Relative Exactness
> > > Give people some credit.
> >
> > I eagerly await the day when I find that an RTK survey has been made in a way that isn't markedly substandard in some significant way.
>
> Not looking is the surest way of not finding.
Oh, actually, I examine the quality of every survey I retrace. I appreciate your view that I should be calling up surveyors asking if they know of any RTK surveys that have been particularly well done so that I can ask the landowners for permission to resurvey the tract for research purposes. This is, however, not how the real world works.
> You have (by your own admission) never used it and have no idea of its capabilities (in the right hands) - only examples about the wrong hands. If it sucked as bad as you make it out to, it would have never survived.
Actually, I think the mechanism is that a significant fraction of the folks using RTK either don't know what sort of results they're getting or can plausibly deny all knowledge of the fact.
> I appreciate your extolling of the virtues of PPK; I find it invaluable when needed. But talk about a single hammer.
The real discussion is about using survey measurement techniques with well-characterized uncertainties in a way that allows realistic estimation of the uncertainties in results. Conventional measurements and post-processed GPS solutions definitely fit that bill. The average RTK user isn't concerned about all that technical, mathy-type stuff.
Relative Exactness
Kent,
If the surveyors in question can not do something as simple as set up his RTK base properly, or select a suitable coordinate system in his data collector, then I doubt he's going to fare any better trying to post process static work, or use least square adjustment software to analyze/improve the quality of his work.
You have a point that RTK may be a magnet for poorly trained crews pushing buttons and accepting whatever number pops up. But the failure to train is more along the lines of educating the crews about basic geodesy, coordinate systems, scale factors, geoids, blunder and error detection, etc ... pretty much the same stuff you have to know in order to do successful post processing and get results that resemble the real world.
I have no doubt that if you were ever to give RTK a try, you'd be able to master it pretty easily, given what you already have to know about GPS post processing/adjustments. In two hours, you'd probably be able to observe 30 second occupations on each of your monuments, at least twice, separated by an hour or so, which would give you more redundancy in your LSA. You'd be able to find a suitable DC package that would export co-variance data into Starnet so you could analyze your data in the same way you analyze your rapid static results.
Relative Exactness
> You have a point that RTK may be a magnet for poorly trained crews pushing buttons and accepting whatever number pops up.
> In two hours, you'd probably be able to observe 30 second occupations on each of your monuments, at least twice, separated by an hour or so, which would give you more redundancy in your LSA.
Actually, on that site the two occupations separated by an hour would add at least three or four hours to the total length of time when travel from point to point is taken into account. When you add in the logistics of either pre-surveying multiple base locations and moving the base or shuffling repeaters around, I'd probably double that number for RTK. On most large tracts post-processed GPS solutions are actually more efficient because radio link isn't a factor and the base doesn't need to be shifted. Of course, there is also the added bonus of improved quality control.
> You'd be able to find a suitable DC package that would export co-variance data into Starnet so you could analyze your data in the same way you analyze your rapid static results.
Yes, that certainly ought to be the case in the ideal world we can all imagine. My strong suspicion is that when RTK is rigorously analyzed, the quality of the work is just not as good as can be had by other means for similar effort. If you need coordinates right there, on the spot, RTK is pretty much necessary. If you need accurate coordinates right there, on the spot, RTK is likely to be an unhappy adventure.
Is 1:5000 really that accurate
I don’t have near the problem with your definition of exactness as I do with your statements at the end of the OP that seem to define 1:5000 as being extremely accurate; leastwise that's how I interpret them. And, I am intrigued that it takes a total station to readily determine a systemic RTK failure. I must have either missed something in my education/training/experience or lost something in my retirement.
Relative Exactness
> Uncertainty. The only true certainty here is that you have no direct hands on experience in the use of the object of your ire.
Actually, if you've followed the RTK threads at all, you'll know that the question of estimation of uncertainty has been raised quite a few times and the answers have been all over the board from folks who do rely upon RTK as their main means of positioning anything.
Beyond that, we have discussed ways of testing RTK processor estimates of vector uncertainty and I can think of only a couple of actual RTK users who seemed remotely aware of the topic. It's pretty clear that there is no correlation between use of RTK and familiarity with uncertainty estimation techniques beyond categorizing them as:
- good as it is gonna get, or
- whoopsy!
Relative Exactness
I can find things from most of the participants in this thread with which I agree:
1. RTK is a tool. When properly used within its limitations, it can be a valuable tool.
2. The user has to know the limitations of the tool, under what circumstances they're likely to adversely affect the work at hand, how to mitigate or avoid those affects, and how to test the work performed to ensure that it meets the professional standard of practice.
3. A significant percentage of RTK users don't know about 1 and 2 above, or choose to pretend they don't exist.
4. The percentage described in 3 have left -- and will continue to leave -- messes behind that will have to be cleaned up by someone. Those messes may appear soon after they leave the site, or they may take generations to come to light.
5. A few anectodes don't constitute a statistically valid sample.
Item 5 aside, I'd like to pass along a tale related to me by a colleague a couple of days ago. It concerns a roughly 1,000-acre master-planned development that his firm has been staking in assorted units for the last 10 years or so. Although the original boundary survey was measured with static GPS by a competent operator, most of the horizontal construction layout has been performed by an assortment of crews of varying levels of competence using RTK. In the last 5 or so years, a commercial RTN has been the source of positioning.
It seems that the crews have had only a limited understanding of datums, epochs and projections, but as the work progressed from north to south in the roughly mile-and-a-half long subdivision no one seemed to mind. However, as the most recent units adjoin the southern boundary, my colleague asked that the crew check into the boundary monuments just for fun. When a half-foot discrepancy appeared, there was much wringing of hands and gnashing of teeth.
