Lunar Soil Characterization Consortium Research Articles

Feature Engineering for estimating the maturity of lunar soils from spectroscopic data

Existing algorithms for estimating the maturity of lunar soils are not optimized for data from any of the orbital sensors which are currently active. This paper addresses this issue by proposing an algorithm for estimating soil maturity (IS/FeO) using spectroscopic data at the spectral resolution of the Moon Mineralogy Mapper (M3). As part of this method, four key spectral parameters for estimating IS/FeO are identified and used to train a Support Vector Regression (SVR) model. The physical significance of each parameter is discussed, and the equation of the predictive hyperplane is provided for increased transparency. The proposed method outperforms state-of-the-art algorithms and returns a coefficient of determination (R2) of 0.92 over the Lunar Soil Characterization Consortium (LSCC) dataset.

Chang’E-5 samples—New “ground truth” for the quantitative inversion of lunar remote sensing

<p indent="0mm">The Chang’E-5 (CE-5) mission successfully returned <sc>1731 g</sc> of lunar soil samples from the northeast of the Oceanus Procellarum. The newly returned lunar soil can provide a new ground truth for the quantitative inversion of lunar remote sensing data. In this study, the mineral and chemical compositions (FeO, TiO₂, Th) of the landing region were obtained from the remote sensing data and compared with the laboratory analysis results of the CE-5 lunar soil samples. The differences between them (especially the olivine content) indicate that there is still room for improvement in the methods for retrieving the elemental and mineral abundances using remote sensing data. Through a systematic comparative analysis, we discovered that: (1) the geochemical characteristics of high FeO, medium TiO₂, and abundant Th in the CE-5 lunar soil are significantly different from the existing Apollo lunar soil, which can provide unique ground truth for remote sensing research; (2) The CE-5 lunar soil represents mature lunar soil developed on the young basalt unit on the moon, with a small proportion of distantly sourced ejecta, representing the mare basalt composition characteristics of the landing region; (3) The CE-5 lunar soil is very fine, which meets the sample standard <sc>(&lt;45 μm)</sc> of the Lunar Soil Characterization Consortium for providing ground truth. Notably, the CE-5 landing region also exhibits unique characteristics: (1) it is located in the mid-latitude region on the western hemisphere of the moon, different from the low-latitude region where the United States and the Soviet Union sample return missions landed; (2) it is located on the late-stage lunar basalt unit; it is different from the existing spectral calibration area. Therefore, the construction of the CE-5 landing area as a new remote sensing calibration site based on the measured data of its lunar soil can significantly expand the geographical scope and global representation of the lunar calibration sites. Based on the results of the remote sensing and sample studies, we delineate the calibration area in the unit where the CE-5 landing site is located (Em4), thus providing a reference for establishing a remote sensing calibration area in the future.

An improved method combining Fisher transformation and multiple endmember spectral mixture analysis for lunar mineral abundance quantification using spectral data

Mineral abundance quantification of lunar surface materials using remotely acquired spectra can be achieved with spectral unmixing methods, but this has remained a challenging problem. The existing unmixing methods neglect the existence of widespread glass material and the spectral similarity of different minerals. To solve this problem, this study proposes a method to calculate the main mineral abundances in lunar regolith based on reflectance spectra by utilizing Fisher transformation combined with multiple endmember spectral mixture analysis (MESMA). A set of spectra of lunar main constituents, including plagioclase, pyroxene, ilmenite, olivine and glass material, were selected from the RELAB spectral database and used as endmembers. The endmember spectra were utilized to compute the projection vectors, with which the spectral characteristics can be transformed into low-dimensional Fisher space in which intraclass spectral variability can be minimized while the interclass variability of different mineral spectra can be maximized. The endmember spectra and the lunar soil spectra in the Lunar Soil Characterization Consortium (LSCC) dataset were both transformed into Fisher space, in which MESMA was executed to acquire the fraction of each endmember. MESMA allows for comprehensive endmember expressions by representing a type of mineral using multiple spectra rather than a single endmember. Compared with laboratory mineral measurements, this approach estimated the mineral abundances with determination coefficient R2 values equal to 0.88 and 0.65 in the lunar mare and highland regions, respectively. After validation with laboratory data, the method was further applied at the regional scale where Apollo 12 and Apollo 16 landed by using M3 images, and the estimated accuracy of different minerals was similar to that of the laboratory spectra.

