Ground or Grid?

[E_Page]

First thing you should do is go back and complete your field survey. It sounds to me like your collected a bunch of topo from your temp control point. Some of that topo happened to include shots on some monuments that will be significant to resolving your boundary. Get some redundancy with your base set at another station and then collect shots on all your monuments again.

I've followed behind several supposed boundary and control surveys performed like this and have found that it is not uncommon to find significant discrepancies by measuring directly between the points collected like this.

Without knowing how you set up your collection, your question can't be answered. Did you start the base telling it you were "here" and assign it a coordinate (like N=10,000, E=10,000), or did you set it up to work in CCS?

[JeffroH]

Perhaps your larger distance is a slope distance and not a horizontal distance?

[Gromatici]

I just figured out that for a Trimble R8 when you set it to "no projection" "no datum" it measures in "ground" and "true" bearings (as calculated at your LAT/ LONG and using the theta angle for mapping and scale factor combined with your elevation factor to scale up to "ground" coordinates)

http://www.californiasurveyors.org/clsa ... ue+azimuth


Read your manual or talk to your dealer or support tech about what the software is doing. Without knowing how you set up a "job" in your DC, we can't tell you much. Sounds like your in ground, but NGS monuments aren't always perfect nor your measurements. I like to have two bases, or set up on two different points and re-shoot some in common before I feel I performed a professional survey verses a "push-button" survey that you see so much of these days.

Maybe you're in CCS83, but the two NGS monuments are in two different epochs. Here is SB, that can make a difference of 1.5'!

[Brian Christensen]

The problem with RTK is that when a vector is measured(one base station), it is akin to a side shot, no way to prove the position of the point at the end of that vector, regardless of the residuals and DOP displayed on your survey controller.

[Gromatici]

If you shoot monuments and control using a 4 x 4 method (4 shots using four different integers) then I'd say you can be as sure it's right as doubling an angle. When you use this method you loose link and regain it for every one shot on the point plus rotating the rod to average the error in a 40' bubble (which is about 0.04' at 2M).

The odds that after "breaking" the radio link and resolving the integers (the factor by which the equation is solved for time, N, E, Z between the base station and the rover) and being in the exact same wrong spot are so great that it probably will happen once in a 10,000 years. As a result we usually shoot all control and monuments 3x with RTK as a way to check the results, plus it adds redundancies in our least squares adjustment.

Then of course, you could hit the point again from another base station.

A typical job we do we can see an error elipse of 0.03' or less. We've gone back after a year and check within that, but occasionally will be off by 0.04' or 0.05' in a mile. Vertical can be a lot looser, and sometime we'll see .08' but the vertical component is always harder to get a good read on so we will adjust in the Total Station measurement with greater weight.

[dmi]

No you cannot do what you are describing. You can if you insist, but it is pretty much WRONG. I suggest you do some reading, perhaps the Caltrans manual on GPS.

[Jim Frame]

"The odds that after "breaking" the radio link and resolving the integers (the factor by which the equation is solved for time, N, E, Z between the base station and the rover) and being in the exact same wrong spot are so great that it probably will happen once in a 10,000 years."

Those odds may depend a lot upon how long you wait between shots. If your reinits are done without allowing sufficient time for the SV constellation to change significantly (say, 30 minutes or more), I don't think you're gaining much in the way of redundancy. Multiple reinits in quick succession might expose a bad integer fix, but they won't expose a weak constellation or multipath condition.

.

[7702]

["plus rotating the rod to average the error in a 40' bubble (which is about 0.04' at 2M)."]


Isn't it easier to check and adjust your level bubbles before you start surveying? That's one less thing to worry about.

[Jim Frame]

The problem with a 40' bubble is that even when it's properly adjusted and visually centered you can still be a couple of hundredths out of plumb at a 5-foot HR. I swapped out the 40' bubble on my prism rod for a 10' bubble about 10 years ago, and it made a noticeable difference in the standard errors of my control points. I decided that a 40' bubble is okay for topo but not suitable for control.

