GNSS RTK Accuracy

#21 · December 23, 2023, 6:29 pm

Si es asi, depende de la formacion en ese momento de los GPS

knder Surveyor

#22 · December 23, 2023, 11:35 pm

For control with RTK I collect a session, then rotate the rod 180° and collect another. I collect until the 3d coordinate quality number reaches 0.02' which is usually something under 3 minutes. Perhaps as little as 90 seconds under very good conditions. If it gets to 3 minutes I usually cut it off there, whatever the quality number says, and resolve to return under different conditions - or maybe traverse in.

An alternative to traversing is to set 3 temporary points surrounding the point of interest that are within line of sight but have better sky, hit those with RTK, resect an instrument position, and tie the point of interest. If I can establish 3 suitable points within a few feet of the point of interest I can simply tape the offsets and Starnet takes care of the rest.

#23 · February 22, 2024, 10:21 pm

We only have SurvPC 6, but we have found running GNSS Analysis and doing repeated measurements of ~15 seconds until we get ExH values of 1" to 1/10' with a Qrank of 1 produces results acceptable for our uses. This is while using Surefix/Fixed+ with a BRX7. We would confirm things with a total station initially until we grew comfortable with it. It seems to be acceptable for our cadastral purposes.

SurvPC 7 seems to have a number of features to build on it:

https://www.youtube.com/watch?v=qHbsgCjOi5Y

Not Javad levels of "whizz bang," but it looks decent enough if you want to manually tweak your results on the fly. I believe TBC allows you to tweak things after the fact, but I do not know if it can do something in the field. I imagine it, Leica, and Topcon all have something similar but cannot verify that from my personal experience.

#24 · October 27, 2024, 9:41 am

@ rover83

The second condition depends on whether the user has specified “auto tolerances” or has manually set the tolerances in the survey style. If auto tolerance is selected and the operator is using a Trimble receiver, Access knows its precision specifications and will enforce those limits on the data points being collected epoch-to-epoch. So if you are set to collect 30 epochs for an observation with an R12i, and after 10 epochs the data points begin to fall outside the default tolerances, it will throw that “poor precisions” or “position compromised” warning, and prompt you to either store (using the “good” in-tolerance positions) or remeasure.

Our crew is now using a new Carlson RT5 tablet running Carlson SurvPC 7. There are no Florida minimum technical standards for GNSS mentioned in 5J-17.052 Standards of Practice – Boundary Survey Requirements. The FDOT Surveying and Mapping Handbook, P. 111, states a horizontal positional accuracy of 2 cm ~ 0.066’ (Table C3) for the kind of work we do. When using RTK this past week our epochs were interrupted by Error: 0.105 exceeds Hrms, which is why I posting to this thread. Do you think 0.066' is a reasonable tolerance to set for Hrms?

The standard deviation settings for horizontal and vertical positioning shown on the SurvPC 7 video attached to this thread are set to H: 0.020' and V: 0.030'. What should I use?

#25 · October 27, 2024, 2:35 pm

Quote from OleManRiver on October 27, 2024, 2:35 pm
@field-dog what confidence level have you set up. 95% 68% DRMS. Those Hrms and Vrms are precisions not accuracy. The goal is to bring the precision and accuracy into one. Redundancy can aid in achieving this and also setting up a network style survey to perform a correctly weighted LSA on observations. For instance most surveying requirements are at 95% confidence. So in the field the crews data collectors are set at DRMS. Which for a quick math check whatever the data collector shows they can almost double that to get close to 95%. I qa/qc and process adjust at 95% always. Especially for control and property corners. We have rules on Topo for horizontal and vertical. I think you can set up the Carlson to do this as I was helping a guy once with Carlson and he was able to do this. Make the change from 68% to 95% in the data collector software. Most modern day receivers spec at 8mm horizontal and 16mm vertical 1 or .5 PPM depending on base and rover or network RTK. That’s at the APC. Perfect world conditions. Then we have the Bubble on the rod to deal with. The ionosphere troposphere and the DOP values etc. Multipath latency delay in corrections and a host of URE and other things that can creep into the equation. Redundancy is your biggest asset in quantifying the precisions and accuracy with GNSS . Time as well. You can repeat back to back observations very precisely but until the full change in constellation happens that can be misleading. When I perform a boundary survey with GNSS. I am always checking my observations results against the record the evidence and redundant measurements while also watching the precisions. Now I don’t get worried and try and say his or her measurements were bad mine are better when following the evidence. I want to make sure I check myself and that the math bearing and distance and called for evidence gets me to the same intent . I have to basically meet the same standards here as an ALTA requirement of .07’ + 50 PPM if I perform any boundary survey with anything other than a compass rule so which is why I utilize LSA as through redundant observations and properly weighted adjustment is the only way in which I can justify my boundary on that measurement side of the house. Look at it as accuracy of my survey but also the accuracy of performing and holding the right evidence . One thing for sure is that I would rather have bad measurements on a boundary survey vs not holding the correct evidence or making the correct call . Most LS and PLS tell me they have never been in trouble for a bad measurement but a bad decision. Not to say our measurements should not be to the best of our ability. Many surveys that I have traced in rural areas were last performed with a chain and probably not even a digital total station but a 20 second transit in many cases. Out west they performed immaculate distances over long lines. In the south and east. More briars hills and swamps made taping more difficult. If I miss some old surveyors distance by a link or two or even a foot if all evidence seems to point to that same called for monument and it seems undisturbed I am not setting another monument at a correct distance and direction but holding what I found.

