Abstract
Quantifying geochemical provenance of the Beaverhead Group in southwestern Montana using unsupervised machine learning on a sparse dataset
Abstracts with programs - Geological Society of America, Vol.54(5)
Geological Society of America, 2022 annual meeting; GSA connects 2022
10/2022
DOI: 10.1130/abs/2022AM-383093
Abstract
A solid understanding of both the depositional age and provenance of the Late Cretaceous-Early Cenozoic Beaverhead Group is vital in analyzing the nature and timing of flat-slab subduction during the Laramide orogeny. Several methods including stratigraphic studies, structural analysis, palynology and detrital zircon dating have previously been used to constrain these two factors. By using XRF spectrometry to obtain major element compositions and LA-ICP-MS to obtain trace element compositions, we hope to further constrain the provenance of the Beaverhead Group using conglomerate clast geochemistry in conjunction with clustering analysis, pattern recognition and other unsupervised machine learning techniques. The Beaverhead Group is up to 15,000 feet (about 4.5 km) thick and is largely classified by thickly-bedded conglomerates. Clasts in the conglomerates are compositionally heterogeneous, but two major lithologies are limestone and dolostone, previously interpreted to be derived from several thick Paleozoic and Early Mesozoic carbonate units. Unless they contain diagnostic fossils, however, these carbonate clasts are useless in traditional provenance considerations because their possible source units are temporally and spatially wide-ranging. By utilizing geochemistry and machine learning, however, we hope to sufficiently resolve provenance relationships between these carbonate clasts and their ubiquitous source rocks. In total, 51 carbonate source samples and 21 carbonate clast samples were collected and analyzed. To capture any variation in sediment transport style, the spatial distribution of these samples covers various possible source-to-sink pathways from proximal alluvial-fan-type deposition to distal mass-flow-type deposition. Although initial findings have been able to distinguish distinct geochemical signals among our source rock samples, they have failed to make any robust quantitative conclusions about provenance. If our study is successful, we will interpret the data in terms of Laramide tectonics to better constrain the timing and mechanisms of the event.
Details
- Title: Subtitle
- Quantifying geochemical provenance of the Beaverhead Group in southwestern Montana using unsupervised machine learning on a sparse dataset
- Creators
- Zachary Vig - University of Iowa, Department of Earth and Environmental Sciences Iowa City, IA USA United StatesEmily Finzel
- Resource Type
- Abstract
- Publication Details
- Abstracts with programs - Geological Society of America, Vol.54(5)
- Conference
- Geological Society of America, 2022 annual meeting; GSA connects 2022
- Publisher
- Geological Society of America (GSA)
- DOI
- 10.1130/abs/2022AM-383093
- ISSN
- 0016-7592
- Alternative title
- Geological Society of America, 2022 annual meeting; GSA connects 2022
- Language
- English
- Date published
- 10/2022
- Academic Unit
- Earth and Environmental Sciences
- Record Identifier
- 9984375351502771
Metrics
12 Record Views