Activity Feed › Discussion Forums › GNSS & Geodesy › RTK Point Movement
RTK Point Movement
Posted by jonathan50 on April 11, 2021 at 10:06 amSo I was taking RTK shots for some corners and MH. I am taking RTK readings with duration of 1-2 minutes at 1 second intervals. I am just using a rod with bubble for level. Since I am a strong coffee drinker, I can’t seem to hold the rod steady. I can see it moving as I try to place the bubble in the center of the levelling vial. But after processing the RTK shots, I see that the deviation of repeated points are in the <0.01m.
Anyone care to give their take on this?
drew-r replied 2 years, 11 months ago 17 Members · 28 Replies- 28 Replies
Correct me if I am wrong, but it doesn??t sound like you are using a bipod to hold your rod on the point. For critical shots such as the ones you are describing I always use a bipod to hold my rod steady. I also make two observations on corners and control points, 3- minutes each with the rod rotated 180?ø in between shots. Also, take care when shooting MH??s with GNSS. I shoot them with the robot or run levels through them if I need vertical.
I use the bipod even with the robot, partly because it’s usually windy, partly cause like you I’m not the most steady sometimes and I’m just OCD and want to he sure everything I set is dead on.
Under a centimeter sounds good enough to me but I can not relate to trying to hold a rod for two minutes without a bi-pod or better.
Some corners and a manhole can be very different things depending on which corners and manhole.
Skip the coffee!;)
- Posted by: @jonathan50
But after processing the RTK shots, I see that the deviation of repeated points are in the <0.01m.
Are you talking delta values (inverses) between two separate observations, or standard deviation of the weighted mean solution of the same vectors?
During an RTK observation (whether 5 seconds or 3 minutes), most field software will monitor the solution epoch-by-epoch and only allow you to store an observation if the specified number of sequential epochs meets precision tolerance. It’s performing quality control in real-time; by definition a stored solution will have a small standard deviation. 1 cm is on the tight end of things, but perhaps conditions where you are working are ideal for satellite observations.
If you are observing a point twice in a row using the same method, it’s not surprising that they are very, very close to each other since satellite conditions are effectively the same. If you come back even 20-30 minutes later and re-observe, you will likely see more than 1 cm of difference, especially in the vertical.
I too recommend a bipod…
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman Less than a cm, you are golden. For argument’s sake call your normal “miss” 6mm, it depends what you are surveying, but most tasks are accomplished with this type of error. 0.02′.
If you need more accuracy then it’s time to break out your robot or total station and survey conventionally.
For very tight control such as building corner layout I use a Trimble S-6 and a short prism pole with a bubble that’s been checked and if needed adjusted. I have not used RTK for that since a few frustrating attempts years ago. Even today with the R-10 it would be iffy and that is a very accurate RTK system.
If your hair is tingling, back off a touch on the coffee. The best solution is a bipod or steady sticks (if you have an actual issue). Check the rod bubble to ensure it isn’t tge problem.
The one to two minute occupations are not the best way to get tight control. After 20 seconds you’ve hit the point of diminishing returns on an RTK observation. You will get more reliable results by taking two 20 second observations, rotating the rod in between. If the point is critical consider a second set of shots under a different constellation, or even gathering static to process against your base.
I was expecting that the MH points would be off between RTK positions by more than 1 cm because it’s a 1.5m pole so any small movement would equate to larger movement of the receiver at the top. So are you saying that the receiver is able to get a standard deviation from all the points within the 2 minute rtk data and compute for the center point. I am pretty sure that I am not able to hold the pole vertical for more than a few seconds at a time.
As to re-observing it at another time to get different satellite positions, I think that would defeat the purpose of RTK. An RTK shot should give you a position that is within 1-2 cm anytime of the day or week or year as long as you use the same base position.
Rotating the rod only tells you if your rod is out of level. In the golden age of GPS the antenna orientation might have a slight effect on the results but probably not enough to show up in the error budget of RTK. With todays antenna design I think that issue is pretty much mute.
@john-putnam I actually asked this question with Trimble. With the R10, R10-2, R12, and R12i, there is not enough difference to be measurable within surveying workflows.
- Posted by: @jonathan50
As to re-observing it at another time to get different satellite positions, I think that would defeat the purpose of RTK. An RTK shot should give you a position that is within 1-2 cm anytime of the day or week or year as long as you use the same base position.
I’m not sure how independent measurement “defeats the purpose” of RTK. The idea is to get repeatable results, which means you need some checks on your work.
As far as the 1-2cm assertion, what site conditions are you working in? What constellations are you tracking at both base and rover? What time of day? How’s the space weather differ at those times? How long of a baseline? All those things will constantly impact your precisions.
The satellites are the true “control points” for real-time measurements, and while the time difference has rapidly shrunk over the years, the only way to get an (independent) redundant check (assuming same base point) can come from measuring from a different constellation orientation, which means waiting some time in between measurements.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman Jonathon, if your coffee shakes are random, i would expect the same positioning that a steady rod would give you with RTK. However if you only shake in one direction, then your position will be biased in that direction relative to a steady rod RTK shot.
Any advantage gained by occupying for as long as 120 seconds is negated by not using a bipod to steady the pole.
If you can’t swing a bipod for whatever reason use steady sticks.
I read RTK points twice, rotating 180?ø between. The purpose is to compensate for any rod non-plumbness, or rather to prove that the rod was plumb in the first place.
1cm -2cm accuracy might be great for section corner stones in rural surveys. Not so good for curb shots in an urban setting.
Elevations from RTK are always far less precise that the corresponding horizontal positions, and 2cm-4cm accuracy on a manhole rim is going to be too loosey-goosey for many applications. The total station is a better tool for urban design topography for this reason.
In todays RTK environment it’s not necessary to wait to get different geometry.
- Posted by: @mightymoe
In todays RTK environment it’s not necessary to wait to get different geometry.
I still haven’t seen any actual studies or white papers bearing this out, and this does not square with actual data that I have processed over the past couple of years, at least not enough for me to agree with this as a blanket statement.
While I have found that horizontal results tend to be repeatable under different geometry about 75% of the time, if we are looking for vertical repeatability, that separate observation will greatly improve the final result.
Add in variability in site conditions, whether we’re running base/rover vs. RTN, single base RTN vs multi-station, MSM vs 2-constellation, etc…
Part of it depends on your goals. For our organization at least, standard procedures need to be such that desired precision is reached at least 95% of the time. That means as independent of a check as we can get, and we still get much better results with properly weighted vectors under different conditions.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman It is certainly true that there is less to be gained. Bad initializations have become almost unknown. But to get maximum accuracy from RTK it is still recommended. Granted, in some applications “maximum accuracy” may amount to hair splitting.
The time it takes for a constellation to change has come down, but it’s still a thing. You could force a change by disabling and enabling various birds but the effort would hardly be worth it.
Putting a minimum of 20 minutes between observations made an observable difference in the quality of mean positions two years ago. That was the last time I had any use for RTK. I haven’t seen any new developments that would change that since. I would be interested to hear if that’s not the case.
Very true – it’s really just making sure that if one of those twenty occurs, we can catch it before the crews are completely done with the project, so they can go reobserve.
For us it’s the right balance between overkilling observations (reducing efficiency) and having to return to the site after the field work was officially wrapped up. A decent trade-off for in-town work. For high-stakes or out-of-town work, we will repeat observations four times throughout the day, and that has always yielded quality results.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman
Log in to reply.