The ability to measure TiO₂ remotely is important for mapping the composition lunar basalt flows globally, and for placing lunar samples into a regional and global geologic context. Comparing Clementine UVVIS-ratio (415/750 nm) with Lunar Prospector derived TiO₂ data, however, yields a less than ideal correlation, which would suggest that either the UVVIS ratio has poor predictive capabilities with respect to TiO₂ composition or poor accuracy of the Lunar Prospector TiO₂ data. Established uncertainties of the Clementine UVVIS data are approximately 1%, while the reported relative errors for Lunar Prospector neutron spectrometer data are on the order of 5%. Thus, we investigate the possibility of whether the greater uncertainty of the Lunar Prospector neutron data could cause the poor correlation between the two data sets. The sensitivity of the TiO₂-UVVIS correlation to data accuracy was measured by adding randomly-distributed noise to the Clementine UVVIS data, and then comparing this modified Clementine data with the “noiseless” Clementine data. The comparison was then evaluated for the level of noise needed to produce a similar amount of scatter observed in the Lunar Prospector TiO₂ and Clementine UVVIS-ratio trend. The results of this study indicate that Lunar Prospector would have to possess significantly more than 5% uncertainty to match the observed poor correlation between Lunar Prospector and Clementine data sets. On this basis, we concluded that algorithms that depend solely upon correlations between UV and visible spectral parameters and TiO₂ concentration have inherently poor predicting power.

Date Published: 30 December 2004
PDF: 8 pages
Proc. SPIE 5660, Instruments, Science, and Methods for Geospace and Planetary Remote Sensing, (30 December 2004); doi: 10.1117/12.579003

Published in SPIE Proceedings Vol. 5660:
Instruments, Science, and Methods for Geospace and Planetary Remote Sensing
Carl A. Nardell; Paul G. Lucey; Jeng-Hwa Yee; James B. Garvin, Editor(s)