Mineral Content Estimation of Lunar Soil of Lunar Highland and Lunar Mare Based on Diagnostic Spectral Characteristic and Partial Least Squares Method

Extraction of mineral and rock information of lunar regolith is of far-reaching significance to the study of material composition, geological structure and historical evolution of lunar regolith. Visible and near-infrared spectra can reflect mineral composition information, and can be used to extract mineral composition and distribution characteristics of lunar regolith. In this paper, the LSCC (Lunar Soil Characterization Consortium) data of lunar regolith is taken as the research object. The partial least squares (PLS) regression model is used to estimate the spectra of lunar regolith measured in RELAB laboratory of Brown University. The mineral contents of plagioclase, pyroxene, olivine, ilmenite, agglutinate and volcanic glass in lunar regolith have been optimized and retrieved. The LSCC spectra of lunar regolith have been pre-processed by multivariate scattering correction (MSC), which highlight the spectral features of lunar regolith. The optimal number of principal components has been selected by cross-validation test. The PLS regression have been established for samples from lunar highland and lunar mare respectively. Two-thirds of samples have been randomly selected as experimental group to establish the prediction relationship between the spectra of lunar regolith and mineral content. The remaining one-third of samples have been used as verification group to further validate the prediction relationship. The results show that the partial least squares regression model has high accuracy and good stability. It is of theoretical and practical significance to optimize the inversion of mineral content in lunar regolith using spectral data of lunar regolith.

Two Simulated Spectral Databases of Lunar Regolith: Method, Validation, and Application

Our simulated lunar regolith spectra database, based on the Hapke AMSA radiative transfer model (RTM), is a large supplement to the limited number of lunar spectra data. By analyzing the multiple solutions and applicable scopes of the Hapke model by means of Newton interpolation and the least square optimization method, an improved method was found for the simulation of spectra, but it remained challenging to use to invert mineral abundance. Then, we simulated the spectra, mineral abundance, particle size and maturity of 57 mare and highland samples of the Lunar Soil Characterization Consortium (LSCC) in size groups of 10 µm, 10–20 µm and 20–45 µm. The simulated and measured spectra fit well with each other, with correlation coefficients greater than 0.99 and root mean square errors at a magnitude of 10-3. The parameters of mineral abundance, particle size and maturity are highly consistent with the measured values. Having confirmed the reliability of our simulation method, we analyzed the mechanism, reliability and applicability of the “spectral characteristic angle parameter” proposed by Lucey using the simulated spectral data of lunar regolith. Lucey’s method is only suitable for macro analysis of the entire moon, and the error is large when it is used for areas with high abundance of forsterite or ilmenite. In the spectral simulation of lunar regolith, olivine was subdivided into forsterite and fayalite, and the two end-members were mixed to approximately estimate the effect of the chemical composition of olivine on the spectra, which has been confirmed to be feasible.

An Empirical Model to Estimate Abundance of Nanophase Metallic Iron (npFe0) in Lunar Soils

Lunar soils gradually become mature when they are exposed to a space environment, and nanophase metallic iron (npFe0) generates within them. npFe0 significantly changes the optical properties of lunar soils and affects the interpretation of the remotely sensed data of the lunar surface. In this study, a correlation analysis was conducted between npFe0 abundance and reflectance spectra at short wavelengths for lunar soil samples in four size groups based on their spectral and compositional data, collected by the Lunar Soil Characterization Consortium (LSCC). Results show that 540 nm single scattering albedo (SSA) of lunar soils correlates well with their corresponding npFe0 abundance for each size group of lunar soil samples. However, it is poorly correlated with npFe0 abundance when all size groups were considered because of the strong interference from grain size variation of lunar soils. To minimize the effect of grain size, the correlation of npFe0 abundance with the spectral ratio of 540 nm/810 nm SSA of all size groups for LSCC samples was calculated and results show that a higher correlation existed between them (R2 = 0.91). This ratio can serve as a simple empirical model for estimating npFe0 abundance in lunar soils. However, bias could be introduced to the estimation result when lunar soils possess a high content of agglutinitic glass and ilmenite. Our future work will focus on improving the model’s performance for these lunar soils.

