Any benefit to running more than one base?

So Single base RTK works well enough, but always in search of more accuracy.

If one were to run three base stations with known or unknown coordinates, each sending corrections to the next base in the triangle.

Then the rover taking corrections from one of the bases in the triangle.

Would this increase accuracy or stability of the corrections?

This is right up your alley to answer!

1 Like

I don’t know that this is possible with Reachview as you would get conflicting corrections from multiple coordinates. We accomplish this with Topcon because of localization so the bases are reporting from the same network error constraints when they swap from one to the other depending on proximity. I don’t think Reach has this capability either.

The rover would only get corrections from the one of the three bases, no conflict there.

Its if the base one correcting base two, base two correcting base three, and base three correcting base one, has any benefit.

1 Like

I don’t think this is the case if they all have the same configuration. As you get closer to base two and lose RTK from one you will be on a different basis unless you really pay close attention to the accuracies and have manual coordinates set.

Just a thought, but with the Topcon we also have the capability of using a repeater. As long as it is also configured correctly it will throw the corrections from one base further. Has anyone tried something like this with Reach. Would be a good option.

Also it might be worth thinking about a caster…

1 Like

This would only be possible for PP (after the fact), real time no. You would also have to buy a good PP software such as Javad Justin. It’s a great PP package, but learning curve a little steep. Best out there from the super geodesists at Javad. Rtklib would not work.

As I’ve said in many posts here, the whole purpose for 3 units ( two as static, one as rover) is to have a closed polygon so that PP you’ll know the accuracy of the rover point. Not that I don’t trust RTK, but it’s always good to know what the true accuracy is for the rover point. I use my rover with the RTN, and I have realtime data via RTK, but I also PP against the two local static units to verify.


Hi all,

Thanks for the discussion! I have some thoughts in this regard as well :slightly_smiling_face:

Its if the base one correcting base two, base two correcting base three, and base three correcting base one, has any benefit

In such a setup, at the last step (when the base three passed corrections to the base one), we should compare initial coordinates with new coordinates.

For example, we’ve got a significant difference in the base one position. However, I cannot think of how we can define which coordinates are the right ones. We could just accumulate the error during the previous steps.

As Brayan said, this task might be resolved by post-processing. Because in this case, we can distribute this difference between all bases according to baselines. However, it can’t be done via ReachView in real-time.


For a software like GNSS solution this would be beneficial in a PPK situation.
Compute a dynamic VRS (baseline of few meter from your rover on the entire track) hence getting a short baseline for the entire run.


Ok very interesting, what if we flipped it around.

One base on a known point. Three separate rovers mounted together in a triangle ten centimetres apart.

You write a program to average the three rovers to find the center of the triangle.

Does the output of the programs have better live accuracy than a single rover?

1 Like

Sorry, I may have misunderstood your scenario. I think the 3 rovers’ positions are similar to taking 3 RTK measurements of the same spot. You will just compute the 3 triangle points to get the centroid. All 3 points are just a single RTK result. I still don’t think you would get any better or worse position accuracy as compared to averaging 3 continuous shots. You would however get additional IMU parameters - pitch, yaw, roll.

1 Like

And some redundancy if one should fail because of e.g multipath

When used on a moving rover, it’s harder to take 3 measurements on the same spot, having tripled the hardware allows for more parallel solutions (but probably not synchronised between each other, so it’s still not 3 measurements of the same spot, but they’ll be closer).

I think a position based on IMU/INS/GPS fusion/filtering can result in better performance (in precision agriculture), doesn’t help with incorrect fixes off-course…, but I think it could provide a more robust-realtime position (with less time-delay, and some compensation of noise/error/jitter on GPS-solution).

Trimble has a system called VerticlePoint RTK. It was created for the people leveling rice fields. They want them perfectly flat.

The issue is, even with RTK, there is some drift in height or elevation throughout the day with a single base configuration. You can see a +/- 1 to 2 inch drift throughout the day on height. They somehow combine 3 base stations to eliminate this. I have never used the system but was told it worked very well.

There was a guy from Australia or New Zealand on New AgTalk forum that came up with a cost affective method of correcting for this. He mounted his base station globe on an electric mast (basically, an over priced electronic linear actuator) that was controlled by the base station Ht. If it changed by more than 1/4 inch, it moved the globe up or down to that Ht. The output of the base remained constant and it corrected for drift this way.

Took a while to find the old post but here is a link to it.


This is where mmGPS or laser assisted GPS is best. This would be a great interface for Emlid.

1 Like

Yes I know. GPS replaced laser in this application. Laser has its downfalls as well. Keeping the transmitter stable if it is windy, variance in the light if it is hot due to heat shimmer etc. These were two issues that the rice fields ran into all the time. GPS was thought to be the “savior”. VerticlePoint or the other method makes it work very well.

I have been around the automated grade machinery since its inception. I actually know how to set target to control slope using a sight rod or the string method. I started with lasers when they were manual level units with tracker heads.

I have tried to sum up some of the options here:

1 Like

This topic was automatically closed 100 days after the last reply. New replies are no longer allowed.