Adding some uncertainty to my orbit propagation model


I have created some synthetic range data from the generator classes within Orekit to perform orbit determination with. I want to add some uncertainty to the propagation model used in the orbit determination process such that it is different from the model used to generate the initial range data. Essentially, I would like to add some noise to the drag coefficient for that period of time. I did think of going into the space weather file directly and changing the coefficients there but it felt like a bit of a hack. Is there a more appropriate way for me to do this?


You may try to use TimeSpanDragForce, that allows to set several DragSensitive models with some switch dates between the models.

Hi @PeterBartram,

Just to add to Luc’s answer.
If I understand well what you’re trying to achieve, you could:

  • Implement the DragSensitive interface and add the noise on the drag coefficient(s)
  • Or extend the DragForce class and add the noise directly on the acceleration



I have an additional comment.
Since Orekit 11.0 it is possible to initialize a covariance matrix together with the initial guess for an orbit determination based on a batch least squares. To do this, a SequentialBatchLSEstimator must be used instead of the classical BatchLSEstimator.
This estimator uses a SequentialGaussNewtonOptimizer for which an initial covariance matrix, related to the initial state vector, can be set.

Best regards,

1 Like