Probabilistic Spectral Reconstruction of Trans-Neptunian Objects from Sparse Photometry: A Framework for Taxonomy, Survey Optimization, and Outlier Detection

Abstract

Near-infrared (near-IR) spectroscopy provides critical constraints on the surface composition of trans-Neptunian objects (TNOs), but spectroscopic observations remain limited compared to broadband photometry. We develop a probabilistic latent-space framework to quantify how much spectral information is retained in sparse photometric measurements. Using a principal component representation trained on a sample of near-IR spectra, we model the spectral manifold of TNOs and perform Bayesian inference in this reduced space to reconstruct full spectra from photometry while propagating uncertainties. Leave-one-out cross-validation demonstrates that the dominant modes of spectral variability are low-dimensional: 4 to 5 principal components capture the structure relevant for taxonomic classification, while 8-10 components improve spectral reconstruction fidelity and uncertainty calibration. For most objects, the reconstructed spectra achieve empirical credible-interval coverage of 95 percent across wavelength. This suggests the diversity of near-IR spectral shapes is governed by structured, correlated surface processes rather than stochastic variation. Practically, we apply this framework to survey optimization, quantifying the information content of JWST/NIRCam filters to identify optimal configurations (e.g., F090W, F115W, F410M, F460M) for TNO taxonomy. Additionally, we demonstrate the pipeline's capability to detect and reconstruct rare spectral types, such as the peculiar Neptune Trojans 2006 RJ103 and 2011 SO277, by allowing constraining photometry to select low-probability intermediate models from the continuous topological manifold. Ultimately, this framework bridges the gap between sparse photometry and spectroscopy, providing a statistically rigorous tool to map the compositional structure of minor planets in upcoming large-scale surveys.

Accepted for publication in The Astronomical Journal. 25 pages, 13 figures. The generative spectral reconstruction example is publicly available on GitHub at https://github.com/sevenlin123/spectra_reconstructor

References in corpus (18)

• XGBoost: A Scalable Tree Boosting System
• SciPy 1.0--Fundamental Algorithms for Scientific Computing in Python
• Astropy: A Community Python Package for Astronomy
• The Astropy Project: Building an inclusive, open-science project and status of the v2.0 core package
• The Astropy Project: Sustaining and Growing a Community-oriented Open-source Project and the Latest Major Release (v5.0) of the Core Package
• Solar System evolution from compositional mapping of the asteroid belt
• The taxonomic distribution of asteroids from multi-filter all-sky photometric surveys
• Fast likelihood-free cosmology with neural density estimators and active learning
• Massive optimal data compression and density estimation for scalable, likelihood-free inference in cosmology
• Likelihood-free inference with neural compression of DES SV weak lensing map statistics
• First Cosmology Results Using Type Ia Supernovae From the Dark Energy Survey: Photometric Pipeline and Light Curve Data Release
• The Bimodal Colors of Centaurs and Small Kuiper Belt Objects
• Reanalyzing the visible colors of Centaurs and KBOs: what is there and what we might be missing
• Col-OSSOS: z Band Photometry Reveals Three Distinct TNO Surface Types
• Col-OSSOS: The Colours of the Outer Solar System Origins Survey
• OSSOS XX: The Meaning of Kuiper Belt Colors
• JWST Spectrophotometry of the Small Satellites of Uranus and Neptune
• JWST Spectroscopy of a Blue Binary Cold Classical Kuiper Belt Object