Skip to main content
SearchLoginLogin or Signup

Asteroid characteristics and classification by combining Gaia photometry and spectroscopy

Presentation #503.06 in the session Asteroids: Main Belt (Oral Presentation)

Published onOct 23, 2023
Asteroid characteristics and classification by combining Gaia photometry and spectroscopy

Photometric inversion has been extensively used to infer the spin and shape properties of asteroids from dense-in-time ground-based lightcurves. The high-precision sparse-in-time photometry provided by the Gaia space telescope has enabled studies of asteroid phase functions. Gaia Data Release 3 (DR3) contains photometric observations of more than 150,000 asteroids (Tanga et al., Astron. & Astrophys. 674, A12, 2023). Here we consider the rotation periods, pole orientations, shapes, and photometric slopes for over 22,000 asteroids that have at least 25 Gaia observations, using ellipsoid and general convex shape models. We characterize the solutions by using a Markov chain Monte Carlo approach (Muinonen et al., Astron. & Astrophys. 642, A138, 2020).

Generally, shapes are not well constrained along the shortest c axis, arising from viewing aspects of the observations. We investigate the distributions of b/a axial ratios for different sizes and rotation periods. Pole latitudes are more robustly constrained than longitudes, but both are susceptible to having erroneous mirror solutions. We find a small population of super-slow rotators (rotation periods exceeding 1000 hr) at small sizes (H ~ 14 mag and smaller). Further results will be included in manuscripts in preparation (Cellino et al., MacLennan et al.).

For maximum accuracy of asteroid classification, we complement the photometric slopes with DR3 spectroscopy (Gaia Collaboration, Astron. & Astrophys. 674, A35, 2023). The full spectroscopic dataset includes 57490 asteroids. After selecting asteroids with both spectra and photometric slopes and after filtering for the highest-quality spectra, we are left with 5783 asteroids. To avoid challenges from merging data from the Gaia Red Photometer and Blue Photometer, we divide the spectra into independent red and blue parts. Removing seven wavelengths giving rise to anomalous behavior, we are left with the wavelengths of 638, 682, 726, and 770 nm for the red part and 418, 462, 506, 550, and 594 nm for the blue part. The blue and red parts are normalized to unity at 550 and 682 nm, respectively. We then standardize the spectroscopic and photometric parts, and apply principal component analysis to the data. After constructing a feed-forward, fully connected neural network, we find that the most reasonable classification (maximum number of classes with high overall accuracy) comes from classifying the asteroids into the taxonomic classes C, S, X, and D. The accuracy of the classification is about 82%. Finally, classification with only the spectroscopic parts would decrease the accuracy.

No comments here