We were able to successfully run a survey with our two RS2 receivers in RTK mode using Lora, but now want to better explore the accuracy and uncertainties by PPK processing using our RS2 base. It seemed like the RTK position was based on an initial 2 minute occupation, and that the GNSS position wandered quite a bit more than was represented by this initial occupation- leading to high uncertainty in our surveyed points. We don’t unfortunately have access to any known points to set up or survey.
We have followed the directions to convert the files off the base & rover to RINEX, but now we’re unsure how to correct the coordinates of our survey points. We have the survey file exported from the phone that has timestamps for our point occupations, but we’re not sure how to match these to our PPK corrected rinex files.
Is there a guide we missed somewhere on how to do this properly? I searched a bit but couldn’t find anything.
I think you’ll probably start here and use the base logs to define the globally accurate base point and then you can shift everything else once that has been defined. Others here are much more experienced with PPP and I am sure can help with the process.
Thanks Michael. It seems we need to do this prior to beginning our survey so that we can entire the precise base position into the reachview 3 app. I was hoping there’s a way to do this in tandem with the rover survey.
Correct, this is a separate process up front and the amount of time it takes to turn around varies but I have heard it come back within 30 minutes at times.
Why do you need absolute global accuracy? As long as you use the same coordinate on the base or a coordinate from one of the points RTK’d from that original point each time you return you will be consistent. If you didn’t document the point then you should be able to get it from PPK as the reference point in the RINEX header.
We’re using the survey points as GCPs for a drone survey. The drone also has its own PPK process. Unfortunately I think our large source of repeatable error between our drone survey models and the locations of the GCP’s is due to the variation between the GCP’s correction (due to it being based on a short occupation time to solve for the base) and the drone’s correction based on the base’s entire occupation of the site.
I’m not sure how to extract timestamps or otherwise to perform a PPK correction for our GCP’s.
It is definitely best to have a local base in that case. Are you setting your GCP’s and then immediately flying the drone? If so let the base keep running and use it to correct both. Let it run for at least 10 minutes.
Yes it generally runs for at least ten minutes before we survey in the GCPs.
I think with the amount of scatter/wander we have seen in the base location file we will have this issue with any RTK solution unless it is a very long occupation time.
Even a very long time averaged is going to give you up to a meter of scatter but like I said as long as you use the same manually entered coordinate for return visits and the same base log to PPK the drone you will be on par in relativity. Entire surveys can be corrected at a later point to be globally accurate if it is really necessary.
It should be possible. We could summarise the process this way:
First visit: set-up base, average for some time and use that as base position. Let it log raw data for the entire survey time (hopefully 3-4 hours). Record GCPs in RTK according to the averaged base position, run your drone mission.
Back in office, post-process base raw log. Shift GCPs according to corrected base position.
Subsequent vists: enter corrected base position manually, work in RTK. For added security, log raw data and double-check against other known points if available.
The only thing I couldn’t see from what you posted is if your drone’s PPK relies on the RS2 or not.
In that case you should be able to use the base raw log to post-process the drone track. You just need to specify the corrected position, in what I assume you’re using is RTKLib, from the options/positions tab.
If you project all the points, base and GCPs, to a “cartesian” system, then you can simply apply the dX, dY, and dZ between the averaged and corrected base positions to the GCPs that were taken relative to the averaged base. That can be done in Excel.