Gaia DR3 Asteroid Reflectance Spectra with Probabilistic Classification

A supervised, probabilistic taxonomic classification for the full Gaia Data Release 3 asteroid sample. The dataset models spectral class likelihoods using multivariate kernel density estimation on principal components derived from Gaia DR3 reflectance spectra. It was created by Marco Delbo and published on the figshare platform in April 2026.

Use Cases

• Identify family interlopers within six major asteroid collisional families using the Gaia-derived spectral classes.
• Study the compositional structure of the main asteroid belt by analyzing the distribution of major spectral complexes (C, S, X) and end-member classes (B, D, A, L, K, V).
• Link asteroid populations to meteorite groups by leveraging the fully probabilistic taxonomic classification for the full Gaia DR3 sample.
• Validate future Gaia data releases (e.g., DR4) by applying the established methodological framework for spectral classification.
• Detect olivine-rich A type material within specific families like Flora and Eunomia based on the spectral classification results.

Strengths

• Probabilistic classification provides likelihoods and a quantitative measure of confidence for each spectral class.
• Validation against objects with known spectral classes demonstrated good performance for distinctive types like S-complex, D, V, and A.
• The dataset offers a resource for the full Gaia DR3 asteroid sample, establishing a framework for future releases.

Limitations

• Classification performance is limited for spectrally continuous classes (B-C-complex, K-L-S-complex, X-complex) due to mixing from Gaia's limited wavelength range.
• The dataset is relatively small at 12.0 MB, which may limit the complexity of analyses compared to larger spectroscopic archives.
• The exact number of rows (asteroids) is unspecified, making it difficult to assess sample size for statistical modeling.

Provenance

Source

Gaia Data Release 3 (DR3) asteroid reflectance spectra, with a reference set from the literature.

Collection Method

Supervised classification using principal-component representation and multivariate kernel density estimation on reference spectra.

Time Range

Covers the observational period of Gaia DR3.

Freshness

Last updated April 2026, with relevance for validating the upcoming Gaia DR4 expected by the end of 2026.

Geography

Observations of asteroids, primarily in the main asteroid belt.