Aryzbe Najera, University of Texas at El Paso; Ana C. Chavez-Lopez, University of Texas at El Paso; Leah A. Porras, University of Texas at El Paso; Miguel Velez-Reyes, University of Texas at El Paso
Keywords: DIRSIG, SDA, Unresolved Space Objects, Satellite Characterization
Abstract:
Hyperspectral remote sensing is an approach to extract quantitative information about Resident Space Objects (RSO) to support Space Domain Awareness (SDA). Measured spectral signatures contain information about material composition, pose, spin rate, among other parameters associated to the RSOs. The high spectral resolution provided by hyperspectral imaging allows extracting properties associated to resident space object even if they are not spatially resolved. Consequently, precise signature interpretation may allow us to understand and react appropriately to changing situations in the space domain.
Unmixing has been widely used in terrestrial hyperspectral remote sensing to extract subpixel material composition information in low to moderate spatial resolution images. Existing algorithms take advantage of spatial domain information to extract material spectral signatures and their spatial distribution. However, for hyperspectral imagery of unresolved RSOs (URSO), few pixels in the image contain information about the space object, therefore, spatial domain information is not available for unmixing. Temporal domain information where RSO data is collected a function of time and might provide the needed information to perform unmixing analysis.
This paper presents experimental results on hyperspectral unmixing for extracting material composition RSO from unresolved imagery. The experiment utilizes simulated data generated by a physics-based image simulator to allows to have a better understanding and look at changes in the ground-based measured spectral radiance for RSO and URSO satellite to material composition and satellite orientation. The Digital Imaging and Remote Sensing Image Generation (DIRSIG™) simulator is used to simulate ground-based hyperspectral observations of resident space objects. Using a computational tool allows us to perform different studies not possible with actual measurements from systems in operation. The main simulation scene parameters consist of a (1) CAD model for a DirecTV-10 satellite (Boeing 702 Bus); (2) the spectral signatures for the materials assigned to the satellite’s surface, (3) the geographic location of the observer (USAFA-16 Telescope) and satellite via its Two-Line Element (TLE); (4) atmospheric conditions; and (5) the instantaneous field of view (IFOV) determined by the focal length of the optical system. We study the effects of spatial resolution and material composition in the performance of the unmixing process as well as address identifiability of materials from the data. We also discuss the resources needed to implement the simulation model in the DIRSIGTM environment.
Date of Conference: September 17-20, 2024
Track: Satellite Characterization