Statistical exploration and volume reduction of planetary remote sensing spectral data

Modern planetary scientific experiments are providing a huge amount of information that requires extensive analysis and evaluation in order to understand its scientific content. In this work we evaluate the ability of a previously developed and evaluated technique to explore a large data set in order to identify the cluster representatives of the spectral variability of a selected Martian region and obtain a reduction of the original data volume ∼2 × 104. We also demonstrate that the cluster centroids are able to quantitatively reproduce previously interpreted atmospheric, topographic, and compositional properties. The results build additional confidence in the application of the statistical techniques.