Oh Great!
> While closure and repeatability of measurements are not a direct measurement of accuracy & precision, they offer better confidence in the quality of work than abstract analysis.
That's hardly true. First, the analysis of errors is not abstract, dealing as it does directly with the actual measurements of the survey. What is abstract is using a very indirect figure of merit like the error of closure and thinking that is directly connected to the actual error in the measurements.
I know of no surveyor who has any sophistication in the analysis of errors in survey measurements who has tossed least squares adjustments overboard and fallen back into using compass rule adjustments. The reason for this is the the former is just more useful.
Oh Great!
> The thing I am noticing, looking at page 105 of Ghilani et al, is that they are assuming you are shooting a fixed target for all pointings in your set. Whereas the bob-on-a-glass is not fixed -- it is in a random "best intent" spot for each shot, which hopefully averages to something that is very close to the point.
Well, if in your test the mean of the prism/bob pointings differed by 0.015 ft. from the better estimate of the direction, that would suggest that the the prism/bob is probably prone to systematic error that won't be minimized with repetition.
> With a fixed target that has the same centering error for each pointing, the centering error cannot be reduced in size by taking multiple pointings. (edit, took this out of quotes because it's a paraphrase, not a direct quote)
If you're going to use a mean value of the angle and distance in the adjustment, then the way to evaluate the centering error of the prism/bob is by computing the centering error of the same number of pointings as you intend to average.
> But with the handheld plumb bob (or the handheld rod) we are making what was a fixed error into a random error, one that differs in a random direction with each pointing, and can be reduced by taking multiple pointings.
If the mean error of a series of pointings doesn't actually approach zero, then any systematic error inherent in the method would have to be modeled in some way. How that is done depends upon the actual nature of the error.
Oh Great!
Nothing about the chi-squared test, huh?
Are we beyond the limits of your memorization there?
Oh Great!>HEY KENT!
hey Kent, a half bubble out is still a half bubble in! Kind of like the broken clock that is right twice a day 🙂 or the blind squirrel that can still find a nut 🙂 Looks like you're the blind squirrel this time! (found your nut)
Oh Great!>HEY KENT!
> hey Kent, a half bubble out is still a half bubble in!
Okay. At least he's trying to figure out what he's doing.
Oh Great!
> Nothing about the chi-squared test, huh?
No, sorry. The test procedure you described neglected the fundamental of actually determining the target centering errors of the prism pole in tripod. So, what you were comparing was the direction of a target centered by two processes of undetermined quality, one of which may have comparatively large systematic errors in it (bob/prism).
Oh Great!
> The test procedure you described neglected the fundamental of actually determining the target centering errors of the prism pole in tripod.
A much more straight forward test would be to actually point the instrument manually at the ground mark (a tack in a tall hub ought to work perfectly well) and then take angles and distances to the prism/bob that is being fished over the ground mark. From those, you can directly compute the centering errors.
Oh Great!
What you are repeatedly demonstrating is that:
a) you do not comprehend that those standard errors can be extracted from a large enough data set using the chi squared test, and
b) you are more concerned with making yourself look smart and making other people look wrong than making a useful contribution.
Oh Great!
> What you are repeatedly demonstrating is that:
>
> a) you do not comprehend that those standard errors can be extracted from a large enough data set using the chi squared test, and
It should be obvious that simply taking a whole bunch of pointings to a prism/bob fished over a ground mark doesn't tell you much of anything about the actual centering errors of the prism/bob with respect to the ground mark. Measuring the direction to the ground mark by pointing the instrument at it (manually) and then letting the robot measure to the prism/bob over it does.
It's such an easy test that I'm wondering why you haven't bothered to do it. Is there some good reason why not?
Kent is missing the point
> > While closure and repeatability of measurements are not a direct measurement of accuracy & precision, they offer better confidence in the quality of work than abstract analysis.
>
> That's hardly true. First, the analysis of errors is not abstract, dealing as it does directly with the actual measurements of the survey. .
Kent, When previously control points and monuments are resurveyed and the results are consistently withing error tolerances (compared with previous values), then you are checking the error of your entire surveying system and methodology. No amount of theoretical statistical analysis of individual survey component errors can ever achieve the confidence level of this empirical comparison.
Also, while good closures can be achieved by compensating errors, consistently good closures can not. While periodic calibration of equipment and baseline checks should not be discouraged, if it ain't broke, don't fix it.
Kent is missing the point
> Kent, When previously control points and monuments are resurveyed and the results are consistently withing error tolerances (compared with previous values), then you are checking the error of your entire surveying system and methodology. No amount of theoretical statistical analysis of individual survey component errors can ever achieve the confidence level of this empirical comparison.
Well, what you're doing is you're remeasuring the angle and distance between some existing marks that you previously surveyed and are finding that some "error tolerance" isn't exceeded? So, to arrive at that "error tolerance" you had to make some sort of error analysis, right? But you just don't want to do it with actual numbers?
I like the way that you use statistical terms like "confidence level" while disclaiming that statistical methods are of any use. :>
Oh Great!
