Spectral Discriminant Analysis Of Martian Simulated Minerals And Brines

Spectroscopy is an important method to identify the composition,abundance and distribution characteristics of Martian surface materials in molecular level.For planetary exploration,multi-load collaborative analysis is essential to improve the detection accuracy of unknown targets and verify their self-consistency.Raman spectroscopy is a powerful tool with sharp peaks,high resolution,and ability to detect micro-regions.Near-infrared spectroscopy can detect various minerals and easily record the signal of minerals,especially H2O and OH.The type of minerals and its distribution on Martian surface can provide clues to the evolution of the Martian environment,and help us to understand the origin of Mars and the history of geological evolution.Brine on Martian surface can provide the information on the evolution of climate/hydrology and water cycles.It is of great significance for identification of the types of minerals/rock,brines and its soluble types by combining with Raman and NIR spectroscopy.Based on the fact that Raman and NIR spectroscopy will be carried by ESA ExoMars and NASA SurperCam in 2020,it will have important scientific significance for combining the Raman with NIR spectral data for Martian minerals identification,and using Raman spectroscopy to analysis Martian surface brines and other water-related substances.Firstly,this paper introduced the characteristics of Mars and its environment,the exploration history of Mars,exploration instruments for minerals and the distribution of Martian minerals.And then introduced the principle of vibrational spectroscopy(i.e.Raman and near-infrared spectroscopy)and the characteristics of vibrational spectra of Martian minerals.Based on Raman and Vis-NIR spectroscopy,the two detailed studies were carried out as following:1)Discriminant analysis of Martian Surface Minerals Based on Spectral Data Fusion MethodIn the study,25 simulated Martian samples(i.e.silicate,carbonate and sulfate minerals)were prepared.The Raman and Vis-NIR spectra of these samples were obtained under laboratory conditions and some data pre-processing methods such as baseline correction,Savitzky-Golay smoothing,standard normal variate(SNV)scaling were used to produce a high-quality representation spectra.Then based on that established a Raman and Vis-NIR spectra database of Martian silicate,carbonate,and sulfate minerals.Firstly,the information-rich spectral bands with higher signal-to-noise ratio and less overlapping were selected(i.e.,Vis-NIR:430-2430 nm;Raman:130-1100 cm-1)during the clustering analysis.Secondly,Soft Independent Method of Class Analogy(SIMCA)and Principal Component Analysis-K-Nearest Neighbor(PCA-KNN)were respectively built based on selected Vis-NIR,Raman and two kinds of their fusion data(i.e.,coaddition fusion and concatenation fusion).The accuracy of SIMCA model was enhanced from 82.7%(Vis-NIR)and 82.7%(Raman)to 93.1%(coaddition fusion)and 96.5%(concatenation fusion).The accuracy of PCA-KNN model was improved from 72.4%%(Vis-NIR)and 72.4%%(Raman)to 79.3%(coaddition fusion)and 91.2%(concatenation fusion).Therefore,the mineral clustering results show that the spectral fusion method can exert the data advantages of Vis-NIR and Raman respectively,and can greatly improve the clustering model's prediction accuracy of Martian minerals.2)Quantitative analysis of Martian surface brines based on Raman spectroscopyIn this study,72 kinds of Martian simulated brine samples of perchlorate,chlorate,sulfate,carbonate and nitrate with different concentrations were prepared and their Raman spectra were obtained under laboratory conditions.And then built a Raman database of Martian simulated brines(i.e.,perchlorate,chlorate,sulfate,carbonate and nitrate)through spectral pre-processing of baseline correction,Savitzky-Golay smoothing and Gauss fitting.The quantitative analysis models were established based on the normalized Raman peak area of lsymmetric stretching vibration mode of anionic groups in brine solutions and brine's concentration using the Raman self-referencing technique by using the strong and broad characteristic envelope peak of water at 2800-3800 cm-1 as the internal standard peak.The results show that there is a good linear relationship between v1normalized Raman peak area of anionic groups and brine concentration,and the linear correlation coefficient R2 of the quantitative curves are all better than 99%,this means that it is feasible to use in-situ Raman spectroscopy to identify and quantitatively analyze of Martian surface brines.Systematic experimental investigations have shown that the quantitative analysis curves established under the laboratory environment were independent of the Raman test conditions(laser energy,focusing distance),cation species,etc.This means that the quantitative analysis model of simulated Martian brines established in this study is very robust and can be applied to the remote sensing and in-situ missions.Finally,the detection limit of brine solutions under the experimental system conditions was studied,which also has great significance for the Mars 2020 exploration missions.