Covariance-based Uncorrelated Track Association

When the Air Force Space Surveillance Network observes an object that does not correlate to an entry in the Space Object Catalog, it is called an Uncorrelated Track (UCT). Some of these UCTs arise from objects that are not in the Space Catalog. Before a new object can be added to the catalog, three or four UCTs must be associated so that a meaningful state can be estimated. Covariance matrices can be used to associate the UCTs in a more statistically valid and automated manner than the current labor-intensive process. To perform the association, the state and covariance from a UCT at a certain epoch must be propagated to the epoch of another UCT state estimate. The uncertainty in UCT state estimates can be large due to limited data and tracking geometry. When the UCT is propagated more than a few hours, the in-track uncertainty becomes much larger and the uncertainty quickly becomes non-Gaussian if it is expressed in Cartesian coordinates. The state uncertainty expressed in elliptical curvilinear coordinates remains Gaussian for longer propagation times and larger in-track separations than when expressed in Cartesian space. Simulations show that automated covariance-based UCT association using curvilinear coordinates performs very well even with time-varying measurement biases.