A Two‐Step Artificial Intelligence Approach for Correcting Space Weathering Effects in the Lunar Reconnaissance Orbiter Diviner Christiansen Feature Image

Abstract Space weathering substantially distorts the Christiansen feature (CF) observed by Lunar Reconnaissance Orbiter (LRO) Diviner thermal infrared radiometer, obscuring the intrinsic compositional and thermophysical signals of lunar surface silicate minerals. In this study, we develop a two‐step correction framework designed to remove space weathering effects from a previously topographically corrected LRO Diviner CF map. The method integrates Kaguya 750 nm reflectance, optical maturity (OMAT), FeO, H‐parameter, and npFe 0 parameters to model nonlinear space weathering relationships across immature, moderately mature, and mature CF pixels. The corrected CF values exhibited reduced dependence on npFe 0 abundances, enhanced correlation with bulk FeO abundances, and improved internal consistency within compositionally homogeneous regions. Comparisons with correction results based on Kaguya OMAT and FeO scale factors indicate that the proposed method more effectively suppresses space weathering induced variability while preserving compositionally diagnostic CF signatures. Persistent low CF anomalies in highland regions suggest additional controls beyond npFe 0 accumulation, potentially related to basaltic dust contamination, subsurface compositional heterogeneity, and silicate amorphization processes. The resulting CF product offers a refined thermal infrared perspective on lunar surface composition, indicating clearer basaltic distributions, particularly within the South Pole Aitken basin.