Skip to content

RTK Correction Networks Accuracy

Graeme Cox explains use of wide area RTK Correction Networks for surveying and landforming.

Related article: Practical Test on Accuracy and Usability of Virtual Reference Station Method in Finland


Hi, it's Graeme Cox here.  I've just had a client who's inquired about using a Virtual Reference Station or VRS for the correction signal to do ag earthworks or OptiSurface earthworks. They've inquired about the accuracy so I did a bit of research and found this paper from Finland.

VRS system is basically a network of base station spread out over a large area typically spaced about 70kms apart or similar. This is an example here of a network in Finland.  They're usually using the mobile network to send the correction to the earthmoving equipment. I'm concerned about that because we recommend a base station right beside the field for the highest accuracy. When you get the nearest base station will potentially be 35 kms away so I'm a bit concerned that the accuracy of this system would not be recommended for earthworks.  Remember we're pushing the accuracy limits of the GPS system when we're doing OptiSurface landforming. 

Here are some results that show the accuracy.  They basically took 2000 measurements all around that network and this is the type of accuracy results they got.  This is the northing error and this is the easting error.  And this is the vertical error as you can see it is worse as you'd expect.  Based from all of them you'd think 95% confidence is probably +/-75mm. 68% confidence which is often quoted as probably more around 35mm.

What's interesting in this paper is they also looked at the distance from the nearest reference station.

Figure 8. Connection between accuracy and distance to nearest reference station

This is the horizontal error and this is the vertical error. You can see the horizontal at 95% confidence which is the upper enveloper here, it does increase as you go over time although the average error here probably does increase that much but were talking at 95% confidence and it does increase.  Maybe will flatten a bit here somewhere which would make sense.

But what we're more interested in is the vertical error for earthmoving and you can see that when the base station is very close, we have a 95% confidence of around 25mm and then it increases quite rapidly for 12kms up towards 100mm for 95% confidence.  We halve that for 68% confidence or one standard deviation. 40mm halved equals 20mm and then 100mm halved equals 50mm.  Based on that, I wouldn't recommend VRS for doing OptiSurface type earthworks because that amount of error is quite significant.  That's why we like to have a base station right beside the field so then the base line will be about a kilometer away so we're operating down this region here.

OK. I hope that's useful. If you got any questions, just shoot an email to and we'll try to answer them.

Feedback and Knowledge Base