Machine learning approaches for classifying lunar soils

We examine the ability of machine learning (ML) techniques to determine the physical and mineralogical properties of lunar soil using reflectance spectra. We use the Lunar Soil Characterization Consortium (LSCC) dataset to train and assess the predictive power of classification models based on their type (Mare soil and Highland soil), particle size, maturity, and the dominant type of pyroxene (High-Ca and Low-Ca). Nine ML algorithms including linear methods, non-linear methods, and rule-based methods (three from each) were selected, representing a range of characteristics such as simplicity, flexibility, computational complexity, and interpretability along with their ability to handle different types of data. Fifteen spectral parameters were initially introduced to the models as input features and a maximum of four features was selected as the best feature combinations to classify lunar soils based on their types, particle size, maturity, and the type of pyroxene. The Support Vector Machine with radial basis function (svmRadial) and the penalized logistic regression model (glmnet) performed well for all target variables with high accuracies. Band depths and Integrated band depths at 1 μm, 1.25 μm and 2 μm, band position of the 1 μm band, along with four band ratios (band tilt, band strength, band curvature, and olivine/pyroxene) are important features for classifying soil type, grain size, maturity, and type of pyroxene from reflectance spectra. This study shows that proper preprocessing and feature engineering techniques are crucial for high performance of the predictive models.

Modal analyses of lunar soils by quantitative X-ray diffraction analysis

We used X-Ray Diffraction (XRD) and Rietveld refinement to determine the modal mineralogy of 118 lunar regolith samples (<150 µm size fraction) from all landed Apollo missions. Data were calibrated with mineral mixtures and compared to results based on an X-ray digital imaging procedure for six soils obtained by the Lunar Soil Characterization Consortium. Agreement between XRD and digital imaging for all minerals detectable in the six soils is excellent (R2 = 0.953). XRD-based ternary plots (plagioclase-total pyroxene-olivine) vary from plagioclase-dominated (highlands as represented by Apollo 16 samples) to substantial mafic abundances at the mare sites. Olivine varies in relative abundance, with the Apollo 17 mare sites having the largest abundances. Olivine reaches 20 wt% at Apollo 17, but is a minor component at Apollo 14. The results agree with trends in mineral abundances obtained from reflectance spectroscopy for the Apollo sites. In a global context, however, the spectral data display a trend of increasing olivine at roughly constant pyroxene/plagioclase, reaching values of 40% olivine in the plagioclase-pyroxene-olivine ternary plot (e.g., Eratosthenian flows in Procellarum), indicating the presence of significant volumes of olivine-rich rock types on unsampled regions of the lunar surface.

Improved space weathering maps of the lunar surface through radiative transfer modeling of Kaguya multiband imager data

Space weathering is a continuous process operating on the surface of the Moon producing nanophase and microphase iron on the rims of mineral grains and within agglutinates. The spectral effects due to these nanophase and microphase iron particles in the lunar regolith are well characterized in the visible and near-infrared wavelengths. In this work, we used the Hapke radiative transfer technique to model multispectral data (415–1550 nm) from the Kaguya Multiband Imager to map the nanophase and microphase iron abundances across the lunar surface. In order to produce these maps, we developed a reflectance model for silicate host particles as a function of FeO content using magnetic and Lunar Soil Characterization Consortium (LSCC) data. We tested the accuracy of our radiative transfer technique and found that the model can predict the nanophase and microphase iron abundances within 0.1 wt% Fe and 0.6 wt% Fe, respectively. By using a FeO and ilmenite map and our relationship between host particle reflectance and FeO abundance, we produced a nanophase, microphase, and submicroscopic iron abundance map. We found the approximate saturation limit for nanophase and microphase iron with respect to FeO. In general, the nanophase and microphase iron saturation limit increases with FeO content, except in soils that have FeO content of 6–15 wt%, the nanophase iron saturation is nearly constant. We also observed that the microphase iron abundance saturates before the nanophase iron abundance. These maps could be used to determine relative age and maturity, but with limitations. Our analyses of these maps showed that (1) swirls contain low abundances of nanophase iron, but no anomalous microphase iron abundances, (2) nanophase and microphase iron abundances are lower at higher latitudes, which suggests lower solar wind and micrometeoroid impact flux at these latitudes, (3) decreased nanophase iron abundances in the nearside suggest a lower solar wind flux due to Earth's magnetotail, and (4) in contrast to Mercury, the microphase to nanophase iron abundance ratio is lower on the Moon due to lower surface temperatures and less intense space weathering.

