Compositional units on Mercury from MESSENGER spectral observations: comparison of clustering techniques

Abstract

The Mercury Atmospheric and Surface Composition Spectrometer (MASCS) obtained spectra of much of the surface of Mercury during the first two MESSENGER flybys of the planet [1]. The dataset have not been corrected for any effect due to observing geometry, but only converted to reflectance [2]. The characterization of spectral units is performed by statistical techniques. In order to extract the spectral shapes of the primary surface components exposed in the surface area analyzed, we applied an R-mode factor analysis [3] [4]. That leads to the evaluation of the eigenvectors of the covariance matrix and their abundances along the track. The results indicates that the NIR spectral range is carrying less information than the VIS portion and that the eigenvectors are unchanged if the full wavelength range is selected rather than limiting observations to the VIS. The analysis shows that seven eigenvectors are needed to reconstruct the original data, where each eigenvector can be regarded as a representative of a spectral class that varies in abundance along the track. The first eigenvector always displays a strong positive or “red” slope, carrying the effects associated to the viewing geometry and all eigenvectors show distinctive spectral signatures. The eigenvector abundances show marked geographical variation and a strong correlation with surface units mapped by MESSENGER’s Mercury Dual Imaging System (MDIS). We apply a decorrelation technique (Mahalanobis transformation [5]) to remove dependence on observation angle in the retrieved eigenvector abundances and then used the corrected abundances to classify or cluster the measurements and to identify spectral units. We used three clustering techniques and then we compare the output from the algorithms. At the same time, we make use of newly available high-temperature spectra from our Planetary Emissivity Laboratory [6] to assist in the identification of the components of each unit. Application to data from the first flyby provides us with confidence in the ability of these techniques to extract physical properties of surface materials.