In this paper the classic approach based on linear covariance analysis using State Transition Matrix (STM) is presented. However, it is explicitly written that the STM should not be computed as solution to the differential equation Phi_dot(t) = A * Phi(t) with A being the Jacobian of the spacecraft dynamics with respect to the spacecraft state, but using finite-differences because in this way the effects of the finite maneuvers are included in the STM.
I would like to ask if in Orekit the STM is evaluated only as solution to the differential equation Phi_dot(t) = A * Phi(t), and in this case how to replace it with finite-difference method approach.
If any valid alternatives exist for propagating the state covariance through a finite maneuver they are welcomed.
As far as I know the effect of a finite maneuver is taken into account when computing the state transition matrix with a numerical propagator in Orekit.
You can even get the derivatives of the thrust parameters with respect to the initial state.
Unless I don’t fully understand your question, there’s no need for you to use finite-differences.
thank you for your reply. I am happy to know that the effects of the maneuvers are already included in the evaluation of the STM.
However, there is still an open point that maybe was not clear in the question and it is the following: imagine that I have uncertainty information (variance) related to the thrust magnitude and pointing direction. How can I include such uncertainties in the numerical propagation and propagate these additional variances as the classic state covariance?