Comparison of two partial least squares-discriminant analysis algorithms for identifying geological samples with the ChemCam laser-induced breakdown spectroscopy instrument.

ChemCam, a laser-induced breakdown spectroscopy (LIBS) instrument on the Mars Science Laboratory rover, will analyze the chemistry of the martian surface beginning in 2012. Prior to integration on the rover, the ChemCam instrument collected data on a variety of rock types to provide a training set for analysis of data from Mars. Models based on calibration data can be used to classify rocks via multivariate statistical techniques such as partial least squares-discriminant analysis (PLS-DA). In this study, we employ a version of PLS-DA in which modeling is applied in a defined classification flow to a variety of geological materials and compare the results with the traditional PLS-DA technique. Results show that the modified algorithm is more effective at classifying samples.