Please enlighten me on this fact. I am processing some RTK points tjat were also checked with a total station in the course of the topography survey. The RTK points were done when the client wanted to get a rush preliminary data result. Now that we are doing the topography survey using total stations for areas that are near tall trees and buildings, I am getting variations in the 0.10 to 0.30 meters. Not acceptable for our acuracy requirements. Vertical datum in total station control points were run via level lines including the GPS base that was used in the RTK survey. The base was located less than 50 m from longest RTK point.
Just wanted to know if this variation is normal? All RTK points recorded have Fix confirmation.
No, that is not normal for RTK. You should be 1-3cm horizontal accuracy with a Fixed position.
I suspect you were using RTK in difficult conditions such as around trees and buildings. This can hurt RTK accuracy.
However, the RTK receiver should not "lie" to you. It should not tell you fixed when the accuracy is 10-30cm, because it is not.
Is there a "confidence" setting in the receiver that you can set higher (for example, to 99%)?
> Please enlighten me on this fact. I am processing some RTK points tjat were also checked with a total station in the course of the topography survey. The RTK points were done when the client wanted to get a rush preliminary data result. Now that we are doing the topography survey using total stations for areas that are near tall trees and buildings, I am getting variations in the 0.10 to 0.30 meters. Not acceptable for our acuracy requirements. Vertical datum in total station control points were run via level lines including the GPS base that was used in the RTK survey. The base was located less than 50 m from longest RTK point.
> Just wanted to know if this variation is normal? All RTK points recorded have Fix confirmation.
RTK 1 cm precision horizontal and about 1 to 3 cms vertical.
Could have incorrect projection or have not converted coordinates from grid to plane to use in total station. Likely inexperienced in knowledge of understanding and use of GPS, as RTK is a piece of cake.
RADU
This type of error is to be expected if you are near tall obstructions, but it's unpredictable. Sometimes the answers are true, other times they can be rubbish. It all depends on the obstructions and the satellite geometry at the time.
If your controller displays the satellite positions then you can get a good idea of what might happen. The controller should also give you a guide in terms of the accuracy it thinks it is getting. My experience is that when it says up to 15mm. then the readings are OK, above 15mm. assume the accuracy is 2 to 3 times worse than it states. Once it gets above 50mm. then the point can be anywhere.
I've done a number of experiments using a GPS on top of a total station prism, reading both at the same time and the results bear out the above.
If you can set your point recording to take in several epochs (say 5 readings over 5 seconds) and average the results you can inspect the variation and decide on site if the point might be OK.
If you have to pick up essential detail in very poor locations for GPS consider doing it using offset lines - you take two points in the open to form a line pointing into the point you want, taping from the nearest GPS point into the detail point. A last resort if you don't have a total station available.
DEPENDS...
Sometimes a static "fixed" can be wrong also.
When in doubt, we used to practice dumping the antenna and reacquiring the satellites and getting a fresh fix when doing RTK photo control. I expect for most topos that would be too much. But now you could have a mess. Of the millions of RTK shots taken each year I wonder how many have serious errors. It's like one of those things some "surveyors" do not really want to know. In the early days of GPS the watchword was always redundancy. Nowadays expediency has cut back on that. It's a bit like building so much confidence in your total station that you no longer double angles or close traverses. My rule of thumb was not to trust any RTK solution that did not fix immediately. If the receiver has to sit and cook for longer than a few seconds you should verify. Of course that's a photo control station not a ground shot in a topo.
Anytime the GPS signals are interrupted there is a chance the wrong integer will be fixed. That is more likely w a weaker constellation. (IMHO)
Should make for interesting discussion today. I'm wondering what Paul will say:-)
You've been listening to the salesman, thinking fixed means a good shot. No it doesn't mean that, it just means it is probably a good shot. RTK systems have improved greatly over the years and have better error trapping but still you can have fixed shots that are off the .1-. 3m, which sounds about right for a "bad" fix. Back in the early days a bad fix happened every day, now it's rare, but you need to be careful and check, check, check everything.
thanks for the heads up on coordinate transformation, checked the settings and all are good.
i have used RTK in bathymetry survey. You don't have such errors because you can readily check the rtk shots because the ellp hgt is almost the same or changes slowly as the tide changes. But in topography survey you can't really check for hor or vert accuracy on the fly, you just rely on the fix code, accuracy in hor and ver according to the data controller is less than 20mm.
we also checked the total station as the error may be with the instrument but direct reverse readings are OK.
Now if you combine rtk and total stations you get variations that are visible at the points where these 2 datasets overlap.
bad part is the client already got the rtk data so have to explain this discrepancy. this is why i am against giving partial data:-$
> No, that is not normal for RTK. You should be 1-3cm horizontal accuracy with a Fixed position.
>
> I suspect you were using RTK in difficult conditions such as around trees and buildings. This can hurt RTK accuracy.
>
> However, the RTK receiver should not "lie" to you. It should not tell you fixed when the accuracy is 10-30cm, because it is not.
>
> Is there a "confidence" setting in the receiver that you can set higher (for example, to 99%)?
>
> > Please enlighten me on this fact. I am processing some RTK points tjat were also checked with a total station in the course of the topography survey. The RTK points were done when the client wanted to get a rush preliminary data result. Now that we are doing the topography survey using total stations for areas that are near tall trees and buildings, I am getting variations in the 0.10 to 0.30 meters. Not acceptable for our acuracy requirements. Vertical datum in total station control points were run via level lines including the GPS base that was used in the RTK survey. The base was located less than 50 m from longest RTK point.
