A geophysical crustal model of central Italy for geoneutrino predictions in Borexino

Predicting the geoneutrino signal at underground detectors requires a local crustal reference whose precision matches the experimental uncertainty. The current Borexino reference is 15 years old and deterministic. We replace it with GUESS, a probabilistic 3-D model of the Gran Sasso region, built by Bayesian inversion of multi-source geophysical data: ground and satellite gravity, deep exploration wells, seismic profiles, receiver-function Moho constraints, and high-resolution topography are fused into a posterior distribution of density and lithological label per voxel. The model partitions the volume into six lithologically distinct units, each with quantified uncertainty on volume, mass and average density, and is validated against three independent in situ density benchmarks. The structural uncertainty on the predicted geoneutrino signal of the major units is constrained to 1%–3%, an order of magnitude below the uncertainty on uranium and thorium abundances. The dominant source of error in the absolute crustal geoneutrino prediction therefore shifts from crustal geometry to elemental composition, defining the next priority for the Borexino determination of the radiogenic heat budget.