Author Topic: Accuracy Questions  (Read 5377 times)

Offline TheBitterEnd

  • forum hero
  • *****
  • Posts: 1455
  • KCC
    • KCC - Join an active club
Accuracy Questions
« on: October 12, 2012, 08:10:26 pm »
I've read the BCRA survey grading system but am left with a few questions:

How does one know if station position is within a given distance of it's actual position (e.g. 10cm for grade 5)?

For grade 5 the measurement should be taken from within 10cm of the station - is this irrespective of leg length?

Obviously it's not always possible to do a closed loop but what would be an acceptable loop closure error?

'Never argue with a fool, onlookers may not be able to tell the difference.' — Mark Twain

Offline cavermark

  • forum hero
  • *****
  • Posts: 1382
Re: Accuracy Questions
« Reply #1 on: October 12, 2012, 08:23:24 pm »
If your station is a fixed point (e.g. dot of tippex on the wall, tip of a stal etc.) then your tape or disto measure to these points +/-10cm. compass needs to be vertically above or below it within +/- 10cm and clino horizontally level +/-10cm.

Offline graham

  • Retired
  • forum hero
  • *****
  • Posts: 10943
  • UBSS, Speleo-Club de Perigueux, GSG, SUI
    • UBSS
Re: Accuracy Questions
« Reply #2 on: October 12, 2012, 09:53:30 pm »
I've read the BCRA survey grading system but am left with a few questions:

How does one know if station position is within a given distance of it's actual position (e.g. 10cm for grade 5)?

By ensuring that the datum point of your instrument (centre bearing of your compass eg) is within 10 cm of the marked station.

For grade 5 the measurement should be taken from within 10cm of the station - is this irrespective of leg length?
Yes.

Obviously it's not always possible to do a closed loop but what would be an acceptable loop closure error?

I cannot think of a simple answer to this one, but we generally get less than 1% using DistoX.
Caving is for Life not just for Christmas

Offline Les W

  • Hard cavin'
  • forum hero
  • *****
  • Posts: 5269
  • Wessex Cave Club, UCET
    • Wessex Cave Club
Re: Accuracy Questions
« Reply #3 on: October 12, 2012, 10:25:59 pm »
Obviously it's not always possible to do a closed loop but what would be an acceptable loop closure error?

I think this is up to you. A large misclosure will of course be due to errors, and it's acceptability will bepend on what the survey is intended for. Clearly a large misclosure should be investigated if a very high level of accuracy is required.
Often you will know where the suspect readings are likely to be (perhaps in that really nasty bit where you couldnt get both eyes out of the water...  :o ). It is normally ok for most survey uses (i.e. for navigation, to look pretty on the wall, to be used to show locations in a book or for samples or similar) to be reasonably inaccurate as the information you are trying to show is not changed (significantly) by those inaccuracies (also remember the limitations of drawing in a small scale...) however if the reason for the survey is to show a potential connection, relationship of passages or other nearby caves, etc. Then you need to have quite high confidence in the survey.

As Graham said, 1% would normally be considered quite good by most surveyors, on expeditions 3% or greater will be fine in most cases. It is worth considering how anal you can get about these things when in most cases it actually doesn't matter that much...  :-\

It is always worth trying to be as accurate as you sensibly can given the various time and other contraints.
I'm a very busy person

Offline graham

  • Retired
  • forum hero
  • *****
  • Posts: 10943
  • UBSS, Speleo-Club de Perigueux, GSG, SUI
    • UBSS
Re: Accuracy Questions
« Reply #4 on: October 13, 2012, 08:31:29 am »
Obviously it's not always possible to do a closed loop but what would be an acceptable loop closure error?

I think this is up to you. A large misclosure will of course be due to errors, and it's acceptability will bepend on what the survey is intended for. Clearly a large misclosure should be investigated if a very high level of accuracy is required.
Often you will know where the suspect readings are likely to be (perhaps in that really nasty bit where you couldnt get both eyes out of the water...  :o ). It is normally ok for most survey uses (i.e. for navigation, to look pretty on the wall, to be used to show locations in a book or for samples or similar) to be reasonably inaccurate as the information you are trying to show is not changed (significantly) by those inaccuracies (also remember the limitations of drawing in a small scale...) however if the reason for the survey is to show a potential connection, relationship of passages or other nearby caves, etc. Then you need to have quite high confidence in the survey.

As Graham said, 1% would normally be considered quite good by most surveyors, on expeditions 3% or greater will be fine in most cases. It is worth considering how anal you can get about these things when in most cases it actually doesn't matter that much...  :-\

It is always worth trying to be as accurate as you sensibly can given the various time and other contraints.

Someone (Pete Cousins?) did a study which showed that, counter intuitively, accuracy (as defined by loop misclosure errors - I have to be careful not to get shouted at by the mathematicians) does not decrease when the caving gets harder.

