Nate The Surveyor, post: 382936, member: 291 wrote: Because apparently you have not done what I have done. Have you used and tested RTK GPS yourself? Which one? Have you tested the Javad unit? I have.
Actually, what you described is something far short of testing. It's a cute anecdote, but not much more. Sorry, but that's the way it is.
Kent McMillan, post: 382924, member: 3 wrote: Actually, to be clear, the tough pill to swallow is how carelessly RTK tends to be used and how little RTK users seem to be able to document about the uncertainties of RTK-derived positions aside from "I checked those numbers and they looked good", which is a subprofessional answer. The whole trend toward "number box" technologies devoid of any real Q/A trail is bizarre from a professional perspective, to say the least.
I understand your concern, but would argue that at some point it becomes irrelevant. There were those who claimed raw data files could never replace the written notes in a fieldbook. The fact is that if the position is repeatable within a given tolerance as shown by repeated measurements, then a least squares analysis really doesn't add anything to the analysis, and in fact may be inferior to the actual repetition of a certain location. It is bizarre only in the sense that it has actually become possible. But it is possible not because of lack of Q/A, but because of its inclusion. The Q/A is merely of a differing form than we may be used to, and of course all processes can be abused. As you yourself have noted, a least squares analysis can be manipulated very easily to say what you want it to say. Manipulating the standard error of equipment in a least squares is similar to manipulating the time spent and the settings used in an rtk position. Either can be done randomly, or properly arrived at through experimentation.
Duane Frymire, post: 382941, member: 110 wrote: The fact is that if the position is repeatable within a given tolerance as shown by repeated measurements, then a least squares analysis really doesn't add anything to the analysis, and in fact may be inferior to the actual repetition of a certain location.
I think that the more professionally defensible answer is that as we rely increasingly upon the geodetic coordinates of things, those reports of positions are of value only to the extent that they have uncertainties attached to them and are known to be reliable within those uncertainty limits. Just repeating an RTK shot in some uncharacterized environment that differs by less than 0.10 ft. horizontally is grossly inferior to a process that is based upon an uncertainty estimate derived by accepted procedures that is validated by some independent means.
I do realize that the RTK bus is being driven by surveyors wanting to mainly work faster with accuracy and reliability as secondary interests at best, but when conventional observations are going to have to be combined with GPS-derived positions, the independent validation of uncertainties comes along for essentially no additional cost if one has the means to do it.
No one has mentioned this yet either but there is a relative positional accuracy calculator built in that can test the RPA of any given shot or averaged position.
I do not use least squares adjustment but I do need to be able to prove to the state board that I meet their requirements for boundary surveying. This is an exportable form that checks all selected shots to see if they fall within the user defined tolerance. No it is not an adjustment but it for sure more of a check than just reshooting corners over and over.
mattsib79, post: 382946, member: 1138 wrote: No one has mentioned this yet either but there is a relative positional accuracy calculator built in that can test the RPA of any given shot or averaged position.
I do not use least squares adjustment but I do need to be able to prove to the state board that I meet their requirements for boundary surveying. This is an exportable form that checks all selected shots to see if they fall within the user defined tolerance. No it is not an adjustment but it for sure more of a check than just reshooting corners over and over.
If you are performing ALTA/NSPS surveys, though, you'll need better evidence than just the raw processor estimates. The power of adjusting GPS vectors, whether static, PPK, or RTK in combination with conventional distances, angles, and height differences is that the conventional observations provide an excellent means of validating the realism of the raw processor estimates or of showing that they are not realistic and altering them accordingly.
Kent McMillan, post: 382947, member: 3 wrote: If you are performing ALTA/NSPS surveys, though, you'll need better evidence than just the raw processor estimates..
Golly Kent, didn't you know that RTK processes everything and makes the point "spot on" without all that silly "kneading and rolling" you talk about? C'mon man, get with the program..
We have not only integrated a sophisticated magnetic locator in the TRIUMPH-LS, but we have also streamlined the whole process. First the ÛÏStakeoutÛ screen will guide you toward the target. Then the ÛÏMagÛ screen locates your underground target and gives you its estimate of the coordinates of the underground target and a button to save it ÛÏas stakedÛ. And finally in the ÛÏCollectÛ screen you can survey the target point.
I would guess that there would be various interpretations of the above to various readers. There is no question that Javad has been and is one of the major innovators in GPS . He has also always paid attention to price point to allow surveyors enter the field of geodetic positioning. There is always a setback of quality work when the novice or untrained consumer expects total precision and accuracy.
With the aging of the surveyor population maybe in 10 years or so with self guided RTK and vehicles, surveyors could be working well into their nineties.
wrote: We have not only integrated a sophisticated magnetic locator in the TRIUMPH-LS, but we have also streamlined the whole process. First the ÛÏStakeoutÛ screen will guide you toward the target. Then the ÛÏMagÛ screen locates your underground target and gives you its estimate of the coordinates of the underground target and a button to save it ÛÏas stakedÛ. And finally in the ÛÏCollectÛ screen you can survey the target point.
There you go! Now all that's missing is the Post-fit Boundary Utility that will "process" the magnetic signals against some stored representation of the boundary to rotate and scale the boundary picture to fit the found magnetic signals. Add in the Video Mapper Utility and the small parcel survey that costs hundreds of dollars it A THING OF THE PAST!
Kent McMillan, post: 382943, member: 3 wrote: I think that the more professionally defensible answer is that as we rely increasingly upon the geodetic coordinates of things, those reports of positions are of value only to the extent that they have uncertainties attached to them and are known to be reliable within those uncertainty limits. Just repeating an RTK shot in some uncharacterized environment that differs by less than 0.10 ft. horizontally is grossly inferior to a process that is based upon an uncertainty estimate derived by accepted procedures that is validated by some independent means.
