Mainstream Lunar Regolith Simulants Incompletely Represent Lunar Bulk Chemistry─A Subcompositional Analysis of Lunar Prospector GRS Data

Regolith simulants are substrates designed to mimic the surface regolith of planetary bodies and are commonly used in the design and testing of in situ resource utilization (ISRU) technologies for engineering, construction, precious materials extraction, and life support. While numerous regolith simulants have been designed since the Apollo missions and methods of quantifying simulant accuracy have been developed, no analysis has yet been done to make a global map of the most chemically accurate simulant for each region of the moon. By conducting an Aitchison nearest neighbor analysis of compositional bulk chemical data of commonly used simulants and Lunar Prospector Gamma Ray Spectrometer data, 2° by 2° tiles of the lunar surface were assigned the chemically closest simulant. A Monte Carlo analysis was further conducted to determine the robustness of each simulant assignment. With Lunar Prospector points having an average closest simulant assignment Monte Carlo probability of 0.44 and a standard deviation of 0.21, analyzed simulants had generally low success in completely representing the bulk chemistry of the lunar surface. Highland simulant NU-LHT-2 M and mare simulant LMS-1/LMS-2 were notable exceptions, having both wide geographic spread and comparatively high robustness values. A confusion matrix analysis revealed that these simulants were compositionally interchangeable even for tiles nominally represented by other simulants. Nonetheless, the development of broad highland and mare simulants has left much of the lunar surface, such as regions with low and very high Al2O3, very high FeO, high CaO, and high TiO2, closely unrepresented by any simulant. Bulk chemistry remote sensing data should be considered as a factor for simulant development in addition to traditional data sources such as returned samples, and the compositional data approach should be used in conjunction with classical metrics such as the Figure of Merit (FoM) to provide a comprehensive assessment of simulant accuracy. Moreover, simulants should be focused on specific regions of interest to better test ISRU technologies since the use of nonrepresentative or generalized simulants will skew the performance of such technologies.