I have a two SparkFun GPS-RTK-SMA Breakout - ZED-F9P (Qwiic) modules and we want to use one as a base and the other as a rover. The problem is that we first need to fixed a good position with an accuracy of less than 1cm. Our way of doing this is to configure the base in Survey-In mode and wait.
But how long does it usually take to achieve an accuracy of 1 cm or less?
We have been operating the base in Survey-In mode since approximately 6:00 p.m. on November 3 until today, and the accuracy we have achieved is 0.04 m.
You won’t achieve a Base Position with 1cm accuracy with a Survey-In procedure.
This boils down to Random and Systematic Errors with GNSS.
A survey-in doesn’t have the benefit of external data to correct those errors.
For 1cm Accuracy, you have 4 basic choices:
Store a lot of RAW data for a PPP solution (CSRS-PPP)
Store a lot of RAW data for a post processed differential correction (OPUS)
Start with a known survey monument and RTK your New Base location with the Rover
Establish the new Base Coordinates as a RTK solution, with the benefit of a real-time correction source (NTRIP, etc).
#1 and #2 are the only ways to have a real chance at a true 1cm accurate solution. Both will take several weeks to accomplish a defendable position.
It’s very hard to validate a position from #3 and #4 to qualify it as a 1cm accurate solution. There are additional methods such as PPK, L-Band, etc - but they are all similar to the list above.
However, if 1cm precision is fine for you, then things get exponentially easier.
IE: less than a day for #1 and #2, and minutes for #3 and #4
The Survey-In procedure becomes viable for 1cm precision also, but only from that particular position & subsequent setups.
You can quickly confirm by performing a Survey-In and establish a RTK position on a fixed mark with the Rover. Tomorrow, setup the base in a slightly different spot and repeat. The RTK coordinates for that fixed mark might not fall within 1cm day-to-day from different survey-in base locations. However, if you replicate the Base Setup from Day #1 in the future (manually enter the original coordinates), you should expect much better than 1cm precision across missions.
If I accept this accuracy of 2 or 3 cm at the base, this error is also transferred to the Rover, right? But the accuracy between the rover and the base, even if both are, for example, 1 cm further north, will be millimetric, right?
I don’t know if I’m explaining myself clearly, but if I fix the base with that 3 cm deviation, when I move the Rover to two specific points, will it always send me the same positions, or can those 3 cm vary?
Basically, I want to clarify how this error affects the Rover.
The Base will always send the same position in the session, however repeated / subsequent application of Survey-In will result in a different position. If you want consistency you need to remember the prior visit, set that as the fixed location and insure the antenna mounts to a fixed position you can control.
The Base-Line, relative distance between the stations, is what the carrier measurement is constraining. RTK is mainly a geometry problem, where you’re anchoring one end, and not using “corrections”.
Your extended Survey-In has resulted in reporting ~3cm “estimated” accuracy. It’s best to think of that 3cm as a Standard Deviation, and generally a confidence interval of 1-sigma without more details.
The accuracy estimate from any GNSS should not be used to infer or assume the accuracy of the GNSS solution…but rather to show you when something abnormal has occurred. That could be something such as a high PDOP at that time, which you can’t control.
The Accuracy Term is misused far too often in this field, much like Corrections vs Observations as @clive1 pointed out.
If you re-occupy a point with the base antenna at a later date and manually enter the exact same coordinates from the initial Survey-In procedure, the RTK solutions from nearby Rovers will agree across missions. Reminder: that’s precision, not accuracy, and it’s not necessarily repeatable by a 3’rd party without using your exact Base position and coordinates That’s another advantage that a Regional NTRIP has over using a local Base/Rover. I can’t think of a circumstance where I would personally choose a local Base/Rover over the simplicity of a Network Rover (RTN) at a site…because at that point the total station is coming out of the toolbox anyway.
From what I understand:
I have achieved a standard deviation of 2.6 cm. If I now take the MEAN ECEF X, Y, and Z values and use them to set the base as a fixed position.
Once the base position is fixed, we know that it has a standard deviation of 2.6 cm.
Then the absolute position of the rover in the world will also depend on the accuracy of the base, that is, the precision would be:
Error due to the position of the base + 1 cm + 1 ppm
However, the relative position will never be affected by the error generated by the base position. Therefore, if I have a fixed base with an error of 2.6 cm to the north, this error will affect the absolute coordinates of the entire system, with the base and rover displaced 2.6 cm to the north. On the other hand, if I use the relative coordinates to guide the rover locally, that 2.6 cm error will not affect me, and the rover will have a relative accuracy of:
1 cm +1ppm of baseline length.
Yes, my understanding agrees with your summary in the post above.
However, this is where it gets a little hairy. I would agree with that statement if the base error is “within reason”. I typically refrain from commenting about this particular aspect because I haven’t thoroughly tested it myself.
IE: I believe having a base that thinks its position is 1m away from it’s true location could cause problems for Rovers, especially considering repeatability over days/weeks. I just don’t know to what degree. It might be fine for an average user, but you appear to be interested in the specifics (I love that) and care about the integrity of the work output.
My assumption is that error in the Base Position is bad for the rover since the base is essentially saying I’m located at X,Y,Z (with the error) and THESE are the various satellite observations from THAT position, via the RTCM output. That’s a @clive1 question, for sure
That’s another advantage that a Regional NTRIP has over using a local Base/Rover. I can’t think of a circumstance where I would personally choose a local Base/Rover over the simplicity of a Network Rover (RTN) at a site…because at that point the total station is coming out of the toolbox anyway.