> All the repeating, coming back the next day and multiple locations are nice, but what is the real distance between two points compared with the inversed distance of two RTK locations.
> From doing that I would say Kent's 0.04' is optimistic, but possible. Better than that, not so much.
The idea behind that demonstration is that if the standard error of coordinates generated by a particular system is more or less than 0.02 ft., then the value of the expected error goes up or down in proportion. So, if the coordinates are uncertain by 0.03 ft. (s.e.), then the expected error is 0.06 ft. and in about 5% of cases the error will be greater than about 0.10 ft.
I definitely agree that the best way to test system performance is by positioning many marks within, say, 100 ft. of each other (and in a real world environment) and comparing their relative positions to the much more accurate relative positions that can be gotten with a total station.
What many RTK users seem to be doing is assessing the accuracy of their systems by considering the mean of two or three measurements to be essentially without error (which of course is not the case) and treating the mean deviations as a significant measure of positioning quality, instead of using standard statistical techniques.
> > Were the residuals the total horizontal vector from the position to the mean or the differences between the N and E components and the mean?
>
> The latter. I computed them in a spreadsheet from the coordinates returned by TBC.
>
> It's probably worth noting that my unfamiliarity with the system leaves me wondering what kind of GPS survey I ended up with. When I pulled the job file into TBC, the points showed up with discrete vectors from individual reference stations, and I was asked to specify the desired geoid model.
The first observation I'd make is that if the spread between two N or E values obtained from 5 minute sessions at two different times is 1cm (0.033 ft.), then an unbiased estimate of the standard errors of each would be +/-0.023 ft. Since that estimate is based on such a small sample size (n=2, n-1 degrees of freedom), it has a very significant uncertainty itself (although that can be reduced by pooling the estimates from a larger number of repeated points).
I'll be interested to hear what TBC does with the observations since ultimately the object is to be able to assign realistic uncertainties to GPS positions, both to be able to combine them with other observations and to document compliance with standards.
I definitely agree that the best way to test system performance is by positioning many marks within, say, 100 ft. of each other (and in a real world environment) and comparing their relative positions to the much more accurate relative positions that can be gotten with a total station.
To really test the RTK system, make the points much closer, 20 feet or less (ten feet is great, but don't get so close that the total station/robot struggles). Set them in a easy place to measure, say a parking lot. The points then become almost foolproof to measure with a tape. This really tests the capability of RTK. Set the points in a way that you can measure all directions and figure out just what the RTK unit is doing.
If what you see is something you can live with; great. But that hasn't been my experience.
There are situations where an absolute position is more helpful than relative positions among marks. For example, if all your carefully set marks get washed away by Tropical Storm Irene and the remaining marks you referenced are things like "the centerline of town highway 13" and "a 10 inch basswood tree". No, I'm not a surveyor and no, I didn't get involved in reconstructing any property lines after the storm. But I was a crew member on the Red Cross disaster response vehicle from which I took the picture.
> To really test the RTK system, make the points much closer, 20 feet or less (ten feet is great, but don't get so close that the total station/robot struggles). Set them in a easy place to measure, say a parking lot. The points then become almost foolproof to measure with a tape. This really tests the capability of RTK. Set the points in a way that you can measure all directions and figure out just what the RTK unit is doing.
I'd think that you'd want the test points far enough apart that the multipath environment would be significantly different unless you are going to be locating them via RTK at different times.
The other thing that I'd think you'd want would be to locate them via a total station so that you end up with the control coordinates of the test points in a system that is also oriented in the same system as the RTK-derived coordinates, such as both on the same projection. That just means making sure that you have a high-accuracy azimuth to orient the total station shots from.
> I will add that I have seen some RTK systems do quite a bit better than .02N/.02E.
I wonder what method you used to estimate the standard errors of the positions those RTK systems. Would you care to elaborate ?
I'd think that you'd want the test points far enough apart that the multipath environment would be significantly different unless you are going to be locating them via RTK at different times.
The other thing that I'd think you'd want would be to locate them via a total station so that you end up with the control coordinates of the test points in a system that is also oriented in the same system as the RTK-derived coordinates, such as both on the same projection. That just means making sure that you have a high-accuracy azimuth to orient the total station shots from.
Yes, there are certinally other tests to do. And I have done them over and over.
But, if I want to know if I can set two property corners close together, or locate existing ones close together, then I want that test also. In that case I don't care about multi-path or coordinate systems.