@field-dog what confidence level have you set up. 95% 68% DRMS. Those Hrms and Vrms are precisions not accuracy. The goal is to bring the precision and accuracy into one. Redundancy can aid in achieving this and also setting up a network style survey to perform a correctly weighted LSA on observations. For instance most surveying requirements are at 95% confidence. So in the field the crews data collectors are set at DRMS. Which for a quick math check whatever the data collector shows they can almost double that to get close to 95%. I qa/qc and process adjust at 95% always. Especially for control and property corners. We have rules on Topo for horizontal and vertical. I think you can set up the Carlson to do this as I was helping a guy once with Carlson and he was able to do this. Make the change from 68% to 95% in the data collector software. Most modern day receivers spec at 8mm horizontal and 16mm vertical 1 or .5 PPM depending on base and rover or network RTK. That’s at the APC. Perfect world conditions. Then we have the Bubble on the rod to deal with. The ionosphere troposphere and the DOP values etc. Multipath latency delay in corrections and a host of URE and other things that can creep into the equation. Redundancy is your biggest asset in quantifying the precisions and accuracy with GNSS . Time as well. You can repeat back to back observations very precisely but until the full change in constellation happens that can be misleading. When I perform a boundary survey with GNSS. I am always checking my observations results against the record the evidence and redundant measurements while also watching the precisions. Now I don’t get worried and try and say his or her measurements were bad mine are better when following the evidence. I want to make sure I check myself and that the math bearing and distance and called for evidence gets me to the same intent . I have to basically meet the same standards here as an ALTA requirement of .07’ + 50 PPM if I perform any boundary survey with anything other than a compass rule so which is why I utilize LSA as through redundant observations and properly weighted adjustment is the only way in which I can justify my boundary on that measurement side of the house. Look at it as accuracy of my survey but also the accuracy of performing and holding the right evidence . One thing for sure is that I would rather have bad measurements on a boundary survey vs not holding the correct evidence or making the correct call . Most LS and PLS tell me they have never been in trouble for a bad measurement but a bad decision. Not to say our measurements should not be to the best of our ability. Many surveys that I have traced in rural areas were last performed with a chain and probably not even a digital total station but a 20 second transit in many cases. Out west they performed immaculate distances over long lines. In the south and east. More briars hills and swamps made taping more difficult. If I miss some old surveyors distance by a link or two or even a foot if all evidence seems to point to that same called for monument and it seems undisturbed I am not setting another monument at a correct distance and direction but holding what I found.

#26 · October 28, 2024, 10:56 am

Quote from john-hamilton on October 28, 2024, 10:56 am
I do two kinds of real time surveys...with my own local base and also using a VRS.

When I setup a local base, and keep the distances under 5 km or so, I get typically 0.01 to 0.02 m horizontal and 0.01 to 0.03 m vertical accuracies. Even in the woods, although woods shots will degrade with distance much faster than open shots. I usually setup a base within 100 meters for a totally-in-the-woods shot.

Last Thursday I was using a VRS in the midwest. Totally wide open point. Did a 120s observation. Reinitialized. Another 120s shot, 30 cm different. Reinitialized again. 45 cm different than the first one. Switched over to RTX, 180s observation. It agreed 0.01 m with the third observation. I believe there was a weather front to the west of me that was causing the problem. I had NEVER seen that bad of a repeat line using VRS. I always take two shots with a reinit in between. Some projects (highway mapping) the second shot is at a different time of day. The worst I ever see is about 5 cm in the vertical on maybe 1% to 2% of shots. Horizontal is always under 3 cm.

