Use of RTK GPS for topographic survey, site benchmarks, structure stakeouts: acceptable accuracy?

[Spartan5]

For projects on the order of 40 acres in area, or a few miles in length, we are finding some consultants are preparing topographic surveys for site design and construction using RTK GPS. Knowing that GPS is somewhat inherently less accurate, when asked we are being told that vertical accuracy is +/- 0.15'. These are usually utility projects such as sanitary sewer and storm drain, water mains, and stormwater management facilities. As such, we can live with that deviation from the state plane elevation (for the most part) provided it is consistent across the site.

But otherwise, for measuring existing elevations of existing structures (invert, rim, weir, etc.) specifying proposed elevations for the same, and setting local benchmarks for stakeout; I have always understood that this had to be accurate to within 0.01'. In part because we specify elevations to the nearest hundredth. Because of significant figures, it is implicit that there is that level of accuracy in the measurements.

For my first ten years in the field we used a total station and ran level loops of our benchmarks. I don't think this was overkill, but that we are getting work prepared to a lower level of accuracy has me asking this question.

Provided it's not overkill, there should be a standard specification / language to ensure that we get the level of accuracy we want from our surveyors. Does anyone have a good reference to a standard?

 

[IRstuff]

If they are "your" surveyors, don't you have requirements for what they must provide? If they can't do it with RTK, there's still post processing, and if that doesn't work, then they just have to do it the old school way.

The big question is whether your ask is too stringent. Seems to me that few structures built can retain 0.12 inch accuracy, just from settling, erosion, etc.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[Spartan5]

Thanks for the reply, IR. The requirements are not as specific as we would like which is the problem I am trying to solve. It doesn't speak to accuracy and is generally understood that the survey be done in accordance with industry standards and sufficient to complete the work in the scope. That's not good enough and I want to fix it.

I agree that few things can be built to retain 0.12 inch accuracy. And I don't expect them to be. That's not particularly the problem. My rule of thumb in the civil world is that if a couple of hundredths of a foot is going to break the design, the design needs to change. But isn't the industry standard to specify hard infrastructure elevations to the hundredth of a foot? Even for things like curb and sidewalk it is critical with things like ADA requirements.

The problem is that the accuracy of the existing points, benchmarks, etc. is plus or minus 2 inches. On a 50' length of pipe at 0.50% slope that's 0.25 feet of fall. If the existing structure elevation is 0.15 feet higher than it really is due to GPS measurement error on the plan, then that pipe could could go in with 0.10 feet of fall at 0.20% if they set the new structure based on a local benchmark elevation. (yes I know they have some play in the structures but that can get chewed up in other places).

My biggest problem is with the site benchmarks not being accurate. If there is a weir for instance that needs to be installed at elevation 300.62. The contractor builds it based on one benchmark to elevation 300.60 (there was a little bit of that settlement that you talked about). The inspector measures it at 300.43 using their level and a different benchmark. What's the elevation of the structure? Was it installed wrong? What goes on the as-built?

 

[IRstuff]

0.01 ft over a "few," say 3 miles is 0.13 arcseconds, which is better accuracy than high-end total stations can provide. A typical construction grade station would have something like a 5 arcsecond accuracy, which, over 3 miles is 0.38 ft.

Seems to me there's a disconnect somewhere in your numbers.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[Spartan5]

Not sure what you're driving at, IR. Out of curiosity, what's your surveying background?

In my past life, when surveying a project of that length using a total station, we weren't taking shots over anywhere near a distance of 3 miles (though we did stretch it a bit on a lake project when we needed to punch some control into out of the way places... I digress). When surveying that length, we would have multiple setups, "traverse points" as we moved from one end to the other (line of sight and all). Setting a traverse point was different than collecting or setting out a normal point. It required for instance, that multiple shots be collected, including flipping the scope 180 degrees and rotating 180 degrees to cancel out error in the instrument. Generally speaking, any benchmarks we set out from a traverse point would be within 1,000 feet of the instrument.

That aside, on a 20 acre site for instance we'd be within 700' of any point from a single setup. And oh for the sites where everything could be hit from a single setup. I digress again. All that said, it sounds like 5 arcsecond accuracy gets you in the range of +/- 0.01 feet when shooting a typical 20 acre site at distances of up to several hundred feet. No?

And, as I mentioned, if we were setting out benchmarks for which having elevation on a constant datum across the project was critical, we'd run a level loop.

I hope that helps clear up the disconnect.

 

[IRstuff]

Partly; I was using the 3mi as the best case tolerance stackup, as a single measurement. In a closure loop, wouldn't be multiple stackups, so the end-to-end tolerance wouldn't be 0.01 ft?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[Spartan5]

The end to end tolerance would not be 0.01 feet on a 3 mile linear survey. We'd be tickled in the field if it closed within a few hundredths. That error will be lost in the wash when we get from one end to the other. Besides, when I am set up on my local control, and taking shots on the infrastructure within a thousand feet of my setup, I know that my elevations are within my tolerance (0.01 feet) relative to the local benchmarks that the contractor will use to construct everything. That is what is most important.

It's not the end to end accuracy that creates the problem though. Look at my examples again. If we have two benchmarks on the site within 300 feet of each other that are off by 0.15 feet we have a problem. When it comes to setting structures, which elevation is correct?

As I understand it, that RTK GPS variance exists regardless of if the points are 3 miles apart or 30 feet apart. So if we had a perfectly level stretch of road at an elevation of 300.00 feet and I set 3 equally spaced benchmarks 600 feet apart; they could be 300.15 feet, 300.00 feet, or 299.85 feet. Likewise, if the same point is checked a few days later, could it not vary by +/- 0.15 feet as well? That's 4 inches of potential vertical error on every shot.

I wouldn't have a problem with this if the standard was for proposed elevations on plans for inverts, tops of structures, etc. to be designated to the nearest tenth of a foot. Likewise the benchmarks. But I always thought significant figures meant something on plans and in computations, ESPECIALLY in engineering. Just like I shouldn't report measurements made with my engineering scale to the thousandth of an inch because it's only accurate to the hundredth; we (engineers/surveyors) shouldn't be reporting elevations that are accurate to the nearest hundredth when they are only accurate to the tenth.

Don't get me wrong. For 98.6753% of the shots on a typical topographic survey, I'm not going to lose any sleep over a couple of tenths. But for things we need to tie into and/or have fabricated and/or affect the function of whatever it is we are building; significant figures, accuracy, and repeatability are important to the success of the projects.

 

[elev8848]

If you have not done so already, you can also look into any published standards and order of accuracy requirements like

http://www.dot.ca.gov/hq/row/landsurveys/SurveysManual/05_Surveys.pdf

Thanks