The problem with accuracy, as such, is that there seems to be no objective way of measuring/defining accuracy, so it can only be guessed at, at the end of the day.
Caving is for Life not just for Christmas

Offline Fulk

  • forum hero
  • *****
  • Posts: 3832
Re: Accuracy Questions
« Reply #5 on: October 13, 2012, 09:53:54 am »
Quote
Someone (Pete Cousins?) did a study which showed that, counter intuitively, accuracy (as defined by loop misclosure errors - I have to be careful not to get shouted at by the mathematicians) does not decrease when the caving gets harder.

I'm not a mathematician, and I'm  not going to shout at you – but I wonder; could it be that random errors in measurement tend to cancel each other out, so even if your survey is not very accurate, a closed loop could end up at more-or-less the right place, to give it an air of spurious accuracy (no matter how hard the going is)?

Offline graham

  • Retired
  • forum hero
  • *****
  • Posts: 10943
  • UBSS, Speleo-Club de Perigueux, GSG, SUI
    • UBSS
Re: Accuracy Questions
« Reply #6 on: October 13, 2012, 10:15:22 am »
Quote
Someone (Pete Cousins?) did a study which showed that, counter intuitively, accuracy (as defined by loop misclosure errors - I have to be careful not to get shouted at by the mathematicians) does not decrease when the caving gets harder.

I'm not a mathematician, and I'm  not going to shout at you – but I wonder; could it be that random errors in measurement tend to cancel each other out, so even if your survey is not very accurate, a closed loop could end up at more-or-less the right place, to give it an air of spurious accuracy (no matter how hard the going is)?

Yes, absolutely, they will. However, if errors are cancelling themselves out then surely there are less errors left? ;) As I said there is no wholly objective way of checking accuracy of a survey. All one can really do is compare independent surveys and assume (big assumption, lots of reasons to discount it) that correspondence between surveys implies correspondence with reality - especially if they were done using wholly different techniques. As nobody regularly carries out re-surveys, closed loops, which can be thought of as independent surveys between end points are as good as we've got.
Caving is for Life not just for Christmas

Offline TheBitterEnd

  • forum hero
  • *****
  • Posts: 1455
  • KCC
    • KCC - Join an active club
Re: Accuracy Questions
« Reply #7 on: October 13, 2012, 06:52:05 pm »
Thanks for the replies. I guess I was misunderstanding it, I took  "station positions identified to less than 10cm" to mean that the accuracy of the station position should be less than 10cm.
'Never argue with a fool, onlookers may not be able to tell the difference.' — Mark Twain

Offline cavermark

  • forum hero
  • *****
  • Posts: 1382
Re: Accuracy Questions
« Reply #8 on: October 13, 2012, 11:09:24 pm »
Thanks for the replies. I guess I was misunderstanding it, I took  "station positions identified to less than 10cm" to mean that the accuracy of the station position should be less than 10cm.

I think it does imply station accuracy too - e.g. if you have to re-use the station to begin a series for a side passage, you need to know it is the same point.
Hence the value of choosing and marking fixed points - the tippex cross etc.

Offline robjones

  • forum star
  • ****
  • Posts: 656
Re: Accuracy Questions
« Reply #9 on: October 13, 2012, 11:39:06 pm »
Quote
Someone (Pete Cousins?) did a study which showed that, counter intuitively, accuracy (as defined by loop misclosure errors - I have to be careful not to get shouted at by the mathematicians) does not decrease when the caving gets harder.

I'm not a mathematician, and I'm  not going to shout at you – but I wonder; could it be that random errors in measurement tend to cancel each other out, so even if your survey is not very accurate, a closed loop could end up at more-or-less the right place, to give it an air of spurious accuracy (no matter how hard the going is)?

Yes, absolutely, they will. However, if errors are cancelling themselves out then surely there are less errors left? ;) As I said there is no wholly objective way of checking accuracy of a survey. All one can really do is compare independent surveys and assume (big assumption, lots of reasons to discount it) that correspondence between surveys implies correspondence with reality - especially if they were done using wholly different techniques. As nobody regularly carries out re-surveys, closed loops, which can be thought of as independent surveys between end points are as good as we've got.

The lecturer that taught me surveying warned us that perfect closure of a loop indicatced errors rather than magical accuracy. By extension, if closure is significantly better than can be anticipated from the usual quality of results from the equipment being used, then errors may have occurred.

Offline mulucaver

  • stalker
  • ***
  • Posts: 251
  • PDMHS, BCA, CSG
    • daveclucas
Re: Accuracy Questions
« Reply #10 on: October 15, 2012, 07:18:09 am »
Don't forget that each survey leg has two stations so an accuracy of 10cm will give you between 0 and 20cms of error.
Dave Clucas
Exploring the World
One Cave at a Time

Offline wookey

  • player
  • **
  • Posts: 105
Re: Accuracy Questions
« Reply #11 on: October 15, 2012, 01:46:40 pm »
Having looked at a lot of largish datasets, I find that for loops longer than about 10 legs you can expect a loop closure error of no more than 2% with conventional instruments if there are no cock-ups. The errors rise quickly with fewer legs and very short loops (3 or 4 legs) often have quite large closure errors (maybe 8%). That's all from memory -  I guess I should actually produce some numbers, but essentially the minor station position, instrument poisition, aiming accuracy and measurement errors seems to cancel out nicely to give a realistic idea of the overall accuracy so long as there are enough of them, but become much more significant (measured in percentage length terms, as we usually do) when the length isn't very long.

