Prognosis of TiO2 abundance in lunar soil using a non-linear analysis of Clementine and LSCC data

We suggest a technique to determine the chemical and mineral composition of the lunar surface using artificial neural networks (ANNs). We demonstrate this powerful non-linear approach for prognosis of TiO2 abundance using Clementine UV–VIS mosaics and Lunar Soil Characterization Consortium data. The ANN technique allows one to study correlations between spectral characteristics of lunar soils and composition parameters without any restrictions on the character of these correlations. The advantage of this method in comparison with the traditional linear regression method and the Lucey et al. approaches is shown. The results obtained could be useful for the strategy of analyzing lunar data that will be acquired in incoming lunar missions especially in case of the Chandrayaan-1 and Lunar Reconnaissance Orbiter missions.