Tectonomagmatic Evolution of Northwestern Mexico, and Applications of Machine Learning in Economic Geology
The work presented here was conducted in two study areas: La Preciosa mining project (Chapters 1 and 2) and Ray Mine (Chapter 3). Chapter 1 of this dissertation contributes to the tectonomagmatic evolution of northwestern Mexico by focusing on the temporal changes in magma source(s) regions, magma generation processes, and tectonic implications. In this chapter, I provide thin section analyses, whole rock geochemistry, in-situ zircon U-Pb ages and Hf isotope compositions of subsurface lithologies from La Preciosa mining project; located in the northwestern corner of Mesa Central, Durango, Mexico. The newly acquired data identified five magmatic events. The first consists of ca. 162 Ma dacitic volcanism synchronous with siliciclastic sedimentation associated with the Nazas volcano-sedimentary province. The second magmatic event (88-61 Ma) is associated with the early Laramide age magmatism. The third event consists of 57-48 Ma rhyodacitic magmatism showing a predominantly positive Hf isotopic composition indicating the dominant role of mantle derived sources. This event is associated with late stage Laramide age magmatism. The fourth event is attributed to the transitional period magmatism and consists of 46-41 Ma andesitic magmatism. Geochemical and Hf isotopic compositions suggest magma source region heterogeneity that can be explained by partial melting of crustally-derived radiogenic sources and relatively non-radiogenic sources. The fifth event, consists of 32-28 Ma silicic volcanism associated with the first pulse of the ignimbrite flare-up of the Sierra Madre Occidental silicic large igneous province. This chapter have been published in Gondwana Research Journal: “Garcia, J.A., Mahar, M.A., Goodell, P.C., Molina, C., Ricketts, J.W., 2021. Petrology, geochemistry, stratigraphy, zircon U-Pb geochronology and Hf isotopic compositions of subsurface lithologies in northwestern Mesa Central, Durango, Mexico: Implications for the tectonomagmatic evolution of northwestern Mexico. Gondwana Research 93, 1-25.” In Chapters 2 and 3, the focus shifts towards the application of unsupervised machine learning and multivariate data analysis techniques to better understand the formation of ore deposits. In Chapter 2, a combined emergent self-organizing map (ESOM)/generalized unified matrix (U*-matrix) approach is employed to identify the geochemical signature associated with various stages of precious metal precipitation (ore paragenesis) at the La Preciosa mining project. A total of 8 high-dimensional clusters were identified using this approach, two of which show high proportions of silver and gold (clusters 3 and 8). Integration of the geochemical associations observed for clusters 3 and 8 with prior petrographic data, allows for the placement of these clusters into a paragenetic context as a stage. Cluster 8 are associated with precipitation of gold/electrum during early paragenesis; and samples in cluster 3 are associated with precipitation of Ag-sulfides, base metal sulfides, and Ag-sulfosalts during late stage paragenesis. Chapter 3 focuses on the application of classical unsupervised multivariate data analysis techniques such as principal component analysis (PCA) in combination with the minimum covariance determinant (MCD) estimator (robust PCA), and k-means cluster analysis, to geochemically identify and characterize the various intrusive events in the diabase suite, and determine the relationship between these events and copper mineralization at the Ray mine. Three intrusive events (C1, C2, and C3) were identified by k-means cluster analysis. Sub-economic hypogene copper mineralization (Cu < 0.1%) are primarily concentrated in clusters C2 and C3, where Fe2O3T is above 12 wt. %. This is attributed to the preferential interaction between earlier mineralizing magmatic fluids carrying a lower copper content and parts of the diabase where iron was more readily available. Copper grades ranging from 0.1 to 0.3 appear to be similarly distributed amongst the three diabase intrusive events (C1=19.58%, C2=13.17%, and C3= 18.11% of the samples). Because most of the readily available iron was likely consumed during the earlier lower copper content mineralizing episode(s), higher copper content magmatic fluids associated with later intrusions would have interacted with the lower iron parts of the diabase (Fe2O3 below 12 wt. %,), leading to localized, higher grade hypogene copper mineralization in intrusive event C1 and parts of C2. Therefore, changes in readily available iron in the diabase appear to be a primary control for hypogene copper precipitation and enrichment.
Garcia, Jose Antonio, "Tectonomagmatic Evolution of Northwestern Mexico, and Applications of Machine Learning in Economic Geology" (2021). ETD Collection for University of Texas, El Paso. AAI28869192.