Settings and Best Practices for Optimizing a Elevation Survey

[Brett Chappell PLS]

Dear Javad Community,
I am performing an elevation survey of 66 monitoring points at a closed garbage site.
The site is 0.8 miles square with 30' of fill above surrounding roadways. This is a 2 part survey. Part 1 will survey the points with digital level equipment. Part 2 will survey the points using GPS, in my case using a Triumph LS with T2 Base. I plan on performing Part 1 first and then localize on a few exterior points and an interior point for Part 2.

My client is interested in the differences between the two surveys to determine which technique to use going forward.

The accuracy requirements have not been spelled out so I consider this a test to optimize RTK
results for a 3-5 minute session. That would make it a 1.5-2 day RTK vs. 5 Day Digital Level survey.

My questions:
1. I would like to know the best settings and occupation time using the Triumph LS.
2. Are there post processing techniques that can optimize the RTK results.
3. Should I use the US satellites only or multiple systems?

Please add any additional suggestions.

Thank you for your time.

 

[Matt Johnson]

Hi Brett, in an area open to the sky you should get very good RTK results without long observations. I had previously posted some test reports of the precision of the LS and T2 system in https://support.javad.com/index.php?threads/occupation-time-and-precision.458/#post-4168. I would expect your results always be within 0.10' vertically even when using short observations of 10 epochs. It would be best to use both GPS and GLONASS satellites. Using a level for landfill elevations seems like overkill to me with the capabilities of RTK. There aren't any post processing techniques that would improve your results and for the best results you want to be using a base station near your rover (short baseline length) rather than a RTN.

 

[Shawn Billings]

You will increase your precision with longer occupation times but perhaps not absolute accuracy. In some recent tests I've done regarding occupation times at ten miles from the base, I've observed that the horizontal coordinates converge in about 4 minutes at that range. Vertical coordinates converge in about eight. Staying longer won't likely improve your position with RTK. I've noticed that just because the solution converges to 0.01 foot doesn't mean the coordinates are accurate to 0.01 foot. This takes us to the age old comparison of accuracy and precision.

To your question, I don't know that 8 minutes will be required at those ranges for the vertical to converge. Beyond this convergence time, additional time on site will provide diminished returns. I'm not speaking authoritatively here, but theoretically. I'd set the receiver to stop by button. I'd watch the coordinates, particularly elevation, as the position averaged (bottom of collection screen). I'd wait for two or three minutes, then I'd see if the elevation changes over the next minute. If the average changes appreciably in the next minute I'd continue for another minute. Eventually the elevation will converge. It won't change with additional time. Once I observed the convergence I'd stop and store.

Beyond this convergence only a repeat under a different constellation will improve your accuracy. The average tool in Cogo>Tools is excellent for combining two or more positions into one, including statistics for accuracy estimates.

 

[Matt Johnson]

I still feel that longer RTK occupation times improves both precision and accuracy. I will try to do some more testing to prove this. I can set the base and rover up at the same height next to each other and repeat the test I linked to. This way we can see both the vertical accuracy and vertical precision between the two receivers.

 

[Brett Chappell PLS]

Hi Matt,
The site is going to be developed for commercial purposes and the site is fairly open. The GPS vs. digital will decide the cost vs. accuracy questions. I think my max occupation time would be 5 minutes. Thanks for the link.

Hi Shawn,
Thanks for the technique. It makes sense and I will probably use your suggested method . The max single baseline will be +/- .5 miles.
I might run two different rovers at opposite ends and create different constellations.

 

[Matt Johnson]

I think I understand now, the points are to monitor if there is any vertical movement over time? If so that makes more sense why they would need to be very accurate.

 

[Shawn Billings]

I don't mean to suggest that longer occupation times won't improve accuracy only that the convergence to the millimeter isn't indicative of accuracy to the millimeter. Repeat observation will help improve accuracy but even then, systematic errors common to both observations may go undetected.

 

[Matt Johnson]

I would agree with that. The accuracy should end up equaling the sum of the RTK precision plus any systematic errors.