StateNoiseCompensation#

class StateNoiseCompensation(continuous_time_covariance: CovarianceMatrix, state_definition=None, sequence_definition=None, max_interval=None, body=None)#

Bases: ProcessNoise

State Noise Compensation (SNC) process noise model.

Implements the SNC approach for modeling unmodeled accelerations in spacecraft dynamics. The continuous-time process noise is discretized using a piecewise constant acceleration model, particularly effective for atmospheric drag, solar radiation pressure variations, and other unmodeled forces.

Process noise model

The unmodeled acceleration is assumed piecewise constant between consecutive measurement epochs \(t_k\) and \(t_{k+1}\):

\[\mathbf{a}(t) = \mathbf{w}_k, \quad t \in [t_k,\, t_{k+1})\]

where \(\mathbf{w}_k \sim \mathcal{N}(\mathbf{0},\, \mathbf{Q}_c)\) and \(\mathbf{Q}_c\) (3, 3) is the covariance of the acceleration uncertainty, in km²/s⁴.

Discretization

Holding the acceleration constant over \(\Delta t = t_{k+1} - t_k\) and approximating the motion as rectilinear over the step, it maps into the position and velocity through

\[\begin{split}\mathbf{\Gamma}_k = \Delta t \begin{bmatrix} \frac{\Delta t}{2}\,\mathbf{I}_3 \\[2pt] \mathbf{I}_3 \end{bmatrix}\end{split}\]

so the discrete-time process noise covariance is

\[\begin{split}\mathbf{Q}_k = \mathbf{\Gamma}_k\,\mathbf{Q}_c\,\mathbf{\Gamma}_k^T = \Delta t^2 \begin{bmatrix} \frac{\Delta t^2}{4}\,\mathbf{Q}_c & \frac{\Delta t}{2}\,\mathbf{Q}_c & \mathbf{0} \\[2pt] \frac{\Delta t}{2}\,\mathbf{Q}_c & \mathbf{Q}_c & \mathbf{0} \\[2pt] \mathbf{0} & \mathbf{0} & \mathbf{0} \end{bmatrix}\end{split}\]

where the row/column partition is \((\mathbf{r},\, \mathbf{v},\, \mathbf{p})\). Static parameters \(\mathbf{p}\) receive no process noise. Integrating white noise over the step instead gives \(\Delta t\,[\tfrac{\Delta t^2}{3}, \tfrac{\Delta t}{2}; \tfrac{\Delta t}{2}, 1] \otimes \mathbf{Q}_c\); the form above is larger for steps longer than a second, which suits an acceleration that is not estimated, and has rank three, which the SRIF needs to factor it through \(\mathbf{\Gamma}_k\). It grows quickly with \(\Delta t\), so across long data gaps \(\Delta t\) should be capped with max_interval.

Covariance propagation

At each time step the propagated covariance is augmented by the discrete process noise:

\[\bar{\mathbf{P}}_{k+1} = \mathbf{\Phi}_{k+1,k}\,\mathbf{P}_k\,\mathbf{\Phi}_{k+1,k}^T + \mathbf{Q}_k\]

where \(\mathbf{\Phi}_{k+1,k}\) is the state transition matrix and \(\mathbf{Q}_k\) captures the uncertainty growth from the unmodeled acceleration over \([t_k, t_{k+1}]\).

Parameters:
continuous_time_covarianceCovarianceMatrix

The continuous-time process noise covariance matrix.

state_definitionoptional

State vector definition. Defaults to None.

sequence_definitionoptional

Sequence definition for multi-leg trajectories. Defaults to None.

max_intervalArrayWUnits or float, optional

Longest interval the noise is accumulated over, in seconds when a float; a longer one, such as a data gap, receives the noise of one this long (see _noise_intervals()). Defaults to None, no limit.

bodyBody or int, optional

Body whose position and velocity the noise drives, or its SPICE ID. Defaults to None, the owner of the first position in the state.

See also

scarabaeus.DynamicalModelCompensation

Exponentially-correlated DMC variant.

scarabaeus.ProcessNoise

Abstract base class.

scarabaeus.ProcessNoiseSettings

Configuration object consumed by the filter.

Notes

The noise drives the position and velocity of body wherever they sit in the state; every other state, of that body or another, receives none.

Dynamic components must be position or velocity, and must precede static components in the state vector.

References

[1]

Tapley, B. D., Schutz, B. E., & Born, G. H. (2004). Statistical Orbit Determination. Elsevier Academic Press. ISBN 978-0-12-683630-1.