Spectral classification of Gaia DR3 Solar System small bodies and application to the search for A-type olivine-rich asteroids

The Data Release 3 (DR3) of ESA's Gaia mission contains 60 518 mean reflectance spectra of Solar System small bodies, spanning the visible wavelength range in 16 bands [1]. Such large and homogeneous dataset is a powerful tool to study the Main Belt as a whole.

We developed a classification method for the DR3 dataset, focusing on the search for new potential olivine-rich A-type asteroids in the Main Belt. Indeed, there is an observed scarcity of purely olivine-rich asteroids in the Main Belt known as the "missing-mantle problem" [2,3]. DeMeo et al. (2019) [4] derived from NIR spectroscopic observations that A-type asteroids account for less than 0.16\% of the Main Belt, for asteroids with a diameter above 2 km. We tested this assertion by exploiting the Gaia DR3 dataset, as this low quantity of olivine-rich bodies contrasts with differentiation theories [5].

We developed a classification method for the DR3 dataset based on a curve matching algorithm that gives the best two spectral classes associated with an asteroid spectrum [6] after a comparison with spectral classes template spectra. To take into account inherent differences existing between DR3 and ground-based spectra, we defined Gaia DR3 template spectra based on the Bus taxonomic scheme [7] to perform the classification.

We filtered the Gaia DR3 dataset to test and apply the classification algorithm on the best-quality spectra only. We considered only spectra with an average signal-to-noise ratio above 30 and without flagged bands from 462 to 946 nm, which left us with 18 739 DR3 spectra. This sample will be refered to as the "filtered DR3 dataset" in the following. We then designed the classification using a sample of objects characterized from NIR or VISNIR spectroscopy in the literature and having a spectrum in the filtered DR3 dataset. Using a trial-and-error approach, we improved the classification of these objects by eliminating certain sub-classes from the Gaia templates and grouping others into larger complexes. The confusion matrix corresponding to this classification is displayed in the figure below. It is quite diagonal, indicating satisfactory results.

To improve the classification of A-type asteroids specifically, we defined a secondary classification step based on the blue part of DR3 spectra only, from 462 to 594 nm. We exploited the fact that this wavelength range does not appear affected by the reddening phenomenon impacting some DR3 spectra compared to ground-based spectra, which allowed us to distinguish between real A-type asteroids and false positives. This secondary step allowed us to classify correctly most A-type asteroids, while keeping the contamination of the A-class low.

We applied this two-steps classification to the 18 739 asteroids of the filtered DR3 dataset, and we obtained a total of 98 potential A-types. Of these objects, 77 had never been characterized with spectroscopy before the Gaia DR3. Considering only objects with a diameter above 2 km, we found a proportion of 0.51% of A-types in the Main Belt, which is more than three times the 0.16% found by DeMeo et al. (2019) [4].

Finally, the two steps classification method we developed gives satisfactory results for the classification of DR3 spectra and allowed to detect new potential A-type asteroids in the Main Belt. It appears that the amount of purely olivine-rich asteroids in the Main belt could be more than three times what previously thought, but this result has to be confirmed by NIR spectroscopy.

Acknowledgements:

This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. The authors acknowledge financial support from CNES, the Observatoire de la Côte d'Azur and the ANR ORIGINS (ANR-18-CE31-0014). 

References:

[1] Gaia Collaboration, Galluccio, L. et al. (2023) "Gaia Data Release 3. Reflectance spectra of Solar System small bodies." In: Astronomy & Astrophysics 674, A35. doi: 10.1051/0004-6361/202243791.
[2] Chapman, C. R. (1986). In: Proceedings of the NASA and CNR, International Workshop on Catastrophic Disruption of Asteroids and Satellites, 103–114.
[3] Burbine, T.H., Meibom, A., Binzel, R.P., 1996. "Mantle material in the main belt: Battered to bits?" Meteoritics and Planetary Science 31, 607–620. doi: 10.1111/j.1945-5100.1996.tb02033.x.
[4] DeMeo, Francesca E., David Polishook, Benoît Carry, Brian J. Burt, Henry H. Hsieh, Richard P. Binzel, Nicholas A. Moskovitz, and Thomas H. Burbine (Apr. 2019). "Olivine-dominated A-type asteroids in the main belt: Distribution, abundance and relation to families." In: Icarus 322, pp. 13–30. doi: 10.1016/j.icarus.2018.12.016.
[5] Neumann,W., D. Breuer, and T. Spohn (July 2012). "Differentiation and core formation in accreting planetesimals." In: Astronomy & Astrophysics 543, A141. doi: 10.1051/0004-6361/201219157.
[6] Popescu, M., M. Birlan, and D. A. Nedelcu (Aug. 2012). "Modeling of asteroid spectra - M4AST." In: Astronomy & Astrophysics 544, A130. doi: 10.1051/0004-6361/201219584.
[7] Bus, Schelte J. and Richard P. Binzel (July 2002a). “Phase II of the Small Main-Belt Asteroid Spectroscopic Survey. A Feature-Based Taxonomy.” In: Icarus 158.1, pp. 146–177. doi: 10.1006/icar.2002.6856.