SPECTRAL UNMIXING BASED CONSTRUCTION OF LUNAR MINERAL ABUNDANCE MAPS

Abstract. In this study we apply a nonlinear spectral unmixing algorithm to a nearly global lunar spectral reflectance mosaic derived from hyper-spectral image data acquired by the Moon Mineralogy Mapper (M3) instrument. Corrections for topographic effects and for thermal emission were performed. A set of 19 laboratory-based reflectance spectra of lunar samples published by the Lunar Soil Characterization Consortium (LSCC) were used as a catalog of potential endmember spectra. For a given spectrum, the multi-population population-based incremental learning (MPBIL) algorithm was used to determine the subset of endmembers actually contained in it. However, as the MPBIL algorithm is computationally expensive, it cannot be applied to all pixels of the reflectance mosaic. Hence, the reflectance mosaic was clustered into a set of 64 prototype spectra, and the MPBIL algorithm was applied to each prototype spectrum. Each pixel of the mosaic was assigned to the most similar prototype, and the set of endmembers previously determined for that prototype was used for pixel-wise nonlinear spectral unmixing using the Hapke model, implemented as linear unmixing of the single-scattering albedo spectrum. This procedure yields maps of the fractional abundances of the 19 endmembers. Based on the known modal abundances of a variety of mineral species in the LSCC samples, a conversion from endmember abundances to mineral abundances was performed. We present maps of the fractional abundances of plagioclase, pyroxene and olivine and compare our results with previously published lunar mineral abundance maps.

Inversion of the Main Mineral Compositions and Subdivision of Tectonic Units on Lunar LQ‐4 based on Chang'e Data

AbstractSpectra are sensitive in detecting main minerals on the lunar surface from visible light to infrared light. Since spectral characteristics of minerals are closely related to their compositions and the maturity level of soil on the Moon, studying the compositions and distribution of elements and minerals on the lunar surface can help to understand the evolution of the Moon through remote sensing technology. The correlation between the spectral characteristics of Chang'e‐1 interference imaging spectrometry (IIM) reflectance images and the mineral contents of LSCC (Lunar Soil Characterization Consortium) lunar surface mineral samples was discussed and the spatial distributions of FeO and Al2O3 contained in both pyroxene and plagioclase on LQ‐4 were studied using the improved angle parameter method, MNF, and band ratio statistics. A comparison of the mapping results of the optical models by Lucey, Shkuractov and other researchers on Clementine and the gamma ray spectrometry data shows that the content error is within 0.6% for lunar mare areas and close to 1% for the highland areas. The tectonic framework on the lunar surface was also investigated. And based on integrated analysis of previous findings on topography of the lunar surface, Chang'e LAM, CCD and LOLA images and the gravity anomalies data (Clementine GLGM‐2), the tectonic unit subdivision was established for LQ‐4, the idea of subdividing the lunar tectonic units was proposed, and this will provide a good foundation for studying the lunar tectonic evolution.

