Mary Ann Moody / Earth & Environmental Sciences / Faculty Mentor: Majie Fan
Grain size analysis has played a pivotal role in interpreting sedimentary environments. The traditional study method that involves manual plotting and visual identification of different grain size patterns can be time-intensive and calls for new statistical analysis. In the western United States, the transition from fluvial (river-transported) to eolian (wind-transported) environments during the late Paleogene was interpreted to be abrupt based on the apparent changes in rock types. However, this transition could be gradual and earlier than the rock-type changes. To identify eolian deposition during the apparent fluvial deposition, this study employs machine learning clustering methods of K-Means and HDBSCAN, alongside dimensionality reduction techniques of principal component analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE), to a large grain size dataset collected in the Flagstaff Rim area, Wyoming. The outcomes of the machine learning classification align well with the traditional grain size pattern identification and show the concurrent presence of fluvial and eolian transport for some time before the abrupt rock type change, suggesting a gradual transition from fluvial to eolian deposition. The results highlight the effectiveness of applying machine learning methods to large grain-size datasets for discerning sedimentary environments and show the trend for future sedimentological research.
Xiangwei Guo
Congratulations on finishing your discover project, Mary Ann! You’ve done an amazing job!
mmm2637
Thank you very much!
Cindy Lou Skipper
Great work Mary Ann!
mmm2637
Thank you Cindy Lou!