Age of RTK corrections - fixed solution accuracy

Does anyone know how the age of NTRIP corrections influences horiz/verlical accuracy of a fixed RTK solution? Would be nice to see a test of that.

I read that correction loss for few seconds or even a minute may still be acceptable for some applications. I ask this because I live in Europe where we have a nice dense CORS network equivalent but most countries charge a fee for its usage based on time (units are seconds). So one could save lots of money by asking for corrections less frequently (maybe every 10 seconds or even less).

I want to survey from moving vehicle over many kilometers in length so I can not use my own base station. My plan is to use a virtual CORS station that will move with my vehicle - my county offers such service.

I found one post with vertical test (unknown setup):

I am interested in similar horizontal test for Reach.

I don’t have an answer for you, but for best accuracy, wouldn’t you want to log your route first, and then download the historical CORS data after you get home? If you need to do it in real-time, then disregard my comment.

Since you are talking about age of corrections, I’ll assume you are talking about RTK processing. RTK solution requires a constant feed of these corrections. Age value is there for you to notice any communication problems, as they influence the solution in a bad fashion.

Reach is able to extrapolate correction data for quite some time, but this is a measure for overcoming communication problems and has not been tested in conditions of constant “correction starvation”.

I am not certain, but most VRS networks provide a virtual reference station close to your rover’s starting point and it doesn’t move after this. Are you sure it will move?

Overall, I would advise stick to post-processing for this test. Should be much easier, cheaper and you will likely be able to use just one reference station. What’s your expected route’s length?

1 Like


Thanks for your reply. My question is more about hacking RTK system. I can understand this is not a priority for Emlid.

Some of the things I am talking about can probably be tested easily on an RTK dataset with RTKLIB - you simply cut the corrections and see how the solution degrades in time. Unfortunatelly I am not really familiar with RTKLIB yet.

I will check if my provider’s VRS can do moving rover. I expect routes with undefined length, maybe even 100s of km (application in autonomous vehicle).

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