Automated classification of interplanetary dust particles: Johnson Space Center Cosmic Dust Catalog Volume 15 |
| |
| Authors: | Jeremie LASUE Tomasz STEPINSKI Samuel W BELL |
| |
| Institution: | 1. Lunar and Planetary Institute, 3600 Bay Area Boulevard, Houston, Texas 77058, USA;2. LANL, Space Science and Applications, ISR‐1, Mail Stop D‐466, Los Alamos, New Mexico 87545, USA;3. Amherst College, Amherst, Massachusetts 01002–5000, USA |
| |
| Abstract: | Abstract– The “Cosmic Dust Catalog,” published by the NASA Johnson Space Center (JSC), describes thousands of interplanetary dust particles subjected to preliminary analysis and with labels indicating their origin. However, only about 80% of the particles are assigned unambiguous labels, the labels of the remaining 20% being uncertain. In addition, the Stardust mission results opened up the possibility that some particles classified as terrestrial contaminants are instead of cosmic (cometary) origin. In this article, we present a methodology for automatic classification of particles on the basis of similarity of their X‐ray energy dispersive spectrometry spectra. The method is applied to the 467 particles constituting Volume 15 of the catalog. A first part of the analysis is to digitize the spectra from their scanned images. The digitized spectra are subjected to agglomerative clustering, which reveals 16 distinct clusters or compositional types of particles. The Sammon’s map is used to visualize the relationship between different clusters; 6 clusters corresponding to cosmic particles and 10 clusters corresponding to terrestrial contaminants are clearly separated on the map indicating overall differences between diverse spectra of cosmic and terrestrial particles. By reconciling labels with the clustering structures, we propose the relabeling of 155 particles including the relabeling of 31 terrestrial contaminants into cosmic particles. The proposed relabeling needs to be confirmed by in‐depth study of these particles. The paucity of particles with firmly determined cometary or asteroidal origin makes it difficult to establish whether the spectra based autoclassification can be utilized to discriminate between cometary and asteroidal particles. The methodology presented here can be used to classify all particles published in the catalog, as well as different samples for which comparable spectra are available. |
| |
| Keywords: | |
|
|