So the short answer is 'aim to get less than 2%, (maybe 1% with digital instruments/data recording)'. If you are doing that then everything is good. But expect to fail in this for short loops. Anything over 15% error is definately a cock-up. Anything over 6% on longer loops almost certainly has mistakes in it. 

Offline jarvist

  • obsessive maniac
  • ***
  • Posts: 424
  • Imperial College CC / Northern Pennine Club
    • Imperial College Caving Club
Re: Accuracy Questions
« Reply #12 on: October 15, 2012, 09:30:06 pm »
With the IC Migovec dataset, we've found very similar behaviour to Wookey above.

Ignoring very small loops, the errors of 'well behaved' (i.e. don't suspect major blunders) surveys seem to be a fairly consistent ~0.1m per leg.

Assuming that human based, transcription or station mislocation / reversed leg errors (blunders) occur at a given rate per survey shot (thus too few to be caught by the Central Limit Theorem, or too systematic in error), you would see the error asymptotically approaching a constant. If all the errors were Gaussian distributed around the true figure, then the error should decrease asymptotically to zero with something like one-over the square-root of the survey shots.

For general (mathematical treatment) of errors, these notes are a very good start:
http://csg.bcra.org.uk/surveynotes.html

For interest, here's the 'Errors' bit from the Sysmig/Vrtnarija IC Migovec data, which is all Suunto instrumented + recorded on paper, over 15+ years, by the relatively inexperienced:
https://gist.github.com/3895033

Offline andrew

  • addict
  • **
  • Posts: 122
Re: Accuracy Questions
« Reply #13 on: October 16, 2012, 09:05:34 am »
Here are the numbers from the Cheddar Catchment. All the data is available at http://cave-registry.org.uk/. I have removed the long list of stations for clarity, but the 4th number along tells you how many station in the loop


Most of this data is with a distoX, although there is some old data and reconstructed data within the set.
Gough's, has metal and electronics in it, so I have let the loops go over 1%. I will generally look closely at any loop with more than 0.5% error. I also look at loops with more than 0.5m error. This has revealed some blunders, even with PocketTopo, where I had not reversed a vertical leg. However, this is probably aiming at better than BCRA grade 5, but I think that the UIS system is probably the one to use when grading surveys,

