Internal structure inference of lunar lave tubes with the MCMC approach

For decades searching for the safe habitation site on the moon have piqued the interest of researchers. Safe habitats should be resilient under different hazards including radiation, temperature fluctuation, and meteorite impacts. The natural tunnels on the Moon called “lava tubes” are considered as good candidates for permanent human shelters. Their hard basaltic roof may provide increased safety from different hazards, and their width can be as large as ~1-2 km, promising the underground storage with little construction effort. Lying beneath the surface, the lava tubes cannot be detected by surface imagery or altimetry, although close association with known rilles, skylights, or rimless pits may add credibility to a potential candidate. However, linear lower magnitude of gravity data indicates the subsurface density deficits and provide another piece of evidence supporting the existence of lava tubes. In this work, we try to find the average Bouguer gravity profiles across the lava tubes from GRAIL mission, for the physical properties of their subsurface density anomalies. Traditional inversion methods based on gradient theory usually leads to the best model with minimum-misfit objective function. However, the results are easily influenced by the initial model and a variety of weighting factors. In this study, we adopt the popular Markov Chain Monte Carlo (MCMC) statistical method to detect possible subsurface density deficits. Different weighting factors are set as hyper-parameters and are quantitatively evaluated for density anomaly influence. Uncertainty analysis is carried out to infer the most likely lave tubes.