BCRA Grade 5

[Thrun Robert IHMD]

I can't proofread my freshly typed text.  I meant to say that
Larry Fish did NOT try to evaluate accuracy or compare the
loop error distribution to any assumed measurement accuracy.


He was only looking for the occurance of blunders.  Depending
on his definition of "standard deviation" for evaluating loops,
he could make a very nice comparison.

Sorry for my mistake.

I'll make a further explanation.  Every program that closes loops
or adjusts traverses gives some indication of the amount of
adjustment needed.  With random errors, one loop may have a small
error, a similar loop might have a large error...  To get some
indication of the overall accuracy or consistency of a survey,
you have to look at the distribution of many closure adjustments.
COMPASS does make some sort of comparison with assumed accuracy,
but Larry did not look at the distribution of the loop errors.

   Bob Thrun


>  -----Original Message-----
>  From: Thrun Robert IHMD 
>  Sent: Monday, January 20, 2003 9:36 PM
>  To: 'Wookey'; Thrun Robert IHMD
>  Cc: 'mellis@tesco.net'; 'cave-surveying@survex.com'; 'lfish@nyx.net'
>  Subject: RE: BCRA Grade 5
>  
>  
>  On Thu 16 Jan Wookey said:
>  
>  >  > My talk in 2000 was the first and only effort to see if
>  >  > surveys actually met the Grade 5 requirements.
>  >  
>  >  I don't think that's true bob. I'm sure there have been 
>  >  efforts to quantify
>  >  the accuracy of surveys. For example I have analysed a large 
>  >  number of CUCC
>  >  calibration readings to try and understand something about 
>  >  the errors in a
>  >  survey. (these are not a real survey 
>  
>  As you say, these calibration shots are not a real survey.
>  They give an indication of what can be accomplished under
>  ideal conditions, but not what was achieved in an actual cave
>  survey.
>  
>  >  And Larry Fish has run statistical analysis on large surveys 
>  >  to see how accurate they really are.
>  
>  Larry was looking for large errors that indicate blunders.  He 
>  did try to evaluate accuracy or compare the loop error distribution
>  to any assumed measurement accuracy.  He has "standard deviation"
>  as one of the choices of criteria for evaluating loop quality.
>  I was not able to find any description of exactly how he defines
>  standard deviation.  If he uses the three-dimensional error 
>  ellipsoid, it would be a very good way to compare the actual
>  closures with an assumed standard.  It would be better than the
>  paper I gave.  I would love to see such a study.
>  
>  >  People are slightly more discerning than that. They call a 
>  >  CC&T survey that
>  >  they took reasonable care over a grade 5 survey. One they 
>  >  took little care
>  >  over (or used inferior equipment) they call grade 3, or 
>  >  maybe 4 if they
>  >  felt it was somewhere between.
>  
>  What instruments would you use if you set out to make a Grade 3
>  survey?  It is obslete except for underwater.
>  
>  For all practical purposes, Grades 5 and 3 are the same.  Both
>  are much better than needed for route finding.  Both are good
>  enough to tell you what valley the cave is headed for or what
>  area to search for entrances or connections.  Neither is good
>  enough for drilling a new entrance at the far end of the cave.
>  The two blend into each other with no clear distinguishing
>  criteria.  I have seen reports that the Grade 2 survey of Ogof 
>  Draenen agrees well with the Grade 5 survey.  The difference is
>  the lack of vertical information in Grade 2.
>  
>  >  
>  >  > The revision brings the BCRA Grading
>  >  > System in line with actual practice and changes the 
>  >  > System from a failed attempt to quantify the accuracy
>  >  > to jargon for the sake of jargon.
>  >  
>  >  I was generally with you till this last sentence. I do in 
>  >  fact agree with
>  >  you that the BCRA grading scheme is not actually very 
>  useful for it's
>  >  intended purpose and a simply serves as a shorthand 
>  >  description of the
>  >  method and instruments used. An actual description of the 
>  >  instruments and
>  >  method used would probably be rather more useful in 
>  >  determining expected
>  >  errors.
>  >  
>  >  However this revision does bring things more in line with 
>  >  practice so I do
>  >  think that's an improvement, and not just 'jargon for the 
>  >  sake of jargon'.
>  
>  I think that Wookey is too accustomed to the BCRA Grading
>  scheme.  Try telling someone who is not a British caver that
>  a survey is Grade 5.
>  
>     Bob Thrun
>