Exploratory data analysis methods of multivariate statistical techniques and spatial visualization are emerging trends in understanding big datasets. In this project, these techniques are applied to a large igneous geochemical dataset from the southern California segment of the Mesozoic Cordilleran arc to better understand magmatic and plate tectonic processes at a subduction zone. A set of 287 granitic samples collected by Baird and Miesch (1984) from the Peninsular Ranges batholith is analyzed for 38 geochemical elements. Patterns in both the geochemical variation and the spatial variation of this dataset are explored. Since geochemical data are compositional in nature, special treatment is needed in analyzing them. Robust principal component analysis for compositional data is used to summarize the 38 geochemical variables into three principal components that are visualized using biplots. The first three principal components appear to be related to extent of differentiation, magma source depth, and possibly solubility, respectively. The first principal component (PC1) accounting for 56.7% of the explained variance, arranges the elements in order of incompatibility. The main associations of PC2 (17.3% of

explained variance) are groupings of rare earth elements, along with Y and Sr – suggesting the effect of deep garnet and shallow plagioclase fractionation in response to pressure and therefore depth. A weak association of soluble elements is found in PC3 (6.7% of explained variance). Spatial geochemical variation is explored by mapping the standard geochemical parameters related to the principal component interpretations as well as the three principal components, and then comparing them. Extent of differentiation is mapped using SiO2 and PC1 and resulting maps show similar patterns. Magma source depth is mapped using Sr/Y, La/Yb and PC2 and similar patterns are found. Alkalinity and possibly solubility and mobility is mapped using K2O/SiO2 and PC3, but these maps do not share similar patterns. Using the exploratory data analysis methods of multivariate analysis and spatial visualization is helpful in understanding geochemical patterns and trends in subduction zones.

LLU Discipline



Basic Sciences


School of Medicine

First Advisor

Clausen, Benjamin L.

Second Advisor

Hayes, William K.

Third Advisor

Nick, Kevin E.

Degree Name

Master of Science (MS)

Degree Level


Year Degree Awarded


Date (Title Page)




Library of Congress/MESH Subject Headings

Geology - Structural -- California - Southern; Batholiths -- California - Southern; Island arcs -- California - Southern; Intrusion (Geology); Geochemical Modeling;

Subject - Local

Igneous rocks -- analysis; Mesozoic Cordilleran Arc; Peninsular Ranges batholith; Geochemical data; Magmatic processes; Plate Tectonics



Page Count


Digital Format


Digital Publisher

Loma Linda University Libraries

Usage Rights

This title appears here courtesy of the author, who has granted Loma Linda University a limited, non-exclusive right to make this publication available to the public. The author retains all other copyrights.


Loma Linda University Electronic Theses and Dissertations

Collection Website



Loma Linda University. Del E. Webb Memorial Library. University Archives