Code: [Select]
REL-ERR ABS-ERR TOTAL-L STS X-ERROR Y-ERROR Z-ERROR STATIONS
100.00% 6.8m 6.8m 2 -6.6m -1.7m -0.6m [1.42@BH_Surface.BoneHole.CheddarCatchment
87.53% 13.5m 15.4m 2 3.5m 11.3m 6.5m [5@bishops_palace.sumps.Goughs.CheddarCatchment
14.71% 7.1m 48.4m 11 -6.5m 1.2m 2.6m [01@BoxCave.CheddarCatchment
6.15% 1.8m 28.6m 2 0.0m 0.0m -1.8m [125@PatonPlace.TyningsBarrow.CheddarCatchment
6.07% 1.8m 29.3m 6 0.0m 0.0m 1.8m [122@PatonPlace.TyningsBarrow.CheddarCatchment
5.52% 1.6m 29.5m 7 1.1m 1.2m -0.1m [26@BeyondZebraAven.EastPassage.UpperFlood.CheddarCatchment
5.05% 1.3m 25.9m 3 0.2m 1.3m -0.2m [2a@TrenchPassage.UpperFlood.CheddarCatchment
4.11% 6.2m 149.7m 25 3.1m 4.3m 3.1m [START@BeyondZebraAven.EastPassage.UpperFlood.CheddarCatchment
2.53% 0.4m 16.1m 5 -0.2m -0.3m -0.2m [36@BeyondZebraAven.EastPassage.UpperFlood.CheddarCatchment
2.44% 0.4m 18.0m 5 0.4m 0.2m -0.1m [38@FullLRUD.EastPassage.UpperFlood.CheddarCatchment
2.00% 0.5m 25.2m 3 0.2m 0.4m 0.3m [1.21@Entrance.Goughs.CheddarCatchment
1.93% 8.1m 422.6m 72 -8.0m -0.7m -1.2m [5.2@CableCrawl.Goughs.CheddarCatchment
1.88% 0.3m 15.5m 4 0.3m 0.1m -0.0m [11@NeverlandRopeClimb.UpperFlood.CheddarCatchment
1.82% 0.7m 36.4m 11 -0.5m 0.4m 0.1m [2.9@SandChamber.Goughs.CheddarCatchment
1.57% 0.6m 38.2m 10 0.5m 0.1m 0.4m [1@NeverlandRopeClimb.UpperFlood.CheddarCatchment
1.49% 1.1m 71.0m 6 0.1m 1.0m 0.4m [1.11@HeartbreakHill.Goughs.CheddarCatchment
1.41% 0.7m 51.2m 10 -0.5m 0.5m -0.2m [23a@RoyalIcing.UpperFlood.CheddarCatchment
1.36% 0.6m 46.4m 12 -0.5m 0.4m -0.1m [1a@RoyalIcing.UpperFlood.CheddarCatchment
1.26% 3.0m 240.4m 29 1.4m 2.7m 0.1m [1.29@Entrance.Goughs.CheddarCatchment
1.24% 0.1m 9.8m 2 -0.1m -0.0m -0.1m [2@DeepPoolTube.EastPassage.UpperFlood.CheddarCatchment
1.04% 0.5m 47.4m 5 -0.3m -0.4m -0.1m [1.14@Entrance.Goughs.CheddarCatchment
0.98% 0.2m 24.1m 10 -0.2m -0.1m 0.1m [1.10@Fonts.Goughs.CheddarCatchment
0.96% 1.1m 112.2m 12 -0.2m 1.0m 0.4m [1.24@Entrance.Goughs.CheddarCatchment
0.94% 0.0m 2.0m 2 0.0m -0.0m 0.0m [1.20@HighTime.2010.Charterhouse.CheddarCatchment
0.91% 1.0m 111.8m 11 0.3m -0.7m -0.6m [1.31@HeartbreakHill.Goughs.CheddarCatchment
0.91% 0.6m 70.7m 9 0.0m 0.6m 0.2m [1.20@Entrance.Goughs.CheddarCatchment
0.87% 0.3m 32.5m 5 -0.2m 0.2m 0.1m [2.4@SandChamber.Goughs.CheddarCatchment
0.85% 0.5m 62.7m 13 0.2m -0.5m 0.0m [1.0@Cheeseroom.Goughs.CheddarCatchment
0.82% 1.0m 125.8m 20 -0.7m -0.8m -0.1m [1.18@RipRiftLadder.2009.Charterhouse.CheddarCatchment
0.82% 0.7m 84.2m 26 -0.5m 0.4m 0.1m [9.7@FarRift.Goughs.CheddarCatchment
0.77% 0.1m 8.0m 5 0.0m -0.0m 0.0m [2.9@SideRift.Reservoir.CheddarCatchment
0.77% 0.5m 64.1m 7 -0.1m -0.5m -0.1m [1.12@StPauls.Goughs.CheddarCatchment
0.73% 0.4m 50.8m 11 0.2m 0.3m 0.1m [1.12@LloydHall.Goughs.CheddarCatchment
0.70% 0.2m 32.4m 9 -0.0m -0.2m -0.0m [1.7@Quicksand_Chamber.2009.Charterhouse.CheddarCatchment
0.70% 0.3m 41.8m 3 0.0m 0.3m 0.1m [1.26@HeartbreakHill.Goughs.CheddarCatchment
0.67% 0.4m 59.6m 7 0.4m -0.2m 0.1m [CC@Goughs_Old.Long_Hole.CheddarCatchment
0.67% 0.7m 102.8m 19 0.6m 0.3m 0.0m [1.8@PotOxbow.2010.Charterhouse.CheddarCatchment
0.67% 0.7m 100.6m 15 0.4m 0.5m 0.3m [1.7@TheOxbow.Goughs.CheddarCatchment
0.61% 0.1m 19.1m 9 0.1m -0.0m 0.0m [1.28@Bone2.BoneHole.CheddarCatchment