.

[Gromatici]

I have the crews use a 10' bubble too, but sometime they forget it. If you do the math a calibrated 40' rod will have a probable error of .045' at 2 meters. (Tangent of 2M x 0.04'00")

Wooley: We're talking about RTK not static so I'm discussing how I eliminate posible error since RTK does a least-squares solution on the fly and there isn't a real way to check it in the post-processing stage like a static control job.

For large control surveys or the mountainous tracts we use static with two bases and the post-processing will catch rod busts and "noise" or muli-path. For the smaller jobs (3 miles or less) we use RTK because it's faster and thus we're more competitive. We do a lot of aerial mapping, topographic mapping and large boundaries so RTK makes it work a lot faster than Static, but sometime we have to mix them in together.

In order to check our work I have the crew shoot monuments and control 3x (modified 4x4 method). It's not from 3 different control points, but it's 3 different shots, breaking the radio link all three times. The idea is that if there is multi-path or radio interference you'll get three different answers and the crew will know there is a problem and tie it out with the gun instead. It's true that it's most likely the same satellite constellations, but we're trying to narrow done Radio interference, multi-path or an out-of-plumb rod and not sub-millimeter constellation differences for a GPS control net.

I do however, when the job needs more than one "base" like to re-shoot some of the previous set control to complete a sound network.

It's all about redundancy and checks. Otherwise your simply trusting that the rod is plumb and that there was no radio interference and multi-path.

For instance, I had a crew that had an out of plumb rod and when I did my starnet the 3 shots had an error ellipse of .2'. I wouldn't have know that if they were not following the method described above.

I find the most common source of errors is out of plumb rods and instruments needing calibration. The other source is lack of knowledge that some surveyors have about the instruments themselves. A 3" gun is only 3" at 2 sigma, which means that you would have to double the angle to get it within 3" of true. If you only taking one shot to something the manufacturer doesn't have that accuracy stated. Then there is the bubble issue on a rod. If you have a 10' bubble there it means it's "accurate" within 0.01' at 6' high.

If your trying to shoot grid lines with a 6' rod your only going to be accurate within 0.01' according to the specs. This is why I use shorter rods and usually set up on a line and sight down it, rather that stake grid lines out radially. +- 3" and a 10' bubble can equal 0.02' on some jobs.

Then there's random errors. Your crew has some drift in the gun, they bump it, or the rod slips. That's why with all our technology you can retrace a guy who did a survey last year and still only check it within 0.08' All the technology in the world won't solve the ultimate problem: end users.

For most jobs I do I have two crews running, doing different tasks. GPS static for the skeleton control framework, RTK control and monument recovery and another crew doing topo or staking. To keep is all from deteriorating I use Star*Net check and see if everyone is keeping it tight and I make sure there are redundancies in the network, otherwise it's just a side-shot like someone said.

[Jim Frame]

"It's true that it's most likely the same satellite constellations, but we're trying to narrow done Radio interference, multi-path or an out-of-plumb rod "

Unless the constellation changes significantly between shots, you're not likely to pick up the mulitpath error.

Dumping the antenna and reinitializing is a good way to check your integer fix, but multiple RTK shots in rapid succession still equals a single sideshot in my book.

[Gromatici]

Experience tells me differently. Usually I'll get two different answers if I'm pushing the envelope (under trees, next to a building, power lines) after dumping the antenna and reinitialize it. We also hired an outside consultant who does this stuff worldwide and even helped write the recently updated code in CA on using CCS83 who told me this method.

Of course there is the 1/1,000,000 chance it could end up with 0.04'.

[7702]

I've had surveyors tell me that they don't like Leica rods because it's harder to hold them steady. It's because they have a more precise level bubble, not because the rod's defective. Pretty amusing.

Also, direct sun will cause the level bubble to drift considerably as well. I always try to shade the bubble when plumbing up the base station.