An improved radiative transfer model for estimating mineral abundance of immature and mature lunar soils

An improved Hapke’s radiative transfer model (RTM) is presented to estimate mineral abundance for both immature and mature lunar soils from the Lunar Soil Characterization Consortium (LSCC) dataset. Fundamental to this improved Hapke’s model is the application of an alternative equation to describe the effects of larger size submicroscopic metallic iron (SMFe) (>50nm) in the interior of agglutinitic glass that mainly darken the host material, contrasting to the darkening and reddening effects of smaller size SMFe (<50nm) residing in the rims of mineral grains. Results from applying a nonlinear inversion procedure to the improved Hapke’s RTM show that the average mass fraction of smaller and larger size SMFe in lunar soils was estimated to be 0.30% and 0.31% respectively, and the particle size distribution of soil samples is all within their measured range. Based on the derived mass fraction of SMFe and particle size of the soil samples, abundances of end-member components composing lunar soil samples were derived via minimizing the difference between measured and calculated spectra. The root mean square error (RMSE) between the fitted and measured spectra is lower than 0.01 for highland samples and 0.005 for mare samples. This improved Hapke’s model accurately estimates abundances of agglutinitic glass (R-squared=0.88), pyroxene (R-squared=0.69) and plagioclase (R-squared=0.95) for all 57 samples used in this study including both immature and mature lunar soils. However, the improved Hapke’s RTM shows poor performance for quantifying abundances of olivine, ilmenite and volcanic glass. Improving the model performance for estimation of these three end-member components is the central focus for our future work.

Modeling the Reflectance of the Lunar Regolith by a New Method Combining Monte Carlo Ray Tracing and Hapke’s Model with Application to Chang’E-1 IIM Data

In this paper, we model the reflectance of the lunar regolith by a new method combining Monte Carlo ray tracing and Hapke's model. The existing modeling methods exploit either a radiative transfer model or a geometric optical model. However, the measured data from an Interference Imaging spectrometer (IIM) on an orbiter were affected not only by the composition of minerals but also by the environmental factors. These factors cannot be well addressed by a single model alone. Our method implemented Monte Carlo ray tracing for simulating the large-scale effects such as the reflection of topography of the lunar soil and Hapke's model for calculating the reflection intensity of the internal scattering effects of particles of the lunar soil. Therefore, both the large-scale and microscale effects are considered in our method, providing a more accurate modeling of the reflectance of the lunar regolith. Simulation results using the Lunar Soil Characterization Consortium (LSCC) data and Chang'E-1 elevation map show that our method is effective and useful. We have also applied our method to Chang'E-1 IIM data for removing the influence of lunar topography to the reflectance of the lunar soil and to generate more realistic visualizations of the lunar surface.

Lunar pyroxene and olivine abundance analysis of Sinus Iridum

Composition distribution on the lunar surface is an important scientific topic. In order to establish pyroxene and olivine quantitative inversion models of Sinus Iridum, we analyze the spectral characteristics of major mineral. First,according to pyroxene and olivine in the 300–2600 nm spectral range of absorption feature, we use Modified Gaussian Model(MGM) model to calculate spectral absorption characteristics at 19 sampling points of the Lunar Soil Characterization Consortium(LSCC), which is also performed detailed chemistry and modal abundance analyses on each of these samples. We obtain five spectral absorption characteristics, and each has three characteristic parameters:absorption center(Center), absorption width(FWHM), and absorption intensity(Strength). Then, the mineral inversion model is established by using linear regression analysis based on LSCC data, and is validated by simulated spectra, which are made by the bidirectional reflectance function of Hapke on M3hyper-spectra data. At last, the quantitative inversion models of pyroxene and olivine have been applied to the Sinus Iridum area, which is the preferred landing station of Chang'E-3 lunar rover. The mineral abundance distribution maps of this area are made.Experiments show that using MGM model fitting analysis method for hyper-spectra remote sensing of minerals identification is a kind of feasible research approach.

Hybridization of partial least squares and neural network models for quantifying lunar surface minerals

