Usually for post processing DGPS in any technique (RTK - Carrier phase, Pseudorange code) the RINEX files used contains data for many consecutive epochs. This leads to very large RINEX files.

Is it possible to apply those techniques for single epcoh? If not, does skipped series will work (thus if epoch is 1 seconds, so each 5th epoch will be recorded thus every 5 seconds)? For both cases, if it's possible, does the accuracy degraded?

The nature of the question is to understand whether the underlying differential GPS algorithm is has look ahead/behind buffer? [i-1, i, i+1].

2 Answers 2


Rate of recording GNSS observation, depend mainly to application you want. For example High-rate observation (like 1s) can be used in disaster monitoring. BUT for Positioning application mainly 15s or 30s is common rate.

The nature of the question is to understand whether the underlying differential GPS algorithm is has look ahead/behind buffer? [i-1, i, i+1].

This question depend to estimation method, In Backward solution or Forward solution No, In combined solution Yes.

  • Hi, can you please elaborate more about the estimation methods? This is the nature of the question: different methods that can with with single epoch.
    – michael
    Jul 6, 2018 at 19:00
  • 1
    @michael , If you want estimate just single epoch and this single epoch has not any Constraint (note that we put this constraint) with previous epoch or next epoch Least Squares estimation is best practice. Jul 6, 2018 at 19:37
  • Buy how you do least squares with single observation (single epoch)?
    – michael
    Jul 7, 2018 at 15:28
  • 1
    what do you mean single observation? or single epoch? (single observation is not mean single epoch), single epoch contain at least four satellite observation Jul 8, 2018 at 6:09
  • Hi, in my terminology single observation=single epoch=X sat from Y constalations that were recorded at that epoch.
    – michael
    Jul 8, 2018 at 8:44

Basic DGPS with simple or double differencing needs simultaneous observations from the base and the mobile receivers. So if you have the same epochs, you don't need the other epochs. It is useless to have a higher frequency with the mobile receiver than with the base receiver, but make sure that your time steps are compatible. Note that DGPS achieves 1 m resolution.

For better (<1cm) resolution, you need to use RTK. In this case, the "integer ambiguity" has to be resolved. In post-processing, this is facilated when the satellite geometry changes, which means that you need a long observation period (without interruption).

To sum up, 1) your precision will be 1 m with a single epoch (DGPS only). 2) You can still solve the "integer ambiguity" if you reduce the frequency. It is not necessary to have high rate (except if you are interested in the movement of your receiver) but long time of observation is recommended (depending on the estimation method).

  • For RTK, are there "estimation methods" that theoretically require only single epoch?
    – michael
    Jul 6, 2018 at 18:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.