Data‐driven taxonomy matching of asteroid and meteorite

Yuki Saito, Peng K. Hong, Takafumi Niihara, Hideaki Miyamoto, Kenji Fukumizu

Meteoritics & Planetary Science
First Published: 19 January 2020


“The matching of asteroids and meteorites is a significant step toward a better understanding of the origin, structure, and history of the solar system. We propose a data‐driven approach for investigating common taxonomic structure between asteroids and meteorites; C‐, S‐, and V‐type for the former, and carbonaceous chondrite, ordinary chondrite, and howardite‐eucrite‐diogenite (HED) meteorite for the latter. In the numerical experiments, by checking whether the taxonomy information of meteorites improves classification for asteroid data, we examine the existence of common structure over the two domains. For this purpose, we compare the resultant accuracies of two clustering methods which are with/without the guidance of meteorite data. We observe that the guidance of meteorite taxonomy improves the accuracy for classifying asteroids, either with the reflectance spectra or major chemical compositions of meteorites. This fact serves as a piece of evidence that there is a common taxonomic structure and links between meteorites and asteroids, supporting a long‐standing hypothesis.”