Statistical covariance realism assessment of LeoLabs’ orbit determination system
We investigate the covariance realism of LeoLabs’ orbital ephemeris data and an applicable metric for consistency monitoring to improve reliability. An overlap analysis has been developed for this work. In this, the Mahalanobis distance between two propagated states is computed over many orbits. We then perform a containment analysis to characterize the Mahalanobis distance distribution. We show that 95.2% of Mahalanobis distances are less than or equal to 4.0, and 97.4% are less than 5.0.
Throughout this research, we verify that LeoLabs’ covariance keeps its realism through a 7-day of propagation window. We also demonstrate that the Mahalanobis distance distribution follows a log-normal distribution. Lastly, we define two applicable consistency monitoring metrics from the containment percentage and fitted parameters of the log-normal probability density function.