Greetings,
What methods do people here use to analytically state the precision of their surveys when using RTK? I have a sense of the precision of the RTK equipment I am using by checking in to my own control and checking between NGS control in the area, but I'm not sure how I can state in statistical language what the precision is.
Since you're in my state, I suggest a review of AZ Boundary Survey Minimum Standards
It doesn't matter how or what you report, it just has to fit those. Remember that GPS is just another tool in the truck. I'll add that the MTS are currently under review for revisions through APLS, but it's not moving too fast.
State requirements aside, I suggest reiterating the manufacturer's specification, which is generally along the lines of 1 cm + 1 ppm horizontal and 2 cm + 1 ppm vertical, both at 1 sigma. In both cases, you might also want to specify the datum reference (i.e., whether the accuracy is with respect to a published datum or merely relative within the survey).
> Greetings,
>
> What methods do people here use to analytically state the precision of their surveys when using RTK? I have a sense of the precision of the RTK equipment I am using by checking in to my own control and checking between NGS control in the area, but I'm not sure how I can state in statistical language what the precision is.
Sort of depends on the purpose of stating the precision. ALTAs? state boundary survey requirements? Just for science?
If you shoot a network between your RTK points with a total station and put the whole mess in to just about ANY sort of least squares, you will get a much better "sense" of your precision. It will also give you the statistical proof & defensibility you need for the job file. Two independent measurement systems checking each other, and enough redundancy to develop some meaningful, defensible statistics. The GPS equivalent of steel tape diagonal checks in layout. Belt and suspenders.
Someone else can probably tell you how to do the same thing purely with RTK vectors and post-processing.
Some RTK units can store the vector data for a shot and you can re-process later, even with multiple base stations (Ian Wilson once told me that he would use a rover and two on-site bases plus nearby CORS to feel confident about RTK -- and send the static base data to OPUS, too, as a check.) Some RTK units seem to just store a coordinate and a sigma, but nothing you can post-process, which makes the checks with an independent method really important, and the least squares to combine the data also really important.
Which leads to how lately I have been pondering an odd definition of precision vs. accuracy:
If the total station network was shot to stand on its own (closed w/cross ties) you can see an amazing precision, right at the spec for the EDM usually. But you don't really know where it is on the Earth. Add some GPS measurements to the mix and suddenly the error ellipses for the combined network Located On The Earth are about as big as the GPS sigmas. In other words, you know your TS network is precise, but you can't pin the whole thing down on the earth any more accurately than your GPS measurements. So if you need the whole network to be Located On The Earth with the same precision as your total station work, you need also some redundant GPS points. Enough of those (average sigma of your GPS points divided by square root of the # of GPS shots tied to the observation network) and eventually it's Precisely Located On The Earth to your spec (until the next readjustment or big quake).
Hope this helps,
-- hb.
You may have to adjust some settings in your dc to have vectors and covariance matrix information saved in your raw data - rather than just coordinates. The when you load the raw data file into whatever office software you are using - LGO, TGO, TBC, Topcon Tools, or whatever it will yield some coordinate quality data along with the coordinates.
Some redundant data and a least squares adjustment report would be still better.
RTK is just another measurement system.
If you have redundancy then you can run an LS adjustment and see the statistics plus the residuals.
The manufacturer specifications are generally at greater than one sigma, with the significant exception of their specifications for their GIS or "resource grade" receivers, which were all speced at one sigma last time I looked.
As my late colleague Ellis Veatch used to point out in every training session he did: "Every RTK location is a side shot and you as a surveyor need to know what that means."
Without redundancy it is no better than the same as a single topo shot from your total station.
> The manufacturer specifications are generally at greater than one sigma
Both Trimble and Leica use RMS for their RTK accuracy specs. RMS is equal to 1 sigma when the mean is equal to the true value, which is not always the case with RTK. So the manufacturers are stating their RTK accuracy at 1 sigma best-case.
Just to be a stickler for nomenclature, I wouldn't use the word "closure" in this instance. To me closure implies that the accuracy and precision of the ending observation depends on the observations previous to it, such as in a level loop or total station closed traverse.
Getting a check shot with RTK on a known point after several observations previous to it doesn't fall into this category. It suggests that the previous shots are correct, but doesn't prove them.
Stephen
RMS is a statistic associated with the result repeating within a certain one-sigma limit. The true answer can be off by 10 feet and still have a good RMS if the answer is repeating within one sigma.
RTK Closure ? There Is No Such Thing
Positional Tolerance, readily done with a post processed static network.
However most RTK users have insufficient equipment.
Having to do a traverse kind of negates the advantage of RTK.
Paul in PA
> RMS is a statistic associated with the result repeating within a certain one-sigma limit. The true answer can be off by 10 feet and still have a good RMS if the answer is repeating within one sigma.
I suspect that manufacturers use RMS instead of 1 sigma because the accuracy of RTK results is related to the accuracy of both the base station position and the geoid model in use. It's less of an issue with horizontal; you have to have a really bad base station position in order to produce RTK positions that are in substantial disagreement with each other. But with a faulty geoid model you can easily get vertical results that don't match reality. (Bearing in mind that in both cases you don't really know what you have unless you take enough independent measurements to determine the mean. 1 sigma isn't very comforting.)
WHY TRY TO RE-INVENT THE WHEEL?
The equipment makers have already done this and reported what it is on their technical datasheets. Why not use that? You are just using the equipment to make measurements for your own purpose and not conducting rigorous scientific examination to prove or disprove manufactures' specs.
RTK has very specific uses as a measurement tool, control surveys would not be one of them. A check shot on known control only shows a correlation between that SINGLE measurement and the published values of the control measured. This cannot be construed on the whole into a statement of positional accuracy for your survey in its entirety.