Bias on measurement angles computed from CPF

[MaximeJ]

Hi @nrol,

@nrol

I found a lead on the 2nd problem : if I change the FramesFactory.getICRF() to a FramesFactory.getGCRF() in the AngularRaDecBuilder I get much better RADEC measurements.

I think it’s a bug in AngularRaDec. Line 171 should probably use transformVector instead of transformPosition (the first one apply only the rotation while the second applies rotation + translation).
Could you please open a bug for this one too ? (two bugs in a post !! Thanks)

@nrol

Anyway, I’m now looking for a way to solve the 1st problem (generating measurements on a list of specific dates)!

Two solutions in my opinion:

1. You already have the reference measurements as Orekit ObservedMeasurements
   → Use a ResidualHandler with your propagator (see this post for an example)
   Afterwards you get a Map of (Observed, Estimated) and you can compare them.

2. You don’t have a list or it’s too difficult to obtain.
   → Use the list of dates in the handler instead of a list of ObservedMeasurement
   In the ResidualHandler before line 90 you need to initialize the observed variable.
   Initialize it like you would do a measurement (with sat, frame, weights, sigmas etc.) and put [0,0] for the observed values.
   When observed.estimate(…) is called, Orekit will compute the estimated (i.e. theoretical) measurement. However the residuals in this case will be meaningless.