I'm pretty much on board with your premise, which is don't use RTK when the distances are short. Ignoring ALTA surveys, which I would only use RTK to gather the topo, I don't use RTK for lot surveys or surveys where the distances are close. In my mind, under average conditions, a RTK GPS shot could have an error of 0.04'. I don't have any data to back this up, just my experience. So say I RTK two property corners that are 20 feet apart, someone could measure that distance better with a 25' carpernters tape than I can with a $20,000 GPS system. Why would I use GPS for this when I can use a total station that does a better job and really doesn't take anymore time? Thats just my line of thinking, bet it right or wrong.
David Livingstone, post: 383818, member: 431 wrote: I'm pretty much on board with your premise, which is don't use RTK when the distances are short. Ignoring ALTA surveys, which I would only use RTK to gather the topo, I don't use RTK for lot surveys or surveys where the distances are close. In my mind, under average conditions, a RTK GPS shot could have an error of 0.04'. I don't have any data to back this up, just my experience. So say I RTK two property corners that are 20 feet apart, someone could measure that distance better with a 25' carpernters tape than I can with a $20,000 GPS system. Why would I use GPS for this when I can use a total station that does a better job and really doesn't take anymore time? Thats just my line of thinking, bet it right or wrong.
To me, that makes perfect sense. The real action is in figuring out how combine GNSS vectors and conventional measurements to get realistic and reliable results, not in trying to RTK everything.
No need to bite
Kent McMillan, post: 383813, member: 3 wrote: Okay, so far you've determined that your RTK works well in wide open settings with long sessions and does not work well in environments with high multipath and rates of cycle slips. Was either a surprise?
Kent, this is with my OLD gear. this is where I did MUCH testing. My new gear is MUCH better.
N
And, I'm not trying to meet ALTA with it. I can meet ALTA in the field. With the old gear.
Nate The Surveyor, post: 383843, member: 291 wrote: Kent, this is with my OLD gear. this is where I did MUCH testing. My new gear is MUCH better.
Okay, what is Shawn doing with your OLD gear and why is he passing it off as something else?
Kent that's rude. When you can't win an argument, you make one up you can win, and put words in other's mouths. Come off of that.
Nate The Surveyor, post: 383848, member: 291 wrote: Kent that's rude. When you can't win an argument, you make one up you can win, and put words in other's mouths.
Uh, actually you were quoting my reply to Shawn dealing with the performance of HIS new Javad unit and wanted me to know that it was your OLD gear. That naturally raises questions and would lead a person to wonder how Shawn got ahold your old gear to test.
Kent, you are off the deep end.
N
Nate The Surveyor, post: 383851, member: 291 wrote: .. stuff ..
Well, here's a link to your post. Note the post you quoted is what most folks, myself included, would assume that you thought you were replying to. I'm not making this up.
ofFor me In Starnet,
After I include GPS (RTK) measurements, terrestrial measurements, including redundancy, radials and or network measurements, I use the inline command .ptol /every
This will provide a positional tolerance between 2 points of every connection whether the measurement was correlated or taken radially (RTK). The Starnet listing file will show whether measurements between pairs have met the A.L.T.A. positional tolerance standards of 0.07' + 50ppm (set in special tab of starnet).
Teddy
Trophy for Teddy... the key is positional tolerance between 2 points, so obviously the measured points need to be closer than 0.07'
I thought many moon ago when I was in school there was a standard requiring lines under 300ft be measured at night with a calibrated steel tape.
I may look for that tomorrow at the office. This would have been during the total station hey day/infancy of GPS.
Steve
Peter Ehlert, post: 383918, member: 60 wrote: the key is positional tolerance between 2 points, so obviously the measured points need to be closer than 0.07'
What?
Shawn Billings, post: 383770, member: 6521 wrote: For 2D, the multiplier from 1 sigma to 2 sigma is about 1.6 not 2.
You lost me there. To get from 1 sigma to 2 sigma you multiply by 2. What we're after is the sigma multiplier from one confidence (e.g., rms) to another (95%). Maybe you were thinking of the 95% confidence for 1-dimensional distributions, which is 1.96 sigma. But that isn't applicable here.
