Feasibility study of rock identification at the surface of Mars by remote laser-induced breakdown spectroscopy and three chemometric methods

The ChemCam instrument that will equip the Mars Science Laboratory ( MSL) rover uses laser-induced breakdown spectroscopy ( LIBS) to remotely identify Martian rocks in the proximity of the rover, and to quantitatively assess their composition. Sample identification is then the first step of the chemical analysis, as a decision aid for monitoring the rover and/ or before a subsequent composition measurement. In this paper we analyze our experimental spectra obtained in the laboratory by three chemometric methods - principal components analysis ( PCA), soft independent modeling of class analogy ( SIMCA) and partial least- squares discriminant analysis ( PLS- DA) - to investigate the feasibility of rocks classification by remote LIBS. If PCA is very interesting for data visualization, SIMCA and PLS- DA enable the making of automatic predictions. We show that SIMCA is less sensitive than PLS- DA, but also more robust when it encounters spectra of an unknown rock. The instrument accuracy during MSL operations will benefit from a combination of the two approaches.