We recently did a project on 10 km of interstate highway to set control for mobile lidar. We ran a second order class I level line through all of the points using invar rod, and double occupied each target with RTK at different times of day. The base was setup about midway through the project. The map below shows the misclosures in the vertical of a minimally constrained level adjustment and a minimally constrained RTK adjustment. Green is under 1 cm, yellow is 1 to 2 cm, and red is 2 to 3 cm. Nothing over 3 cm. I also tied in several NSRS benchmarks that were very obstructed and they hit under 2 cm as well. However, the final product required an accuracy of 0.05 feet (<2 cm), which is why we ran levels and used a one piece invar rod. I have seen projects run by others where they used a regular level rod (i.e. not one piece invar) and the difference between the leveled heights and the RTK was up to 0.20 feet (6 cm) and there was a definite correlation with elevation, caused by the mismatch between sections of the rod. I have also seen mobile lidar projects where they just use RTK, no levels

I do two kinds of real time surveys...with my own local base and also using a VRS.

When I setup a local base, and keep the distances under 5 km or so, I get typically 0.01 to 0.02 m horizontal and 0.01 to 0.03 m vertical accuracies. Even in the woods, although woods shots will degrade with distance much faster than open shots. I usually setup a base within 100 meters for a totally-in-the-woods shot.

Last Thursday I was using a VRS in the midwest. Totally wide open point. Did a 120s observation. Reinitialized. Another 120s shot, 30 cm different. Reinitialized again. 45 cm different than the first one. Switched over to RTX, 180s observation. It agreed 0.01 m with the third observation. I believe there was a weather front to the west of me that was causing the problem. I had NEVER seen that bad of a repeat line using VRS. I always take two shots with a reinit in between. Some projects (highway mapping) the second shot is at a different time of day. The worst I ever see is about 5 cm in the vertical on maybe 1% to 2% of shots. Horizontal is always under 3 cm.

We recently did a project on 10 km of interstate highway to set control for mobile lidar. We ran a second order class I level line through all of the points using invar rod, and double occupied each target with RTK at different times of day. The base was setup about midway through the project. The map below shows the misclosures in the vertical of a minimally constrained level adjustment and a minimally constrained RTK adjustment. Green is under 1 cm, yellow is 1 to 2 cm, and red is 2 to 3 cm. Nothing over 3 cm. I also tied in several NSRS benchmarks that were very obstructed and they hit under 2 cm as well. However, the final product required an accuracy of 0.05 feet (<2 cm), which is why we ran levels and used a one piece invar rod. I have seen projects run by others where they used a regular level rod (i.e. not one piece invar) and the difference between the leveled heights and the RTK was up to 0.20 feet (6 cm) and there was a definite correlation with elevation, caused by the mismatch between sections of the rod. I have also seen mobile lidar projects where they just use RTK, no levels

#27 · October 28, 2024, 6:33 pm

John wat is the geoid accuracy in that area. That 3 cm seems high for what I have been achieving with RTK base a rover in the vertical.now my sites are not as linear either and such so a whole lot to unravel for sure. Also what receivers are you running.

#28 · October 28, 2024, 11:38 pm

this discussion is a an of worms for sure.

VRS is trash for any true accuracies. It’s quite unreliable in most scenarios needing redundancy and reliability better than 20cm horizontal and 50cm vertical. I don’t trust it. I would use it for tops shots only.

I’ve been using Trimble/Leica/Topcon all for the last 20 years. We have yet to achieve better results than post processing and RTK base/rover solution. Now, we just have more access to satellites, better algorithms in the software, less jamming or spoofing and more tools to deal with multi path.

And the same processes I learned in the 2000s still work best

1. buy reliable gear from reputable manufacturers- Trimble has been at it the longest

2. use good software - I love access, TBC, OPUS

3. utilize best practices in the field

Fixed tripods are the gold standard for base and rover

Take rapid static obs at different times of day

Test the final results against a baseline or traverse

#29 · October 29, 2024, 5:21 am

I should add a few comments about the project I mentioned. The purpose of the RTK was for horizontal only, since we ran levels for the vertical. I did 120s, rotate 180°, then another 120s. And then repeated the observation at least two hours time difference. So not the full recommended 180s.

The points were on the 10' wide outside shoulder of a busy interstate (I-70), so a lot of truck traffic on one side and hills/trees on the other. Many points were pretty wide open, but there were definitely some obstructed ones as well.