The goal of this study is to develop an efficient and accurate model for using visible–near infrared reflectance spectra to estimate the abundance of minerals on the lunar surface. Previous studies using partial least squares (PLS) and genetic algorithm–partial least squares (GA–PLS) models for this purpose revealed several drawbacks. PLS has two limitations: (1) redundant spectral bands cannot be removed effectively and (2) nonlinear spectral mixing (i.e., intimate mixtures) cannot be accommodated. Incorporating GA into the model is an effective way for selecting a set of spectral bands that are the most sensitive to variations in the presence/abundance of lunar minerals and to some extent overcomes the first limitation. Given the fact that GA–PLS is still subject to the effect of nonlinearity, here we develop and test a hybrid partial least squares–back propagation neural network (PLS–BPNN) model to determine the effectiveness of BPNN for overcoming the two limitations simultaneously. BPNN takes nonlinearity into account with sigmoid functions, and the weights of redundant spectral bands are significantly decreased through the back propagation learning process. PLS, GA–PLS and PLS–BPNN are tested with the Lunar Soil Characterization Consortium dataset (LSCC), which includes VIS–NIR reflectance spectra and mineralogy for various soil size fractions and the accuracy of the models are assessed based on R2 and root mean square error values. The PLS–BPNN model is further tested with 12 additional Apollo soil samples. The results indicate that: (1) PLS–BPNN exhibits the best performance compared with PLS and GA–PLS for retrieving abundances of minerals that are dominant on the lunar surface; (2) PLS–BPNN can overcome the two limitations of PLS; (3) PLS–BPNN has the capability to accommodate spectral effects resulting from variations in particle size. By analyzing PLS beta coefficients, spectral bands selected by GA, and the loading curve of the latent variable with the largest weight in PLS–BPNN, we conclude that spectral information incorporated into the three models is directly derived from the diagnostic absorption bands associated with the individual minerals. It is concluded that the PLS–BPNN model should be applicable to both Clementine UV–VIS–NIRs and Moon Mineralogy Mapper (M3) data.

Quantifying TiO2 abundance of lunar soils: Partial least squares and stepwise multiple regression analysis for determining causal effect

Partial least squares (PLS) regression was applied to the Lunar Soil Characterization Consortium (LSCC) dataset for spectral estimation of TiO2. The LSCC dataset was split into a number of subsets including the low-Ti, high-Ti, total mare soils, total highland, Apollo 16, and Apollo 14 soils to investigate the effects of interfering minerals and nonlinearity on the PLS performance. The PLS weight loading vectors were analyzed through stepwise multiple regression analysis (SMRA) to identify mineral species driving and interfering the PLS performance. PLS exhibits high performance for esti- mating TiO2 for the LSCC low-Ti and high-Ti mare samples and both groups analyzed together. The results suggest that while the dominant TiO2-bearing minerals are few, additional PLS factors are re- quired to compensate the effects on the important PLS factors of minerals that are not highly corrected to TiO2, to accommodate nonlinear relationships between reflectance and TiO2, and to correct incon- sistent mineral-TiO2 correlations between the high-Ti and low-Ti mare samples. Analysis of the LSCC highland soil samples indicates that the Apollo 16 soils are responsible for the large errors of TiO2 es- timates when the soils are modeled with other subgroups. For the LSCC Apollo 16 samples, the domi- nant spectral effects of plagioclase over other dark minerals are primarily responsible for large errors of estimated TiO2. For the Apollo 14 soils, more accurate estimation for TiO2 is attributed to the posi- tive correlation between a major TiO2-bearing component and TiO2, explaining why the Apollo 14 soils

Radiative transfer modeling for quantifying lunar surface minerals, particle size, and submicroscopic metallic Fe

The main objective of this work is to quantify lunar surface minerals (agglutinate, clinopyroxene, orthopyroxene, plagioclase, olivine, ilmenite, and volcanic glass), particle sizes, and the abundance of submicroscopic metallic Fe (SMFe) from the Lunar Soil Characterization Consortium (LSCC) data set with Hapke's radiative transfer theory. The mode is implemented for both forward and inverse modeling. We implement Hapke's radiative transfer theory in the inverse mode in which, instead of commonly used look‐up tables, Newton's method and least squares are jointly used to solve nonlinear questions. Although the effects of temperature and surface roughness are incorporated into the implementation to improve the model performance for application of lunar spacecraft data, these effects cannot be extensively addressed in the current work because of the use of lab‐measured reflectance data. Our forward radiative transfer model results show that the correlation coefficients between modeled and measured spectra are over 0.99. For the inverse model, the distribution of the particle sizes is all within their measured range. The range of modeled SMFe for highland samples is 0.01%–0.5%, and for mare samples it is 0.03%–1%. The linear trend between SMFe and ferromagnetic resonance (Is) for all the LSCC samples is consistent with laboratory measurements. For quantifying lunar mineral abundances, the results show that the R squared for the training samples (Is/FeO ≤ 65) are over 0.65 with plagioclase having highest correlation (0.94) and pyroxene having the lowest correlation (0.68). In future work, the model needs to be improved for handling more mature lunar soil samples.

