Machine learning thermophysical analysis of asteroid (101955) Bennu

The general lack of fine regolith on asteroid (101955) Bennu was an important discovery of the NASA OSIRIS-REx mission [1]. Indeed, contrary to the expectations from ground-based observations [2], OSIRIS-REx revealed that Bennu had rough surfaces and lacked fine-regolith ponds [1].

The observations by the OSIRIS-REx mission — ranging from high-resolution images to laser-altimetry — offer a precious opportunity to investigate with high spatial resolution the surface of Bennu and study the problem of the general lack of fine regolith. To this end, we apply a two-component thermophysical model [3] to deduce the properties of Bennu’s regolith from data by the OSIRIS-REx Thermal Emission Spectrometer (OTES, [4]). Our approach uses machine learning to distinguish the contribution of two terrains to the total emitted flux: a geological unit composed by particles larger than the diurnal thermal skin depth (a few centimeters on Bennu, [5]), and another unit of smaller particles which corresponds to the fine regolith and could result from the comminution of the former unit.

Machine learning allows efficiently exploring the large parameter space of thermal inertia and relative areal abundance of these two terrains and fit them to the data. We will present the results from our analysis and discuss them in the context of independent evidence from Bennu, asteroid (162173) Ryugu, and other asteroids.

[1] Lauretta, D. et al. Nature 568, 55–60 (2019). [2] Emery, J. et al. Icarus 234, 17–35 (2014). [3] Cambioni, S., et al. Icarus 325, 16–30 (2019). [4] Christensen, P. R. et al. Space Science Reviews 214, 87 (2018). [5] Rozitis, B. et al. Science Advances 6, 41 (2020).