> > What you are repeatedly demonstrating is that:
> >
> > a) you do not comprehend that those standard errors can be extracted from a large enough data set using the chi squared test, and
>
> It should be obvious that simply taking a whole bunch of pointings to a prism/bob fished over a ground mark doesn't tell you much of anything about the actual centering errors of the prism/bob with respect to the ground mark. Measuring the direction to the ground mark by pointing the instrument at it (manually) and then letting the robot measure to the prism/bob over it does.
>
> It's such an easy test that I'm wondering why you haven't bothered to do it. Is there some good reason why not?
If I accept that you are going to sidestep any mention of the chi squared test from here on out, will you agree to experiment with it in your star*net sometime at home while nobody is looking?
I already have long-developed, exhaustive statistics for the centering error of the rod-in-bipod and handheld rod to compare with the bob shots.
I did a more expedient field version to get the HI of the bob, and noted some rough tendencies about the horizontal while I was at it. You are confusing one field anecdote with a comparison of well-developed standard deviations.
Your test as proposed is not one I would have thought of. It would only measure direction, not distance, and the zeniths would not be comparable. Comparing the eyeball-pointing to the ATR in the robot is never fair. (yes, of course you do calibrate the ATR by eye, but that is under optimum conditions.) At anything less than about 200 feet the ATR is so much better than anyone's eyeball it is apples and oranges, so you want to be comparing ATR shots between the glass-on-rod and glass-on-bob.
So by going into the field for a couple of afternoons and shooting a few dozen shots with the bob, and looking at the least squares results, I did that. Happy with it. I will keep developing it.
The moral of the story for the original poster is this:
Whatever the new method is, try it. Tripods, bipods, skyhooks, faith surveying, whatever.
The data will tell you if it works or not. I have found that for my tiny universe, a handheld rod with a robot is as accurate as anything and means less stuff to carry. My plumb bob experiment is just a progression towards a more walkable rig. Shooting a few more sets makes up for the wobble and takes less time and energy than carrying the extra equipment.
I have pondered that by using the handheld rod or the plumb bob, you get more spread but you also get more truly independent measurements. Would you rather take sets to a fixed but slightly off-center target or to a floating target with better "intent" of being over the point?
Kent is missing the point
numbers?
>
> I like the way that you use statistical terms like "confidence level" while disclaiming that statistical methods are of any use. :>
Thanks for the compliment. I'm just trying to keep the discussion grounded in reality.
Kent is missing the point
> I'm just trying to keep the discussion grounded in reality.
Well, I'm not sure how that gets you there when the reality is that the measuring processes that surveyors use have quantifiable uncertainties. Basically, your point was that they don't, and while suggesting that an "error tolerance" is what you use to test the quality of your work. In reality, the two ideas are fundamentally incompatible.
Kent IS THE POINT
ok, it's all about you kent. when i was in elementary school, there was a boy that had to get the last word, under every circumstance. he crossed paths with another boy that didn't like what he had to say. about three seconds later, he found himself hanging by his underwear from a fence post. he left him there for about ten minutes; he got the point.
good thing surveyors are more civilized than that
my point is this, kent. you have a lot of knowledge, but you deliver it in such a down way. what's the general problem with some technical sharing without the insults?
Kent does not suffer fools lightly
It's an argument that doesn't make sense. Either errors are quantifiable (which I believe) or they are not in which case you can't have a tolerance error? How can you measure an unquantifiable subject?
But, Kent does start "nicely" usually, but when the argument gets pointed so does he. It reminds me of the old "Why are you so defensive?" "Because you are so offensive!" If attacked then defense is reasonable.
Kent does not suffer fools lightly
ok, Don. i would never consider HB or Perry fools, though. nor Kent, for that matter. the condescending remarks are beginning to grit like a poorly shucked oyster.
Oh Great!
> Your test as proposed is not one I would have thought of.
Which is why I mentioned it. It gets at the actual answer very efficiently and directly.
> It would only measure direction, not distance, and the zeniths would not be comparable.
Once you've determined what the centering error for the angle measurement, you can then assess whether the method is even good enough to consider using. The zeniths are an entirely separate issue from horizontal centering. There is no reason to test them together with the horizontal components of centering the prism/bob.
> Comparing the eyeball-pointing to the ATR in the robot is never fair. (yes, of course you do calibrate the ATR by eye, but that is under optimum conditions.) At anything less than about 200 feet the ATR is so much better than anyone's eyeball it is apples and oranges, so you want to be comparing ATR shots between the glass-on-rod and glass-on-bob.
What do you suppose that the angular difference between pointing with reticle and with ATR would be at a range of, say 20 ft.? I'd expect that it would not be a significant factor in the centering test.
Kent does not suffer fools lightly
LOL, nor do you (or should you ) take everything I say seriously 😉 If I really started calling names then I would be banned banned banned...
Everyone, me included, has the "fool" inside just waiting to burst out in full glory!!!
Oh Great!
So, still no comment on the chi squared test ?
Why should I continue to answer your questions when you dishonor mine?
If I cease posting, will you respond cogently to my previous questions?
I doubt it. Better check with your invisible friends. Maybe they can refer you to a book on manners.