Statistics – Applications
Scientific paper
Dec 2009
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2009agufm.p21a1193d&link_type=abstract
American Geophysical Union, Fall Meeting 2009, abstract #P21A-1193
Statistics
Applications
[5464] Planetary Sciences: Solid Surface Planets / Remote Sensing, [5470] Planetary Sciences: Solid Surface Planets / Surface Materials And Properties
Scientific paper
During the MESSENGER flybys of Mercury, the Mercury Atmospheric and Surface Composition Spectrometer (MASCS) obtained spectra of much of the surface of the planet. We employ principal component and clustering analyses to identify and characterize spectral units along the MASCS ground tracks. To retrieve the spectral shapes of the components present in the dataset, we applied an R-mode factor analysis, aiming to minimize data covariance. Identification of the different components and their abundances is accomplished by principal component analysis and an evaluation of the eigenvectors and eigenvalues of the covariance matrix. We apply the analysis both to the entire range and also to each channel to monitor possible differences. Full range analysis shows that seven eigenvectors are needed to reconstruct the data within the error. A comparison indicates that the near-infrared (NIR) channel is carrying significantly less information than the visible portion, even if the eigenvectors are essentially the same. The first eigenvector always displays a strong “red” slope, and all eigenvectors show characteristic spectral signatures. Each spectral eigenvector can be regarded as a representative of different spectral classes with varying abundance along the track. We apply a decorrelation technique (Mahalanobis transformation) to completely remove dependencies from observation angles in the retrieved concentration coefficients, since we do not photometrically correct the data. Concentration coefficients, obtained from the covariance matrix decomposition, indicate that spectral units show significant geographical variation and a strong correlation with surface units mapped by MESSENGER’s Mercury Dual Imaging System (MDIS). We computed the hierarchical clustering of the data by a weighted centroid approach. The distance between two clusters is defined as that between each cluster centroid weighted by the number of objects in that cluster, and a centroid is defined by the average of all spectra within the cluster. At each clustering step, data points are included in the nearest cluster and visualized on a bidimensional map. Thus, data exhibiting rare combinations of spectral units, relative to the observed surface, are easily recognized. The next step will be a detailed analysis of each identified unit. At the same time, we make use of the newly available high-temperature spectra from our Planetary Emissivity Laboratory to progress toward the identification of the components of each unit. The latest applications to data from the first flyby give us confidence in the ability of these techniques to extract physical properties of surface material.
D'Amore Mario
Head James W.
Helbert Jérôme
Holsclaw Gregory M.
Izenberg Noam R.
No associations
LandOfFree
Compositional units on Mercury from principal component and clustering analyses of MESSENGER spectrometer observations does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Compositional units on Mercury from principal component and clustering analyses of MESSENGER spectrometer observations, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Compositional units on Mercury from principal component and clustering analyses of MESSENGER spectrometer observations will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1768376