> > Just wanted to know if this variation is normal? All RTK points recorded have Fix confirmation.
my understanding of the situation is this;
most, if not all receivers display coord quality as 1 sigma certainty ~66% error ellipse. to estimate 1 sigma vs. 2 sigma ~95% erroe ellipse, multiply the cq's by a factor of 1.5 usually works
also, i have gotten 'fixed' vectors that sometimes are not accurate, or precise. when in doubt, say getting aggressive with horizon limitations, measure multiple times as one check. as another, store your data as static or kinematic track, then post process. most of the time you will get a better vector, or be able to sort out the good from the not so good (provided you have multiple measurements)
cheers
The answer is it depends... on many factors. I would direct you to the following:
Kevin
Fact is, you can have lots of sats up, and get a bad or poor shot, then, later, say 1 hr later, get a GOOD shot on the SAME point. Now, you have to have a tie breaker on it.
N
My procedure for RTK topo is this:
If it's rough stuff and I don't care a lot about the accuracy-then if it's fixed I'm ok.
If it's something I do want to be done tight (which is most everything) then I survey the topo, checking in before and after set-ups. Then I change HI's, and rover height and go check some locations-not everything but some random shots across the site. It can be a pain, but.....
I do it more to check my fixes than anything else. I used to be more anal about it making sure I had a check shot for each fix but now I just check random shots.
I suggest holding your head high and presenting the new data as your professional product.
Prepare a plot of the rush preliminary data showing a cloud around the data which should be updated by the more accurate data. Just put together a quick statement about RTK close to trees, etc.
We can always say we hate to provide rush, preliminary data, but sometimes, as in this case, it is necessary and most of it was fine with a little cleanup in those areas close to the trees….
Everyone will understand and compliment you for a job well done. Be proud to provide what the client needed with both sets of data. The main thing is for them to be able to revise their preliminary data as quick as they can and not spend too much time on it. Maybe you can think of a way to make that easy while incorporating some Quality Control. They will like that too!
> Please enlighten me on this fact. I am processing some RTK points tjat were also checked with a total station in the course of the topography survey. The RTK points were done when the client wanted to get a rush preliminary data result. Now that we are doing the topography survey using total stations for areas that are near tall trees and buildings, I am getting variations in the 0.10 to 0.30 meters. Not acceptable for our acuracy requirements. Vertical datum in total station control points were run via level lines including the GPS base that was used in the RTK survey. The base was located less than 50 m from longest RTK point.
> Just wanted to know if this variation is normal? All RTK points recorded have Fix confirmation.
There are a host of reasons why you would see this discrepancy. In a perfect world, an RTK shot is centimeter accuracy at one sigma. It's not out of the realm of possibility to see 0.06' (which is one sigma both ways) or up to 0.12' (two sigmas) so all of your points, if done properly should be within 0.12' of each other 95% of the time, imperically speaking. Now, just like any other measurements, if you want them tighter, you need to observe them under different constellations and at different times. A minimum of twice but three times is optimal. For Ultra-Tight Control, RTK is not the weapon of choice. Static is.
As I mentioned, a perfect world. Well, it's not perfect so factor in a tree nearby masking a satellite, pole being out of level, not waiting for a sufficient amount of epochs to be recorded, or just a damn gremlin in the receiver and it's not hard to have RTK points that get WAY out of whack. That's why you check check and recheck.
Redundancy is the key here.
:good:
I agree, however, I wouldn't place any guess at an explanation such as "trees too close" on any file or document, all the client needs to know is that preliminary means preliminary and it was not intended to be a final, reliable product.
It only takes one bad initialization on an important point to ruin your day. No such thing as every shot being good, maybe 96%, but not all. The nature of RTK. Some bad shots won't matter as in topo, but be careful with boundary points if you are using RTK.
Since the variation occurred near Tall trees or buildings, it could be the result of Multipath. When near buildings, the satellite radio wave is directly received by the antenna. But the radio wave can also bounce off the nearby building to the antenna. Both satellite signals (direct and bounced) reach the antenna and distort the readings, thus recording observations that deviate from the normal specs for RTK.
RTK is a tool, and should not be used for everything. Total-Station or reflectorless methods are more proper near buildings and obstructions. Choose your tools wisely ...
FIX Is A Mathematical Solution Using Available Data
Understanding that trees are around, that available data may have come from satellites less than perfectly arranged.
I will say the precision was accepatble for the purposes intended.
Now that your purposes have expanded it is neccessary to reobserve with the appropriate instrumentation.
Some GPS algorithms assume 4-5 satellites are sufficient for a fix. Whether it is a good fix is another matter. With the number of satllites up there, 10 in view occurs quite often. However I have seen 7 or 8 satellites in skyview line up in a row overhead, say SW to NE. With only 1 satellite of consequence to the NW and 1 to the SE a few trees can easily block out the 2 most impurtant signals for a good position.
Paul in PA
To get from 1 sigma to 2 sigma you multiply by 2. You don't use the ratio of confidence percentages because the statistical distributions are heavy in the center and light in the tails.
For a 1-dimensional normal (bell-curve) distribution with sufficient degrees of freedom, 1 sigma is about 68% confidence, 2 sigma close enough to 95% and 3 sigma 99%. For a 2-dimensional distribution those aren't quite the right numbers but I don't have them on the top of my head.
If you are talking vertical and playing around in the trees, it wouldn't surprise me to find that much difference. Is the base station around trees also? A FIX doesn't mean it's always a solid measurement. You can be fixed in a less than perfect environment and get less than desirable results.