0.60% 0.2m 35.7m 2 -0.2m 0.1m -0.0m [1.18@HighTime.2010.Charterhouse.CheddarCatchment
0.56% 0.4m 76.9m 11 -0.2m 0.4m 0.0m [7.0@Tunnel.Goughs.CheddarCatchment
0.53% 0.2m 29.4m 8 0.1m 0.1m 0.1m [3.14@BoulderChamber.Goughs.CheddarCatchment
0.53% 0.2m 30.6m 8 -0.1m 0.1m 0.0m [2.3@SandChamber.Goughs.CheddarCatchment
0.52% 0.1m 10.2m 2 0.0m -0.0m -0.0m [1.7@HighTime.2010.Charterhouse.CheddarCatchment
0.51% 0.5m 105.1m 17 -0.2m 0.5m 0.1m [1.12@PotOxbow.2010.Charterhouse.CheddarCatchment
0.51% 0.4m 78.1m 22 0.2m 0.3m 0.1m [9.21@FarRift.Goughs.CheddarCatchment
0.49% 0.1m 18.0m 7 -0.0m 0.0m 0.1m [1.12@UpperBeddingChamber.CharterhouseWarren.CheddarCatchment
0.44% 0.0m 2.0m 2 -0.0m -0.0m 0.0m [1.21@HighTime.2010.Charterhouse.CheddarCatchment
0.44% 0.1m 17.4m 6 0.0m -0.1m -0.0m [1.21@ChillOutChoke.2008.Charterhouse.CheddarCatchment
0.43% 0.1m 26.9m 4 0.0m 0.0m -0.1m [1.27@Rip_Rift2.2009.Charterhouse.CheddarCatchment
0.38% 0.2m 46.0m 14 -0.0m -0.1m 0.2m [2.16@SandChamber.Goughs.CheddarCatchment
0.38% 0.0m 10.2m 2 0.0m -0.0m -0.0m [1.6@HighTime.2010.Charterhouse.CheddarCatchment
0.37% 0.3m 86.8m 19 -0.1m 0.1m -0.3m [14@FullLRUD.EastPassage.UpperFlood.CheddarCatchment
0.37% 0.0m 6.6m 2 0.0m -0.0m -0.0m [1.20@HighTime.2010.Charterhouse.CheddarCatchment
0.33% 0.9m 281.6m 40 0.8m 0.3m 0.2m [1.0@Zebra.2010.Charterhouse.CheddarCatchment
0.33% 0.0m 11.7m 2 0.0m 0.0m -0.0m [1.24@HighTime.2010.Charterhouse.CheddarCatchment
0.32% 0.2m 77.0m 16 -0.2m 0.1m 0.2m [2.0@SandChamber.Goughs.CheddarCatchment
0.31% 0.1m 18.4m 2 0.0m 0.0m 0.0m [1.15@HighTime.2010.Charterhouse.CheddarCatchment
0.29% 0.1m 29.0m 2 0.1m 0.1m 0.0m [1.8@HighTime.2010.Charterhouse.CheddarCatchment
0.27% 0.1m 34.5m 2 -0.1m 0.0m 0.0m [1.18@HighTime.2010.Charterhouse.CheddarCatchment
0.27% 0.2m 80.1m 15 -0.1m 0.1m -0.1m [1.34@Blackcat.Goughs.CheddarCatchment
0.26% 0.0m 9.0m 2 -0.0m 0.0m -0.0m [1.1@HighTime.2010.Charterhouse.CheddarCatchment
0.26% 0.1m 30.5m 2 0.0m 0.1m 0.1m [1.24@HighTime.2010.Charterhouse.CheddarCatchment
0.26% 0.0m 14.0m 2 -0.0m -0.0m 0.0m [1.14@HighTime.2010.Charterhouse.CheddarCatchment
0.25% 0.2m 77.3m 7 -0.1m 0.0m -0.1m [1.75@Entrance.OldCave.Charterhouse.CheddarCatchment
0.25% 0.0m 9.0m 2 -0.0m -0.0m 0.0m [1.2@HighTime.2010.Charterhouse.CheddarCatchment
0.25% 0.0m 18.1m 2 0.0m 0.0m 0.0m [1.9@HighTime.2010.Charterhouse.CheddarCatchment
0.25% 0.0m 14.0m 2 -0.0m -0.0m 0.0m [1.13@HighTime.2010.Charterhouse.CheddarCatchment
0.24% 0.8m 342.7m 62 -0.7m -0.4m 0.2m [1.0@Rip_Rift1.2009.Charterhouse.CheddarCatchment
0.23% 0.1m 30.5m 2 0.0m 0.1m 0.0m [1.25@HighTime.2010.Charterhouse.CheddarCatchment
0.23% 0.0m 3.2m 2 -0.0m -0.0m 0.0m [1.22@HighTime.2010.Charterhouse.CheddarCatchment
0.22% 0.1m 35.4m 9 0.0m 0.0m -0.1m [3.6@BoulderChamber.Goughs.CheddarCatchment
0.22% 0.1m 29.9m 2 0.1m -0.0m 0.0m [1.6@HighTime.2010.Charterhouse.CheddarCatchment
0.21% 0.0m 6.1m 2 -0.0m -0.0m 0.0m [1.26@HighTime.2010.Charterhouse.CheddarCatchment
0.21% 0.0m 6.9m 2 -0.0m 0.0m -0.0m [1.22@HighTime.2010.Charterhouse.CheddarCatchment
0.20% 0.2m 107.7m 20 -0.2m 0.1m -0.1m [7.3@Tunnel.Goughs.CheddarCatchment
0.20% 0.0m 8.7m 2 -0.0m 0.0m 0.0m [1.8@HighTime.2010.Charterhouse.CheddarCatchment
0.19% 0.0m 8.8m 2 -0.0m -0.0m -0.0m [1.1@HighTime.2010.Charterhouse.CheddarCatchment