Radiative transfer modeling of near‐infrared reflectance of lunar highland and mare soils

With near‐infrared hyperspectral data sets returned from KAGUYA/SELENE and Chandrayaan‐1's Moon Mineralogic Mapper (M3), accurate evaluation and interpretation of lunar data sets with higher spectral resolution has never been more critical. Here we test a new radiative transfer spectral modeling algorithm to determine composition from hyperspectral reflectance spectra of lunar soils. Data for 19 lunar mare and highland soil samples previously characterized by the Lunar Soil Characterization Consortium are used for validation. Spectral fits are made using a goodness of fit metric considering spectral shape, spectral contrast, spectral slope, and iron abundance. High precision fits are achieved for nearly every soil with this algorithm. Using a plot of spectral shape relative to the ratio Mg′ (i.e., molar (Mg/(Mg + Fe)) × 100) determines the winning model and composition. Mg′ is determined with an average difference of ∼11–15 and ∼3–8 units before and after a correction is applied, respectively. Mineralogy is determined with an average difference of ∼5–15 vol% depending upon the mineral constituent.

Mineralogical and chemical characterization of lunar highland soils: Insights into the space weathering of soils on airless bodies

With reflectance spectroscopy, one is measuring only properties of the fine‐grained regolith most affected by space weathering. The Lunar Soil Characterization Consortium has undertaken the task of coordinated characterization of lunar soils, with respect to their mineralogical and chemical makeup. It is these lunar soils that are being used as “ground truth” for all airless bodies. Modal abundances and chemistries of minerals and glasses in the finest size fractions (20–45, 10–20, and &lt;10 μm) of four Apollo 14 and six Apollo 16 highland soils have been determined, as well as their bulk chemistry and IS/FeO values. Bidirectional reflectance measurements (0.3–2.6 μm) of all samples were performed in the Reflectance Experiment Laboratory. A significant fraction of nanophase Fe0 (np‐Fe0) appears to reside in agglutinitic glasses. However, as grain size of a soil decreases, the percentage of total iron present as np‐Fe0 increases significantly, whereas the agglutinitic glass content rises only slightly; this is evidence for a large contribution to the IS/FeO values from the surface‐correlated nanophase Fe0, particularly in the &lt;10 μm size fraction. The compositions of the agglutinitic glasses in these fine fractions of the highland soils are different from the bulk chemistry of that size; however, compositional trends of the glasses are not the same as those observed for mare soils. It is apparent that the glasses in the highland soils contain chemical components from outside their terrains. It is proposed that the Apollo 16 soils have been adulterated by the addition of impact‐transported soil components from surrounding maria.

Quantification of the chemical composition of lunar soil in terms of its reflectance spectra by PCA and SVM

In the second phase of the Chang’E Program an unmanned lunar rover will be launched onto the Moon. When ground scientists get a full understanding of the chemical composition of lunar soil around the rover, they can make more detailed survey plans for the rover and various payloads onboard so as to satisfy their scientific objectives. There is an obvious relationship between the reflectance of lunar soil and its chemical characteristics. Both principal component analysis (PCA) and support vector machine (SVM) models were applied to establishing the relationship between the reflectance spectra and chemical compositions of lunar highland and mare soil samples sent back by Apollo missions 11, 12, 14, 15, 16 and 17 and measured by Lunar Soil Characterization Consortium (LSCC). PCA was used to reduce and select the features of the reflectance spectra of lunar soil samples. Then, these features were put into SVM to estimate the abundances of various chemical components in lunar soil. We also compared the results of our measurement with those obtained by the SVM model [partial least squares (PLS)] and the principal component regression (PCR) model reported in literature. Our studies showed that with the exception of TiO2, the results of prediction of the abundances of chemical compounds in lunar soil by our model are much more reliable than those reported in literature. The reflectance spectra of lunar soil are closely related to the materials from which it was derived.