I do realize that the RTK bus is being driven by surveyors wanting to mainly work faster with accuracy and reliability as secondary interests at best, but when conventional observations are going to have to be combined with GPS-derived positions, the independent validation of uncertainties comes along for essentially no additional cost if one has the means to do it.
I may not understand this correctly, but it seems this would eliminate any and all rtk locations; similar to elimination of any and all sideshots with conventional equipment. That does seem a bit extreme. But I guess it shouldn't be all that difficult either. All the data necessary to process a least squares is being collected for each rtk shot with its corresponding CORS shot. So the G files are there if one wanted to use them. Am I correct that an rtk location can't be analyzed with a least squares program (at least in a meaningful way)?
Does anyone know of a Carlson tutorial on how to do that with their least squares program?
Kent McMillan, post: 382934, member: 3 wrote: The question then is how you adjust RTK-derived positions with conventional measurements and static vector solutions by least squares, which is what professional practice would require
Kent,
Did you miss the part that we are obtaining static vectors...that will load into you favorite least square adjustment software?
Kent McMillan, post: 382953, member: 3 wrote: There you go! Now all that's missing is the Post-fit Boundary Utility that will "process" the magnetic signals against some stored representation of the boundary to rotate and scale the boundary picture to fit the found magnetic signals. Add in the Video Mapper Utility and the small parcel survey that costs hundreds of dollars it A THING OF THE PAST!
Must be a lot of peyote in that part of Texas
I don't get the train of thought that because a position is repeatable it is accurate. You simply really don't 100% know and are assuming based on past experience. I'll use the title insurance analogy here. Title companies only go back 20 - 30 years as they rarely see problems crop up longer than that. The few times they get bit for not going back will be outweighed by the money they save doing it quickly and cheaply. RTK is no different, its a "calculated/insured risk". Whether you decide to assume this risk based on your experience or beliefs is up to you.
I am envisioning the following exchange:
JUDGE/ATTORNEY:
Please tell the court how you arrived at the distance between the two corners.
SURVEYOR:
Well I placed my new $40,000 RTK device on each corner and it told me it was okay to store the location and move on to the next point.
JUDGE?ATTORNEY
What do you mean by that
SURVEYOR
Well it averaged the location a bunch of times and the results were all with 0.10' so it must be correct
JUDGE/ATTORNEY:
So once again did YOU actually measure the distance between the corners
SURVEYOR:
Well no not in the traditional sense, I let the black box tell me the coordinates were okay to store and move on. As I'm not privy to the software code nor would be able to explain the mathematics involved, I relied on the manufacture to calculate the actual measurement, I just report what it tells me
JUDGE/ATTORNEY
So the box measured the distance, how do you know the box is correct.
SURVEYOR
Well it tells me the same answer every time. I have verified this on other jobs against conventional instruments.
JUDGE/ATTORNEY
So on those OTHER jobs the box agreed with conventional measurements, those other jobs are not why we are here, what about this job where you only used the box
SURVEYOR
well it worked great on the other jobs so it must of worked here
JUDGE?ATTORNEY
do you have an independent verification on this job that the box is correct
SURVEYOR
No, but I could go out there today and get the same answer
JUDGE?ATTORNEY
your dismissed
It is certainly an interesting concept and I could see the benefit of the integration, but at what cost? Is the locator as good as the others on the market? If not, I am not interested even if it will make the coffee in the morning and start the truck in the winter.
I think Javad is doing some innovative things, but man, his marketing efforts need some serious attention.
If I can 
just get these two to talk I won't even have to get out of my house. Staying in the truck all days are going to be a thing of the past. No really I like the idea.
Repeatable = accurate??
Is because it the Satellites are moving. It is NOT the same measurement, if the ROVER is there, with sufficient time. That IS what IS going on.
It is redundant. Statistical, and USABLE in Least Squares.
🙂
N
Duane Frymire, post: 382957, member: 110 wrote: Am I correct that an rtk location can't be analyzed with a least squares program (at least in a meaningful way)?
No, that's not the case. All that is required is for the RTK position to be stored as a position or vector with uncertainties expressed in some meaningful way that can be used as a basis for weighting the vector in a least squares adjustment. With GPS vectors, that uncertainty is typically expressed either as a variance-covariance matrix or as standard errors/correlations of the components. The same applies to positions derived from network RTK.
The uncertainties would typically be computed by the processor in the course of getting the solution that is the best estimate prior to adjustment with other conditions. Those processor estimates may or may not be realistic, so part of the purpose of the adjustment is to test them for realism through the residuals generated by the adjustment and to revise the uncertainties if the residuals show the processor estimates not to reflect reality.
John Evers, post: 382962, member: 467 wrote: Did you miss the part that we are obtaining static vectors...that will load into you favorite least square adjustment software?
So, this means that the RTK vectors won't? Why are other manufacturers able to do this and Javad not?
Kent McMillan, post: 382979, member: 3 wrote: So, this means that the RTK vectors won't? Why are other manufacturers able to do this and Javad not?
Wow...putting words into another mans mouth, because I exactly answered your question regarding having static vectors.
Of course you can export RTK vectors with their variance-covariance matrix.
edit...enough already. bye Kent
John Evers, post: 382983, member: 467 wrote: Of course you can export RTK vectors with their variance-covariance matrix.
Well, why didn't you say so? If your point was that adjusting a static solution from the same quickie occupation that generated the RTK vector would be a way of testing the processor-estimated weights of the RTK vectors, I think that is an overly optimistic theory.
What one actually finds when GPS vectors are combined in an adjustment with conventional observations of well-characterized uncertainties is that comparing one method to another that is utterly and completely independent is an excellent way to measure unmodeled errors that the processor-estimated uncertainties don't reflect, while, of course, improving the results considerably.