0.19% 0.0m 11.6m 2 -0.0m -0.0m 0.0m [1.13@HighTime.2010.Charterhouse.CheddarCatchment
0.18% 0.1m 30.7m 5 0.0m 0.0m -0.0m [7.1@BeddingChamber.Goughs.CheddarCatchment
0.18% 0.1m 39.1m 2 0.0m 0.0m -0.1m [1.3@HighTime.2010.Charterhouse.CheddarCatchment
0.18% 0.1m 34.5m 2 -0.0m 0.0m 0.0m [1.17@HighTime.2010.Charterhouse.CheddarCatchment
0.17% 0.0m 8.7m 2 0.0m 0.0m -0.0m [1.8@HighTime.2010.Charterhouse.CheddarCatchment
0.17% 0.0m 17.6m 2 -0.0m -0.0m 0.0m [1.26@HighTime.2010.Charterhouse.CheddarCatchment
0.17% 0.0m 13.0m 2 -0.0m 0.0m 0.0m [1.5@HighTime.2010.Charterhouse.CheddarCatchment
0.16% 0.1m 79.5m 11 -0.1m 0.1m 0.0m [1.81@Entrance.OldCave.Charterhouse.CheddarCatchment
0.16% 0.0m 22.8m 2 0.0m 0.0m 0.0m [1.16@HighTime.2010.Charterhouse.CheddarCatchment
0.16% 0.1m 37.5m 6 -0.0m 0.0m 0.0m [1.31@Blackcat.Goughs.CheddarCatchment
0.15% 0.2m 129.7m 21 -0.1m 0.0m 0.2m [1.0@StPauls.Goughs.CheddarCatchment
0.15% 0.0m 6.9m 2 -0.0m -0.0m -0.0m [1.23@HighTime.2010.Charterhouse.CheddarCatchment
0.15% 0.0m 6.1m 2 -0.0m -0.0m -0.0m [1.25@HighTime.2010.Charterhouse.CheddarCatchment
0.14% 0.0m 29.6m 2 -0.0m 0.0m 0.0m [1.4@HighTime.2010.Charterhouse.CheddarCatchment
0.14% 0.0m 22.1m 9 -0.0m -0.0m 0.0m [1.0@Bone2.BoneHole.CheddarCatchment
0.13% 0.0m 11.6m 2 -0.0m -0.0m 0.0m [1.12@HighTime.2010.Charterhouse.CheddarCatchment
0.13% 0.0m 14.5m 6 -0.0m -0.0m -0.0m [1.17@UpperBeddingChamber.CharterhouseWarren.CheddarCatchment
0.13% 0.0m 14.4m 2 0.0m 0.0m -0.0m [1.16@HighTime.2010.Charterhouse.CheddarCatchment
0.12% 0.0m 3.2m 2 0.0m -0.0m -0.0m [1.21@HighTime.2010.Charterhouse.CheddarCatchment
0.12% 0.0m 18.4m 2 0.0m -0.0m 0.0m [1.13@HighTime.2010.Charterhouse.CheddarCatchment
0.11% 0.0m 14.4m 2 0.0m 0.0m -0.0m [1.17@HighTime.2010.Charterhouse.CheddarCatchment
0.11% 0.0m 36.6m 6 0.0m -0.0m 0.0m [1.26@100Fathom.2008.Charterhouse.CheddarCatchment
0.11% 0.0m 29.6m 2 -0.0m 0.0m 0.0m [1.3@HighTime.2010.Charterhouse.CheddarCatchment
0.11% 0.0m 29.9m 2 -0.0m 0.0m 0.0m [1.5@HighTime.2010.Charterhouse.CheddarCatchment
0.08% 0.0m 29.0m 2 0.0m 0.0m -0.0m [1.7@HighTime.2010.Charterhouse.CheddarCatchment
0.07% 0.0m 11.7m 2 0.0m 0.0m -0.0m [1.23@HighTime.2010.Charterhouse.CheddarCatchment
0.07% 0.0m 18.1m 2 0.0m 0.0m 0.0m [1.12@HighTime.2010.Charterhouse.CheddarCatchment
0.07% 0.0m 13.0m 2 0.0m 0.0m -0.0m [1.4@HighTime.2010.Charterhouse.CheddarCatchment
0.06% 0.0m 39.1m 2 -0.0m 0.0m -0.0m [1.2@HighTime.2010.Charterhouse.CheddarCatchment
0.06% 0.0m 18.0m 5 -0.0m -0.0m 0.0m [1.31@ChillOutChoke.2008.Charterhouse.CheddarCatchment
0.05% 0.0m 8.8m 2 0.0m -0.0m -0.0m [1.0@HighTime.2010.Charterhouse.CheddarCatchment
0.05% 0.0m 17.6m 2 -0.0m -0.0m 0.0m [1.27@HighTime.2010.Charterhouse.CheddarCatchment
0.05% 0.0m 6.6m 2 0.0m -0.0m -0.0m [1.17@HighTime.2010.Charterhouse.CheddarCatchment
0.04% 0.0m 22.8m 2 -0.0m 0.0m -0.0m [1.15@HighTime.2010.Charterhouse.CheddarCatchment
0.03% 0.1m 214.8m 24 -0.0m 0.0m 0.1m [14@PatonPlace.TyningsBarrow.CheddarCatchment
0.02% 0.1m 239.8m 41 -0.1m -0.0m -0.0m [1.0@RipRiftLadder.2009.Charterhouse.CheddarCatchment
0.00% 0.0m 33.4m 3 -0.0m -0.0m -0.0m [LW-62@Longwood_old.Longwood.CheddarCatchment
0.00% 0.0m 191.8m 20 0.0m 0.0m 0.0m [LW-7@Longwood_old.Longwood.CheddarCatchment
0.00% 0.0m 3.6m 2 0.0m 0.0m 0.0m [53@Main.TyningsBarrow.CheddarCatchment