[Steve Martin]

multipath is time variable and has a sinusoidal effect on the inferred range to the satellite. Unlike fixing the wrong ambiguity, residual effects that make it past the signal processor have to be averaged out over time, i.e. a static gps survey. When you are doing a RTK survey, you automatically have a larger effor budget as compared to static gps because you cannot average out the residual multipath effect in the signal.

Rodger Frank had a good presentation about survey accuracies at the CLSA conference a few years ago. He talked about transit and chain accuracies and how things got much better with the advent of the EDM which evolved into total stations. Then GPS came out and accuracies over greater distances were incredible. Next came RTK and accuracies took a step backwards. Granted RTK today is better than it was 10-15 years ago, but it will never quite get to what can be acheived with static GPS.

Back in the early days of GPS surveying, we were taught to always orient the antennas to North. This was before the days of antenna models and I am sure the new models of antennas are much better than geodetic models of old, but the point was that antenna phase center variations would be the same on like antennas oriented in the same direction and this small error would then cancel out. I guess it just goes againt my geodetic survey training to rotate an antenna during a survey, but I suppose that any phase center error is overshadowed by the level bubble error and operator setup errors using a stakeout rod.

Now that I am going down memory lane, a story comes to mind: We were testing out a Psuedo Kinematic GPS survey procedure years ago (look that up in the history books) and the Party Chief was using a new out-of-the-box pole and bi-pod rig. At the end of his route, when he check back into known control, he noticed that the pole was out of plumb even though he carefully centered the bubble every time. He pulled out his plumbbob and sting and estimated that it was out by 0.1' North and 0.1' East. Since at each setup the antenna was oriented North and the rig remained fixed he hoped we could salvage his half day of survey. A comparison of the PK position against the static GPS\optical plumment tribrach indicated a 0.11' North and 0.09' East offset and we were able to enter that antenna offset into the processing software to determine vectors to the station marks.

Moral of the story: always calibrate your rod and bubble before each survey, even if it is a new pole or you just did it last week.

[7702]

["If you do the math a calibrated 40' rod will have a probable error of .045' at 2 meters. (Tangent of 2M x 0.04'00")"]

I don't understand the math on the above statement. 40 minutes of angle in 2 meters results in an offset of roughly 0.076' according to my calculations. Is that what you were trying to determine, or is the "probable error" something else?

However, if one fails to convert 40 minutes to decimal degrees before running the calcs, the result is real close to the 0.045' stated.....????

On another note, it's interesting that technology as complex as GPS (sinusoidal effects, integers, precise timing, satellites, etc.) can be totally derailed by concepts as simple as plumbing a rod.

[Gromatici]

Maybe I'm wrong but the point of this discussion started as a grid-ground and somehow turned into a calculation lesson. I did forget to convert the 40' to decimal and got .045' instead of the .08' (rounded). I guess my point has been lost that the 0.5' could be the difference in grid v. ground or could be measurement protocols and errors caused by RTK itself or rods and what we can do as professionals to eliminate them and discover them. He could have an inch on one point, another inch on the other plus any other issues we've discussed.

50,000 on a new GPS RTK system:

Only good within 2" because of the rod

Priceless.

[goodgps]

Agent 007,
Dear Mr Bond,
If you're using a Trimble System, on a "here" base, you are measuring in "Ground" distances at a horizontal plane, IF you've single point calibrated holding a scale factor of 1. and have chosen a base elevation relative to your survey area. (we use 100 in the mid-valley)
The measured bearings relate close to geodetic at the calibration point.

Although redundancy is a prudent part of any survey method, most likely your survey (collection of points) is pretty good.

We steady stick our poles for each shot using two lath or the bi-pod.

I am only familiar with the Trimble systems. Perhaps Leica or other systems work only in the geodetic with distances scaled to the sphere. check the specs of your particular device or perhaps ask your dealer / representative.

Good luck on the LS when you take it :-)