For a bivariate Gaussian normal distribution with equal sigma in any direction (east, north, or any radial), 95% confidence is 2.445 sigma. The sigma or rms for the difference of two coordinates with equal individual sigma is sqrt(2)=1.414 sigma of one.
So the allowable rms in easting and also in northing for each of the two points is 2 cm/(2.445*1.414) = 0.58 cm, in order to end up with 2.0 cm at 95% confidence for the difference.
Bill93, post: 383991, member: 87 wrote: To get from 1 sigma to 2 sigma you multiply by 2. What we're after is the sigma multiplier from one confidence (e.g., rms) to another (95%). Maybe you were thinking of the 95% confidence for 1-dimensional distributions, which is 1.96 sigma. But that isn't applicable here.
For a bivariate Gaussian normal distribution with equal sigma in any direction (east, north, or any radial), 95% confidence is 2.445 sigma. The sigma or rms for the difference of two coordinates with equal individual sigma is sqrt(2)=1.414 sigma of one.
So the allowable rms in easting and also in northing for each of the two points is 2 cm/(2.445*1.414) = 0.58 cm, in order to end up with 2.0 cm at 95% confidence for the difference.
Except while it's true that an RMS error of 5.8mm in Easting and Northing for each point positioned does give a 95%-confidence relative error ellipse of 2cm between the two points, the RMS error of the coordinates is not the radial distance RMS of each point. That DRMS value is what some RTK manufacturers evidently claim to use in specifying positional accuracy (without explicitly saying so, of course).
So given that the Eastings and Northings are both normally distributed with a standard error of 5.8mm, the uncertainty of the radial distance would be 7.1mm. DRMS, right?
That uncertainty for each point in turn corresponds with an uncertainty of 1.42cm 2DRMS for each and a relative uncertainty of 2cm between the two points.
Kent McMillan, post: 383995, member: 3 wrote: So given that the Eastings and Northings are both normally distributed with a standard error of 5.8mm, the uncertainty of the radial distance would be 7.1mm. DRMS, right?
No, if the coordinate errors are equal and independent (ellipse becomes circular) then Easting is radial, 5.8 rms. Northing is radial, 5.8 rms. Any radial is 5.8 rms. Northeast error is radial, 5.8 rms. The ellipse becomes circular.
I'm interpreting the requirement of 2 cm as not just the error on the distance between points, but the error you'd find if you translated one of the points by your angle and distance measurements to attempt to lay them on top of each other, and looking at how far they mismatched. Isn't that what the spec wants?
What does D in DRMS mean?
Bill93, post: 384037, member: 87 wrote: No, if the coordinate errors are equal and independent (ellipse becomes circular) then Easting is radial, 5.8 rms. Northing is radial, 5.8 rms. Any radial is 5.8 rms. Northeast error is radial, 5.8 rms. The ellipse becomes circular.
Except the distance is the root sum of squares of the N and E components and that is a different case. It's true that one could transform the situation into an orthogonal coordinate system with axes in any orientation and the RMS errors of the X and Y components would still be 5.8mm.
I'm interpreting the requirement of 2 cm as not just the error on the distance between points, but the error you'd find if you translated one of the points by your angle and distance measurements to attempt to lay them on top of each other, and looking at how far they mismatched. Isn't that what the spec wants?
Yes, that's right. the maximum allowable relative positional uncertainty of 2cm is the maximum length of the 95%-confidence relative error ellipse.
What does D in DRMS mean?
I use it to mean Distance Root Mean Squared error, referring to the length of the vector from the actual position to the estimated position, rather than the RMS errors of components of the coordinates.
And of course what I meant was:
Kent McMillan, post: 384052, member: 3 wrote:
Yes, that's right. the maximum allowable relative positional uncertainty of 2cm is the maximum length of the semi-major axis of the 95%-confidence relative error ellipse.
Kent McMillan, post: 384052, member: 3 wrote: distance is the root sum of squares of the N and E components
In a specific instance of one actual error thats true. In the probability though the rms of NE error is the same as E or N rms error because often the N error and E error are not large at the same time. That's what a circular "ellipse" means.
