Characterization of Resident Space object States Using Functional Data Analysis

To date, most characterization techniques (e.g., using photometric light curves) take place using time and frequency domain analyses of data samples generally lacking in the complete information content needed for unambiguous characterization of non-resolved Resident Space Objects (RSOs). In this paper, the information content of multiple measurement types is examined using information theoretic and functional data analysis (FDA) approaches which have shown promise in characterizing the physical and dynamic attributes of space objects from non-resolved observations. With limited data and information, it may still be valuable to understand whether the "state" of an RSO is: (a) active (operational), (b) passive (debris), (c) dormant (a potential threat acting passive), or (4) transitionary between any of 2 of the a-c states. Representative use cases are established, and the information content is examined in a probabilistic context for a set of simulated astrometric, photometric, Long Wave Infra-red (LWIR) and Radio Frequency (RF) observations for a diverse set of object shapes, sizes and dynamics representative of states a-d are used to demonstrate the application and value of FDA. The results confirm the value of these approaches by correctly categorizing independent sets of measurements and quantifying the likelihood of a given combination of observation types as being associated with a specific object. The value and information contribution of each observation type to the characterization is assessed by virtue of the Hellinger Distance metric.