The Science of Non-resolved Space Object Signatures for Space Domain Awareness

This project is focused on leveraging modeling and simulation of non-resolved space object signatures utilizing the DIRSIG software model to better understand spectral signature phenomenology. The non-resolved spectral signatures are typically represented by time resolved spectral light curves which are a function of the space object materials, surface orientations, illumination conditions, and viewing conditions. The DIRSIG software model is being used to generate spectral and polarimetric light curve training data for project collaborators to develop machine learning algorithms to characterize specific properties of unknown space objects. This characterization will ideally help identify the fundamental surfaces, shapes, and materials present on the unknown objects and ultimately aid the sponsor with a more thorough understanding of the object’s purpose, operational status, and possibly future intent.

During the second year of the project, RIT focused on developing a two dimensional signature space related to object surface orientations resulting in each space object having a unique spectral/angular fingerprint. We explored methodologies to simulate the fingerprint, leveraging sponsor provided CAD models, RIT spectral goniometer BRDF measurements of space material surfaces, and the DIRSIG software model. Additionally, we initiated an approach to transition non-resolved spectral field measurements of geosynchronous satellites into this fingerprint signature space. Collaboration with GTRI and USAFA resulted in an initial streaming Machine Learning approach to estimate satellite shape from these spectral/angular measurements and simulations, and was shared with the Space Domain Awareness community in a presentation at the 2025 AFRL VOLTRON conference.

Figure 1: Example demonstrating DIRSIG simulations of non-resolved spectral/direction fingerprints of Mexsat-3, DirecTV-10, Spaceway-1, and Echostar-10 geosynchronous satellites.