Spectral clustering tools applied to Ceres in preparation for OSIRIS-REx color imaging of asteroid (101955) Bennu

Abstract

The OSIRIS-REx asteroid sample-return mission is investigating primitive near-Earth asteroid (101955) Bennu. Thousands of images will be acquired by the MapCam instrument onboard the spacecraft, an imager with four color filters based on the Eight-Color Asteroid Survey (ECAS): b' (473 nm), v (550 nm), w (698 nm), and x (847 nm). This set of filters will allow identification and characterization of the absorption band centered at 700 nm and associated with hydrated silicates. In this work, we present and validate a spectral clustering methodology for application to the upcoming MapCam images of the surface of Bennu. Our procedure starts with the projection, calibration, and photometric correction of the images. In a second step, we apply a K-means algorithm and we use the Elbow criterion to identify natural clusters. This methodology allows us to find distinct areas with spectral similarities, which are characterized by parameters such as the spectral slope S' and the center and depth of the 700-nm absorption band, if present. We validate this methodology using images of (1) Ceres from NASA's Dawn mission. In particular, we analyze the Occator crater and Ahuna Mons. We identify one spectral cluster located in the outer parts of the Occator crater interior showing the 700-nm hydration band centered at 698 +/- 7 nm and with a depth of 3.4 +/- 1.0%. We interpret this finding in the context of the crater's near-surface geology.