The problem appears to be the result of a combination of grid-ground misunderstandings as well as a difference between the boundary epoch (HPGN 1999.51) and the RTN epochs (2007 at one point, then 2010 after a few years). The mess is awaiting the results of protracted discussions between my colleague (the surveyor in responsible charge), the engineering firm that produced the subdivision mapping, and the city in which this disaster is located. Proposed solutions range -- in order of increasing cost -- from filing a Record of Survey on the south units to remapping the entire 1,000 acres. (Fortunately, none of the parties has suggested the "tear it all out and start over" approach.) No one is happy, and the only thing certain is that a lot of money will be spent in reaching a resolution.
The problem is not attributable to RTK, but rather to poorly-trained field personnel operating under grossly inadequate supervision. The word "negligence" hasn't yet been advanced by anyone, but it's early days yet.
Relative Exactness
> The problem is not attributable to RTK, but rather to poorly-trained field personnel operating under grossly inadequate supervision.
:good: :good: :good:
Relative Exactness
> The problem is not attributable to RTK, but rather to poorly-trained field personnel operating under grossly inadequate supervision. The word "negligence" hasn't yet been advanced by anyone, but it's early days yet.
I would think that RTK is a special case under the doctrine of Attractive Nuisance. :>
Is 1:5000 really that accurate
Have you ever heard of sarcasm?
Is 1:5000 really that accurate
> I don’t have near the problem with your definition of exactness as I do with your statements at the end of the OP that seem to define 1:5000 as being extremely accurate; leastwise that's how I interpret them. And, I am intrigued that it takes a total station to readily determine a systemic RTK failure. I must have either missed something in my education/training/experience or lost something in my retirement.
So, are you saying that errors of 1:5000 or more are not consistent with excellent quality work?
As for the use of a total station, that is about the easiest way to pick up a systematic scale error in RTK work that I know of. In the example that I posted, the original survey was evidently made on a custom projection generated from an autonomous position at the base with the Y axis of the grid oriented very nearly geodetic North at the autonomous position and with the projection surface being some developable surface with a standard parallel on the apparent ellipsoid as computed from the autonmous position's ellipsoid height.
What that yielded were grid coordinates in a system that differed (on average) by roughly 160ppm from ground scale. There is no way to have detected that systematic scale error, except by some alternate known scale like that of a distance measurement with a total station.
So, if the surveyor of record sets above average monuments in Texas everything is ok. Or in fact, he produces and certifies substandard work, wouldn't you report him to the State Board. And this still has nothing to do with RTK quality, just project supervision.
Relative Exactness
From the data shown it looks like there may of being float solutions accepted, but to say that RTK fixed position are not accurate enough for boundary work is a joke.
> So, if the surveyor of record sets above average monuments in Texas everything is ok. Or in fact, he produces and certifies substandard work, wouldn't you report him to the State Board. And this still has nothing to do with RTK quality, just project supervision.
Actually, the responsible surveyor is no longer licensed in Texas.
I see this case as attributable to RTK and the willingness of a broad fraction of RTK users to have the person who sold a system to them tell them (or their field personnel) how to use it. I've heard one GPS salesman who represented a major manufacturer speak at a CEU SEMINAR and state that no surveyor would actually want to use the Texas Coordinate System, for example, but would want to cook up some bastardized coordinate system.
RTK ought to fall under the common law doctrine of Attractive Nuisance if it doesn't already.
Relative Exactness
> From the data shown it looks like there may of being float solutions accepted, but to say that RTK fixed position are not accurate enough for boundary work is a joke.
Not really. It depends upon what an acceptable relative uncertainty is and how the RTK system is being used to validate the results.
If the results are being used with no check and the uncertainty in the RTK position is not being tested (which I think describes the majority of RTK use), then it's fair to say that RTK fixed positions are in fact not good enough for boundary surveying. They aren't good enough, that is, if the work is to comply with many accepted accuracy standards.
Relative Exactness
There are situations in tree cover where one must be aware of possible fixed solutions that may be in error and need to take the appropriate measures to ensure that you have a correct location, but I have returned and located corners previously measured or set with RTK in conditions such as you referred to and have never found a difference of much more than 0.1'. I worry more about the rod bubble than RTK errors. This reminds me of older surveyors who were not to sure of the reliability of distance meters vs chained distances when the biggest problem was reflector offsets, temperature and barometric pressure.
Is 1:5000 really that accurate
Yeah, but sometimes I am a little slow in recognizing it. Maybe it was the exactness comments that threw me off.
Is 1:5000 really that accurate
So, are you saying that errors of 1:5000 or more are not consistent with excellent quality work?
I guess that depends on whether you find and accept the corner monuments. It also depends on the terrain, the nature of the monuments and the value of the property.
As for the use of a total station, that is about the easiest way to pick up a systematic scale error in RTK work that I know of.
So am I correct in concluding that your comparisons and conclusions are based upon the use of a total station? Or, is that also sarcasm?
[sarcasm]And this still has nothing to do with RTK quality, just project supervision.[/sarcasm]
Well, if you believe that it's purely coincidental that so much crappy RTK work is being generated and want to blame it on "poor supervision", that's fine. I think alternate explanations are much more useful.
The way that I've observed technology tends to be used in the offices at the survey factories is that it's the field staff who actually use it on a daily basis who know which button to push when and the "supervisor" is usually out of the loop. I think that's more of a comment on the technology than the supervisor.
I call BS
> I think that's more of a comment on the technology than the supervisor.
That like saying that misreading the venier on a transit is Gurley's fault!
Or NOT inputting the correct Barometric and temperature corrections is the fault of Total Station technology.
Get a clue Kent, or better yet, get RTK and move into the modern era.
Loyal