Consider this: can you measure on flat ground a distance of 19.51 feet ten out of ten times with a tape within .02'? Can you measure that same distance with RTK ten out of ten times within .02'?
My answer would be yes and no.
> Consider this: can you measure on flat ground a distance of 19.51 feet ten out of ten times with a tape within .02'? Can you measure that same distance with RTK ten out of ten times within .02'?
>
> My answer would be yes and no.
Well, certainly an RTK system that can't deliver acceptable accuracy of relative position at separations of under 20 feet isn't going to get any better at 100 feet.
I was thinking more about designing a test that would characterize the random errors of the RTK-derived positions so that various other valid inferences can be drawn by purely statistical methods.
LOL Kent wants the manufacturer of the equipment to assign realistic uncertainties to it ;-).
> LOL Kent wants the manufacturer of the equipment to assign realistic uncertainties to it.
Yes, the smoke and mirrors of marketing black box technology isn't really on board with that idea since it kills the sale to say "now you won't meet ALTA standards with this, you know." When Jim imported the observations into TBC, the manufacturer may have actually provided the data necessary to answer the question. Naturally, the variances may still be typically optimistic, but a general least squares adjustment provides a way of rooting out that optimism when conventional measurements are combined with the network RTK positions.
> Here's the answer to the problem that dealt with the repeatability with which a survey marker can be replaced via some positioning process like network RTK that doesn't involve any tie to any other local reference point. I think you'll get a kick out of the most likely error.
> 3. You are called back, say, a year later to remark the corner and use the same technology that can measure the N and E values of coordinates in nominally the same system with the same uncertainties of +/- 0.02 ft. (standard error) each. Say it's network RTK.
> [ . . . ]
> What is the most probable error that you will make in remarking the corner? That is, what is the most likely distance that the new marker you set with, say, network RTK that can measure coordinates with standard errors of, say, +/-0.02 ft. will be from the original position of the marker (of which no trace remains other than the previously determined coordinates)?
>
> ANSWER :
>
> The most likely error in the replaced marker is ... 0.04 ft! In 1000 trials, fully 25% of the replacements were in error by a distance in the range of 0.036 ft. to 0.044 ft.
In an area with significant crustal motion, especially if an episodic event (earthquake) occurred in the year before remarking the corner, the error could be considerably more than 0.04'
> In an area with significant crustal motion, especially if an episodic event (earthquake) occurred in the year before remarking the corner, the error could be considerably more than 0.04'
Yes, without a doubt. Likewise, if the positioning system wasn't actually delivering coordinates with standard errors of +/-0.020 ft. N and +/-0.020 ft. E the expected error could be considerably more. The value of 0.020 ft. was chosen to show that even in a highly optimistic case the most likely error in replacing a marker was surprisingly large. Knowing that you expect to miss a corner by 0.04 ft. pretty well shoots down network RTK for most high-value urban land I'm familiar with.
I offered to previously. I have spread sheets from the various manufacturers I have reviewed over nearly 5 years. I'll start a new thread...
:good:
> I offered to previously. I have spread sheets from the various manufacturers I have reviewed over nearly 5 years. I'll start a new thread...
Sure, that'd be great. Be sure to mention the methodology used.
I enjoy many of the threads on this site and ignore others. Ocassionally I am so impressed by the quality of discussion in a thread that I copy and save it on my computer for future reference. This thread I saved. Thanks to those who contributed.
Kent brings up a good point in that we should know what the expected errors are for any measurement method we use.
This does not mean the RTK or RTN is bad but it may not be appropriate for certain tasks.
Least Squares allows us to actually know how well our various measurements fit together.
I think for topo and monitoring wells and other things which don't require subcentimeter position then these systems are perfectly acceptable. But for boundary tasks that should require smaller errors then we should be using a more appropriate method.
Even with static methods our productivity is light years ahead of where we were in 1970 so spending an extra day or two for a product that will be used for the next 50 to 100 years just does not seem unreasonable to me.
I turn sets for control and boundary monuments and use single angles for topo shots. It is the same sort of thing.
However, some boundaries may not require a high level of precision and accuracy. RTK/RTN may be appropriate in some of those cases. It just depends on the requirements of the situation as determined by the Land Surveyor. I am using a staff compass to run between existing monuments for the purpose of marking large trees. It is fine for that purpose if care and checking are used too. If I find monuments missing then I will get the Total Station out and traverse it to replace those. Be realistic about the capabilities of the system in use and select methods which are appropriate for the task at hand.
:good:
