Abstract:

 Lunar dust contaminant is one of primarily influential factors on the working status of insitu spectrometer on lunar surface. In this work, the influences on the accuracy of quantitative analysis and the NIR reflectance fingerprints of typical lunar mineral mixtures (i.e., orthopyroxene, labradorite, ilmenite and olivine) were analyzed by using different contamination levels of reflectance standards (i.e. pollutionfree, slight pollution, heavy pollution). Significant differences in fingerprints were observed from reflectance spectra of 15 samples which were recorded using the pollutionfree and various pollution levels of reflectance standards, respectively. More differences were found with higher pollution levels of reflectance standards. The prediction precisions were decreased by using PLS model based on pollutionfree samples to predict the contaminated ones. Instead, the prediction precisions were increased by using the PLS model based on contaminated samples to predict the ones at same pollution level. This work shows that the contamination of reflectance standard is the key influential factor on the prediction precision of quantitative analysis. Our results have scientific significance for insitu planetary load design, data rectification and assessment of model predictive ability.

Key words: near infrared reflectance spectrometry, reflectance standard contamination, lunar minerals, partial least squares regression, quantitative analysis