Early stage characterization of on-orbit fragmentation events

Currently, fragmentation events are the predominant source of space debris, thus making Fragmentation Analysis one of the primary Space Surveillance and Tracking services for supporting operators in space traffic management. In order to promptly characterize a break-up event, two essential activities must be performed: the association of a detected object to a fragmentation event and the detection of the event epoch. A stochastic approach is considered fundamental to act operationally in the period immediately following a fragmentation, when data are uncertain and scarce. This study presents a newly developed tool called FATE (Fragmentation Association and Time Estimation), which aims to achieve the two objectives described above for the prompt and complete characterization of break-ups in the early stage after a fragmentation event. The tool is comprised of two algorithms which can be unified in a single pipeline. The association problem is addressed by the first module of the tool, which performs a Chi-Squared test based on the Mahalanobis distance to assess the statistical matching between two distributions found in the right ascension of the ascending node-inclination plane. One distribution is derived from the last available ephemeris of the parent, and the other is built from the angular measurements track of the observed object. The latter is obtained through the admissible region concept, and allows to perform this analysis starting from a single measurements track and without an Initial Orbit Determination (IOD) result. With the information of the associated objects, the IOD process can be performed on the fragments associated to the event, and the second portion of the algorithm, based on an upgraded version of the FRagmentation Epoch Detector (FRED) tool with the novel use of Conjunction Analysis methodologies, can be exploited for the identification of the event epoch. The proposed approach employs Gaussian Mixture Models (GMM) to describe and propagate the parent and fragment state uncertainties throughout the process. The algorithm then computes a set of candidate fragmentation epochs and subsequently ranks them using the Probability of Collision metric to determine the most probable event epoch. In cases where the observed object is associated with the fragmentation event under analysis, the epoch detection algorithm can achieve a more precise estimation of the epoch and location of the event. The performance of the FATE tool is evaluated in a realistic simulated scenario. Concerning the association part, the simulations demonstrate satisfying accuracy, even when the fragmentation epoch is not available yet. Regarding the epoch estimation part, its robustness to increasing Orbit Determination errors and presence of dynamical perturbation is verified as well. The upgraded version of FRED benefits from the GMM-based approach, which evidently reduces the computational effort.