Offline Ouan

  • addict
  • **
  • Posts: 155
    • Thailand Caves
Re: Accuracy Questions
« Reply #14 on: October 16, 2012, 10:40:51 am »
Slightly off topic, but to do with accuracy and the BCRA grades.

How are people calibrating their DistoX?
The calibration routine only calibrates to magnetic north.
For BCRA Grade 5 it should be calibrated to grid or true north.
A calibration shot in or near the cave every survey trip only corrects for variation in magnetic north.

Offline jarvist

  • obsessive maniac
  • ***
  • Posts: 424
  • Imperial College CC / Northern Pennine Club
    • Imperial College Caving Club
Re: Accuracy Questions
« Reply #15 on: October 16, 2012, 12:19:28 pm »
The calibration routine only calibrates to magnetic north.
For BCRA Grade 5 it should be calibrated to grid or true north.
Ah, I wouldn't call this calibration - this is applying magnetic declination. OS maps should state the declination for a given year, and the rate of change.

For Slovenia, I've used this American service (using the International Geomagnetic Reference Field), to calculate for specific years and locations:
http://www.ngdc.noaa.gov/geomag-web/#declination

This doesn't compensate for any local magnetic anomalies.

For utmost accuracy you'd want a local _measurement_ of the magnetic declination, but as its a systematic correction you need it measured to a far higher accuracy than your individual survey shots - i.e. if you try and do it with your inaccurate hand held instruments you'll just introduce additional error.

Offline Rhys

  • Global Moderator
  • forum hero
  • *****
  • Posts: 1476
  • SWCC, RFDCC
Re: Accuracy Questions
« Reply #16 on: October 16, 2012, 12:31:43 pm »
For magnetic declination in the UK, I use this:

http://www.geomag.bgs.ac.uk/data_service/models_compass/gma_calc.html

It's far easier than trying to decipher the description in the margin of a map that might be many years old.

Rhys

Offline Ouan

  • addict
  • **
  • Posts: 155
    • Thailand Caves
Re: Accuracy Questions
« Reply #17 on: October 16, 2012, 01:22:37 pm »
Perhaps I should have gone to the BCRA website first...

"To attain Grade 5 it is essential for instruments to be properly calibrated"

Nice and vague as it doesn't define what a proper calibration is.

Offline graham

  • Retired
  • forum hero
  • *****
  • Posts: 10943
  • UBSS, Speleo-Club de Perigueux, GSG, SUI
    • UBSS
Re: Accuracy Questions
« Reply #18 on: October 16, 2012, 01:27:20 pm »
Perhaps I should have gone to the BCRA website first...

"To attain Grade 5 it is essential for instruments to be properly calibrated"

Nice and vague as it doesn't define what a proper calibration is.

It's the opposite of improper calibration, of course.

More seriously, it will, of course vary from instrument to instrument so it's fair enough not to state. What should be in there, though are what the acceptable errors are.
Caving is for Life not just for Christmas

Offline cavermark

  • forum hero
  • *****
  • Posts: 1382
Re: Accuracy Questions
« Reply #19 on: October 17, 2012, 02:10:03 am »
Perhaps I should have gone to the BCRA website first...

"To attain Grade 5 it is essential for instruments to be properly calibrated"

Nice and vague as it doesn't define what a proper calibration is.

For compass and clino doing a leg between two fixed points (e.g. 2 trees 20m apart) then shooting back the other way will indicate if your instruments are ok in relative terms. Compass bearings should be 180 degrees different and clinos the same angle just positive rather than negative or vice versa.

Offline graham

  • Retired
  • forum hero
  • *****
  • Posts: 10943
  • UBSS, Speleo-Club de Perigueux, GSG, SUI
    • UBSS
Re: Accuracy Questions
« Reply #20 on: October 17, 2012, 08:01:39 am »
For compass and clino doing a leg between two fixed points (e.g. 2 trees 20m apart) then shooting back the other way will indicate if your instruments are ok in relative terms. Compass bearings should be 180 degrees different and clinos the same angle just positive rather than negative or vice versa.

I have old calibration data from RAF astro-compasses used in the 1950s which shows different errors in different quadrants, so something more complex than that may be required. I do not know how the collected that data nor what they calibrated the instruments against, but I can see that certain systemic faults, such as a pivot point which is not quite central, will give a variable error around the card.
Caving is for Life not just for Christmas

Offline cavermark

  • forum hero
  • *****
  • Posts: 1382
Re: Accuracy Questions
« Reply #21 on: October 17, 2012, 08:08:34 am »
For compass and clino doing a leg between two fixed points (e.g. 2 trees 20m apart) then shooting back the other way will indicate if your instruments are ok in relative terms. Compass bearings should be 180 degrees different and clinos the same angle just positive rather than negative or vice versa.

I have old calibration data from RAF astro-compasses used in the 1950s which shows different errors in different quadrants, so something more complex than that may be required. I do not know how the collected that data nor what they calibrated the instruments against, but I can see that certain systemic faults, such as a pivot point which is not quite central, will give a variable error around the card.

I wonder what amount of error you are talking about there Graham - is it likely to make a significant impact on cave surveying (e.g. less than half a degree)?

Would things like off centre pivot points not be due to manufacturing and within a tolerance to be sold? Or would it occur during use?

