SMU Data Science Review
Abstract
In this paper we present a methodology for automating theclassification of spectrally resolved observations of multiple emissionlines with the Atacama Large Millimeter/submillimeter Array (ALMA).Molecules in planetary atmospheres emit or absorb different wavelengthsof light thereby providing a unique signature for each species. ALMAdata were taken from interferometric observations of Titan made be-tween UT 2012 July 03 23:22:14 and 2012 July 04 01:06:18 as part ofALMA project 2011.0.00319.S. We first employed a greedy set cover algorithm to identify the most probable molecules that would reproducethe set of frequencies with respective flux greater than 3σaway from themean. We then selected a subset of those molecules as present in theatmosphere by specifying a selection threshold and one of two selectionmetrics. Our model was able to correctly classify 100% of previously dis-covered molecules in Titan’s atmosphere from this data, including EthylCyanide as reported by Cardiner et al. (2015)[2]. One molecule, Formalde-hyde, was identified in both selection metrics that was not previouslyrecorded in the atmosphere. The results of our methodology allow for astreamlined approach for molecule classification and anomaly detectionin planetary atmospheres.
Recommended Citation
Cocke, Steven; Wilkins, Andrew; McDaniel, Josephine; Santerre, John; and Nixon, Conor
(2020)
"Automated Spectroscopic Detection And Mapping Using ALMA and Machine LearningTechniques,"
SMU Data Science Review: Vol. 3:
No.
1, Article 12.
Available at:
https://scholar.smu.edu/datasciencereview/vol3/iss1/12
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License
Included in
Cosmology, Relativity, and Gravity Commons, External Galaxies Commons, Instrumentation Commons, Other Astrophysics and Astronomy Commons, Physical Processes Commons, Stars, Interstellar Medium and the Galaxy Commons, The Sun and the Solar System Commons
