An Automated Radiative Transfer and Geophysical Inversion Framework for Estimating Asteroid Macro-porosity and Internal Differentiation

Rationale and Objective: Resolving the internal architecture of asteroids provides critical insights into the early solar system's collisional and thermal evolution. Cross-referencing spectroscopic surface mineralogy with bulk densities derived from orbital tracking or radar astrometry isolates macro-porosity and uncovers internal structural profiles. These constraints directly impact planetary defense modeling, resource assessment, and asteroid mining frameworks. This preprint presents an automated geophysical inversion pipeline to extract pore-free regolith grain densities (rho_grain) from visible and near-infrared (VNIR) spectra, mapping internal composition profiles across varied datasets without relying on arbitrary geometric scaling factors. Novelty and Method: The method analytically inverts Hapke's bidirectional reflectance equations to convert raw observations into linearized, directionless single-scattering albedo vectors. To preserve spectral integrity, raw data is interpolated onto a uniform 500-point grid (0.3 to 4.0 microns) using shape-preserving cubic splines, avoiding channel down-sampling or artificial thermal clipping, before emissions are subtracted via an adaptive Near-Earth Asteroid Thermal Model (NEATM). Compositional unmixing relies on a Bayesian Maximum A Posteriori (MAP) optimization loop. High collinearity among mineral endmembers is mitigated by an Active Sub-Matrix Partitioning protocol, stabilized with trace-conditioned Tikhonov regularization and taxonomic priors. Linear mass-balance summation then yields the final grain densities alongside rigorous analytical uncertainty propagation. Results and Contribution: Validation spanned 29 benchmark targets: 10 laboratory meteorites, 14 planetary targets with spacecraft telemetry (Dawn, NEAR Shoemaker, OSIRIS-REx, and Hayabusa 1/2), and several synthetic instrument-bias matrices. For (4) Vesta, processing Dawn orbital spectrograph data yielded a surface grain density of rho_grain = 3.250 +/- 0.120 g/cm^3. Comparing this to Vesta's higher global bulk density (rho_bulk = 3.46 +/- 0.01 g/cm^3) establishes a positive density differential. A multi-phase mass balance loop leveraged this discrepancy to simultaneously isolate an internal Core Mass Fraction of 17.83% and a structural macro-porosity of 16.65%, aligning cleanly with independent gravity science models from the mission. Conclusion: The code handles all non-degenerate target arrays, while automated quality assurance filters intercept featureless M- and D-type spectra to avert rank-deficient optimization failures. This framework provides a reproducible, open-source method for inferring internal structures and density differentials directly from disk-integrated photometry. It establishes the mathematical basis necessary to scale automated geophysical mapping to population-wide minor-body surveys.