Precision of Millimeters with >100km baseline

Hi all,

Just to share my results and tell how satisfied I am with my reach units. I’m from Asuncion, Paraguay and here we don’t have any permanent base reference (CORS), so I have to do my post process using CORS from the neighbor country, Argentina. The nearest one is at 120km in Formosa. The image tells the rest of the story,

I didn’t have a professional grade GNSS at the time to compare the accuracy. But anyways, this result surprised me on how well it maintain a fixed solution and a very tight precision. I will leave here a link to the data if you are interested in playing with.
Also, to be sure of this I did another measurement some kilometres away from the first one and obtained almost same precision results.

Details of my setup:

  • Image: 1.2, ReachView:0.4.9
  • 9.5x9.5cm Aluminum ground plane
  • Systems: GPS, GLO and SBAS
  • Ublox config: GPS_GLONASS_5Hz


Nice results. It is a good demonstration of the resolution of Reach. As Emlid has stated from day 1, l1 single carrier is good at accuracy at short baselines up to 10km.

What RTKLIB results show is not the accuracy of the results, but how precise is the measurement based on the data used.

Reach is targeted toward short baselines less than 10km to get cm accuracy. As the baseline is extended, the precision of the answer may be good, but the answer accuracy not nearly as good.

This is where a good GNSS surveyor shines.

I am not a professional surveyor, but I have done a lot of testing.

Most of target uses of REACH are short baseline where there are small changes in orbits, troposhere, ionosphere, clocks, etc.

You appear aware of improvements in accuracy of results versus precisions.

I have done many tests on L1 long baselines that showed precise answers to cm -mm, but were more like dm to m in actual location of results.

In your situation, if you need accurate GNSS results, and establishing a local accurate base coordinate by a surveyor using dual frequencv equipment
is not feasible, I would follow establishing your own base station using results from the 100 km base over many days in postprocessing.

I hope I did not carry on too much or say something wrong, but wanted to emphasize the many difficulties in long baseline accuracies

Wish you the best.


Many of the errors associated with long baselines could be eliminated by using precise products from IGS.


I agree. It’s better to use actual data measured at the time of the data collection than extrapolating past data from models.

so interesting. I have been exploring GNSS area for more than 3 years and today i got proper explanation why L1 signal only good for < 10 km baseline. Great. Thanks for information