Offline Cave_Troll

  • forum hero
  • *****
  • Posts: 1592
  • SUSS
    • http://www.flickr.com/photos/cave_troll/
Re: Accuracy Questions
« Reply #22 on: October 17, 2012, 09:26:15 am »
I have old calibration data from RAF astro-compasses used in the 1950s which shows different errors in different quadrants, so something more complex than that may be required. I do not know how the collected that data nor what they calibrated the instruments against, but I can see that certain systemic faults, such as a pivot point which is not quite central, will give a variable error around the card.

Over what distance were those differences observed?
I'd expect a greater chance of different errors if you're looking at the errors at two points 10km apart.

Offline graham

  • Retired
  • forum hero
  • *****
  • Posts: 10943
  • UBSS, Speleo-Club de Perigueux, GSG, SUI
    • UBSS
Re: Accuracy Questions
« Reply #23 on: October 17, 2012, 11:05:07 am »
I wonder what amount of error you are talking about there Graham - is it likely to make a significant impact on cave surveying (e.g. less than half a degree)?

Would things like off centre pivot points not be due to manufacturing and within a tolerance to be sold? Or would it occur during use?

I'm afraid I don't have the stuff to hand, it's in our Library. Don't forget that these were ex-WD instruments and should have been of good quality.
Caving is for Life not just for Christmas

Offline graham

  • Retired
  • forum hero
  • *****
  • Posts: 10943
  • UBSS, Speleo-Club de Perigueux, GSG, SUI
    • UBSS
Re: Accuracy Questions
« Reply #24 on: October 17, 2012, 11:06:05 am »
Over what distance were those differences observed?
I'd expect a greater chance of different errors if you're looking at the errors at two points 10km apart.

IIRC the checks were done from the top of a tower & were looking at points about 1 km away in each direction.
Caving is for Life not just for Christmas

Offline jarvist

  • obsessive maniac
  • ***
  • Posts: 424
  • Imperial College CC / Northern Pennine Club
    • Imperial College Caving Club
Re: Accuracy Questions
« Reply #25 on: October 17, 2012, 01:02:08 pm »
There's a very interesting article somewhere (I can't remember where - I think Compass Points, but I couldn't find it yesterday reading the back archive), looking at sources of errors in mechanical instruments. Essentially they came down to offset errors such as the disk slipping, but also much more complex error caused by the jewelled pivot wearing out a (quite possibly non circular) hole in the card (particularly with the heavy clino - I seem to recall it mentioned that you have to do varying inclination readings when calibrating for this reason).

Personally I have not been able to detect any reproducible measurable difference (i.e. any calibration to apply) between the modern Suunto instruments held by Imperial, except for when the instrument is knackered to the point where it is retired anyway. As such I now just take multiple cross-checked measurements between them all to see if any of the compass cards are sticking etc.

However, we wear those poor little things out quickly taking them caving, so they're all pretty young - decades old club instruments may have been oddly magnetised while sitting in the library, balanced for the wrong hemisphere, used to bang in nails, badly jolted rattling in someone's car etc.

Graham - your WD compass certainly sounds like the effect of having some part of the enclosure becoming magnetised (I guess they didn't have such massive stocks of Alu or otherwise non-iron / ferromagnetic based metals back then?) - this happens inevitably to steel boats, and is calibrated by physically placing offsetting magnets in the binnacle.
« Last Edit: October 17, 2012, 01:16:46 pm by jarvist »

Offline jarvist

  • obsessive maniac
  • ***
  • Posts: 424
  • Imperial College CC / Northern Pennine Club
    • Imperial College Caving Club
Re: Accuracy Questions
« Reply #26 on: October 17, 2012, 03:41:01 pm »
Found the one on Suunto Compass errors with a suitably long google search... Can't find the info on the Clino - perhaps I imagined it?

"Errors In The Suunto Compass Used For Cave Surveying - Lang Brod", Compass Points 21
http://utter.chaos.org.uk/survex/cp/CP21/CPoint21.htm#art2

Unfortunately missing the figures...

Offline kdxn

  • menacing presence
  • **
  • Posts: 238
  • BPC, CPC
Re: Accuracy Questions
« Reply #27 on: October 17, 2012, 06:45:32 pm »
You might also want to check the NSS Compass and Tape archive:

http://www.caves.org/section/sacs/SACS/C_%26_T_Back_Issues.html

Some articles on Suunto readings, accuracy, bias and statistics.

Offline wookey

  • player
  • **
  • Posts: 105
Re: Accuracy Questions
« Reply #28 on: October 17, 2012, 10:03:26 pm »
And here is the smk system looplist from CUCC's austria dataset, including a lot of ARGE date. This is 822 loops in 97km of cave. Mostly done with compass+clino+tape, but a fair number of SAP, laser tape and distoX surveys in there now too. http://wookware.org/files/smk-system-loops.txt

It's a lot noisier than my earlier post suggested, with quite a lot of loops worse than 2%, even longer ones. An analysis by instrument type, caving club, surveyor would be interesting. I'd expect ARGE's data to be better and distoX data to be better. But that would require a bit more work than:

cavern smk-system.svx; grep Original smk-system.err | sort -n -k5 > smk-system-loops.txt; scp smk-system-loops.tx wookware:wookware/files

plenty of room for survex to provide stats in a more useful and malleable form.