua.279.0 fixed
in each case the maximum discrepancy between the two results is:

Moved by (-0.11,0.13,0.03): scagua.a.10

Which seems entirely plausible - I wouldn't expect exact agreement in
this case as some of the data has backsights and some doesn't.  In the
data from SCAGUA.th, Survex will treat the data with backsights as
slightly more accurate (because the average of two essentially
independent readings is expected to be more accurate than a single
reading) and that affects the distribution of loop misclosures.

Cheers,
    Olly