Joseph Coughlin (Master Solutions)
Keywords:
Abstract:
Correctly determining satellite health and status is an important aspect in ascertaining satellite anomalies. Spectral data has often been proposed as having the potential to provide one part in the larger puzzle of determining satellite status. This paper presents the results of our study to clarify the utility of spectral data in ascertaining satellite anomalies, as compared to conventional broad-band imagery, using spectral similarity techniques.
This study focuses on the extraction of time varying signature distributions for non-resolved images using spectral similarity techniques. Analysis of the time varying distributions has potential to show the rough material similarities as a function of time without requiring a thorough material end-member extraction for each spectral image. We model time-varying multispectral and hyperspectral satellite signature images for different satellite models, configurations, and lighting conditions in the Thermal Infrared spectral region for a realistic sensor system. We analyze the resultant spectral signatures to determine spectral similarities and material distributions. By varying the model configurations we analyze the presence and distribution of different materials, reflectivity changes, rotation rates, and temporal changes to ascertain the effects of potential anomalies, such as incomplete or improper solar panel deployments.
To clarify the utility of spectral data, we compare the broad-band time variability in the signatures to the spectral similarity time variability. Our initial results indicate that spectral similarity derived from the spectral signatures can yield relevant information to determine satellite anomalies. This paper presents the results of our internal study to clarify the utility of spectral data in ascertaining satellite anomalies.
Date of Conference: September 16-19, 2008
Track: