Paper No. 132-7
Presentation Time: 3:30 PM
MINERAL ASSOCIATION ANALYSIS: PREDICTING THE MINERALOGY OF PLANETARY ANALOGS AND POTENTIAL LANDING SITES (Invited Presentation)
MORRISON, Shaunna1, PRABHU, Anirudh1, ELEISH, Ahmed2, HAZEN, Robert M.1, GOLDEN, Joshua J.3, DOWNS, Robert T.3, PERRY, Samuel4, BURNS, Peter C.5, RALPH, Jolyon6 and FOX, Peter7, (1)Earth and Planets Laboratory, Carnegie Institution for Science, 5251 Broad Branch Road NW, Washington, DC 20015, (2)Tetherless World Constellation, Rensselaer Polytechnic Institute, 110 Eighth Street, Troy, NY 12180, (3)Department of Geosciences, University of Arizona, Tucson, AZ 85721, (4)301 Stinson Remick Hall, 301 Stinson Remick Hall, Notre Dame, IN 46556-7200, (5)Chemistry and Biochemistry, University of Notre Dame, 301 Stinson-Remick Hall, Notre Dame, IN 46556, (6)Hudson Institute of Mineralogy, Keswick, VA 22947, (7)Rensselaer Polytechnic Institute, 5 State St, Troy, NY 12180
In the context of planetary exploration, the growing wealth of mineralogical data resources, both on Earth and beyond, has paved the way for advanced predictive methods, including mineral association analysis [1-4]. This study employs association analysis to predict previously unknown mineral occurrences, deposits, geologic environments, and mineral inventories on Earth, including a planetary analog site. Leveraging mineral association rules, this approach enables the identification of promising locations for discovering specific mineral species, along with their likelihood based on various probability metrics. Moreover, it expands to predict mineral assemblages that may be indicative of distinct planetary conditions, environments, or deposit types. This powerful recommender system aids researchers in pinpointing locations on Earth's surface or other planetary bodies where particular mineral assemblages are likely to exist, even if they have not been previously identified. Such predictive capabilities extend beyond conventional database queries, providing new insights into unexplored terrains and offering a broader understanding of planetary mineralogy. This innovative methodology opens avenues for identifying planetary analogy sites, assessing resources, and planning missions for mineralogical exploration on celestial bodies.
[1] Brin S, Motwani R, Silverstein C (1997) Beyond Market Baskets, ACM SIGMOD Record.
[2] Morrison SM, Prabhu A, Eleish A, Hazen RM, Golden JJ, Downs RT, Perry S, Burns PC, Ralph J, Fox P, Machine learning approaches for predictive mineralogy in Earth and planetary science: A study in mineral association analysis, PNAS Nexus, 2(5):pgad110. doi: 10.1093/pnasnexus/pgad110
[3] Prabhu A, Morrison SM, Giovannelli D (2021) A new way to evaluate association rule mining methods and its applicability to mineral association analysis, AGU Annual Meeting Abstract
[4] Morrison SM, Anirudh Prabhu, Ahmed Eleish, Shweta Narkar, Peter Arthur Fox, Golden JJ, Downs RT, Samuel Perry, Peter C Burns, Jolyon Ralph, and Hazen RM (2020) Affinity analysis of mineral co-occurrence: Predicting unknown mineral occurrences with machine learning, AGU Annual Meeting, ED044-0001
![[Visit Client Website]](/img/gsa/banner.jpg)