Document Type
Article
Publication Date
6-2022
Keywords
machine learning; sand; mineralogy; global prediction; provenance; petrography
Abstract
Petrography has long been used as a tool to decipher the sedimentary provenance of sand and sandstone based on relative proportions of detrital grain types. However, there is not yet a predictive framework that constrains the relationship between sand grain composition and the factors that influence it, including bedrock lithology, elevation, slope, basin area, precipitation, and temperature. We used a globally extensive modal point-count data set of 3,522 Pleistocene to modern sand samples to train an inline series of random forest meta-estimators to predict relative abundances of quartz (Q), feldspar (F), and lithic (L) grain types and eight subcompositions. The trained random forest models were fit to a global data set of fluvial watersheds to produce the global prediction of sand modal composition (GloPrSM) model which predicts Q-F-L of fluvially transported sediment. GloPrSM predicts quartz enrichment in low latitudes (35°N to 35°S), feldspar enrichment near plutonic and metamorphic crystalline terranes in middle to high latitudes, and lithic enrichment near modern and ancient collisional plate boundaries and flood basalts. Our analyses reveal that slope, temperature, and the relative abundance of exposed metamorphic and felsic to intermediate plutonic rocks are the most important predictors of Q-F-L composition. In addition, temperatures higher than 15°C and large drainage areas appear to promote quartz enrichment, while steeply sloping environments favor lithic enrichment. The GloPrSM model represents the first global-scale estimate of fluvial sand compositional modes and illustrates that the spatial distribution of Q-F-L at Earth's surface can be predicted from the first-order factors that generate sediment.
Citation
Johnson, J. I., Sharman, G. R., Szymanski, E., & Huang, X. (2022). Machine Learning Applied to a Modern-Pleistocene Petrographic Data Set: The Global Prediction of Sand Modal Composition (GloPrSM) Model. Journal of Geophysical Research: Earth Surface, 127 (7), e2022JF006595. https://doi.org/10.1029/2022JF006595
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.
