The lunar surface exhibits noticeable variations in photometric properties across different regions, influenced by their specific compositional and physical characteristics. These variations are particularly pronounced at the millimeter to centimeter scale. In this study, we propose a feature subspace-based sample selection method to identify Chang'E-4 Visible Near-infrared Imaging Spectrometer (CE-4 VNIS) in situ observations that share similar photometric properties at the centimeter scale. By constructing a feature subspace through principal component transformation, the CE-4 VNIS observation sites with comparable surface photometric properties are able to be selected from a series of observations. The selection results demonstrate that the majority of observation sites are covered by the lunar regolith, with a few exceptions consisting of rocks, soil bulk, shadows, and rough surfaces. These observations should be excluded from the photometric correction process during phase function fitting. Analyzing the photometric performance of the VNIS data with lunar regolith reveals that reflectance decreases as the phase angle increases below 80°, while above 80°, reflectance increases with increasing phase angle. This phenomenon is likely attributed to stronger forward scattering. Thus, a second-order polynomial is employed to fit the phase function for VNIS data. Consequently, a photometric model is developed for lunar regolith observed by CE-4 VNIS, incorporating the fitted phase function. Experimental results demonstrate reduced differences among observations with varying phase angles, and the corrected VNIS spectra exhibit a phase-reddening effect. These findings validate the effectiveness of the proposed model. A comparison between the corrected data with and without sample selection reveals a 33 % decrease in mean standard deviation for VIS/NIR bands and a 19 % decrease for SWIR bands in the corrected spectra after sample selection. These results indicate the potential applicability of our method for future VNIS observations, enabling the acquisition of photometrically consistent data.
Read full abstract