Weights on the observation time

Hi all,

I’m new to Orekit and just discovering all its functionality. The problem I’d like to solve with Orekit is straight forward orbit determination/fitting. I have an initial orbit (TLE) and some measurements (RA and DEC). While RA and DEC are very accurately measured, the observation time is not precisely known. My goal is to fit the initial orbit to exactly match my observations and then determine the observation time from the fitted orbit. I found several tutorials describing how to use the BatchLSEstimator for this kind of task. However, as far as I can see, there is no possibility to add an uncertainty on the observation time but only on the measured positions. Do you know if there is a way to make the timestamp a free parameter for the estimator?

Thanks a lot,

Hi @shellmich,

Welcome to the Orekit community !

Good to read ! :slight_smile:

Not that I know of and I don’t think it will be easy. But I may be wrong and maybe another user will prove me wrong.

There is a ClockOffsetDriver parameter in the GroundStation object that you may use.
However if you use a batch least-square orbit determination it will give you a global clock offset (identical for all measurements).
So this will be adapted for a “constant” clock bias.

If you want to have a different clock offset for each measurement, you should use a Kalman OD. But then maybe you will have observability problems, meaning that the filter won’t be able to know if the error comes from the measurement values or from the clock offset.

It may be worth a try, I never personally attempted it so I can’t tell you if it will work.

Hope this helps,

Hi @MaximeJ and thanks for your quick response! We (hopefully) corrected for the systematic errors in the timing and are now left with some random errors coming from the acquisition system that we can not correct for. So I guess that the GroundStation parameter won’t work for us but I’ll definitely check the Kalman orbit determination!

If you have random timing errors, these cannot be distinguished from angular errors. The reason for this is that changing a date changes orientation of the Earth, hence changing the angles.

I guess you must just accept them and they will creep into the residuals.