Spatial-Statistical Properties of Geochemical Variability as Constraints on Magma Transport and Evolution Processes at Ocean Ridges

Martin Lee Collier

Spatial-Statistical Properties of Geochemical Variability as Constraints on Magma Transport and Evolution Processes at Ocean Ridges
Collier, Martin Lee
Thesis Advisor(s):
Spiegelman, Marc W.
Earth and Environmental Sciences
Permanent URL:
Ph.D., Columbia University.
The research presented in this thesis employs spatial and statistical properties of major element variability in basaltic lavas and mantle residues to constrain some of the processes and dynamics occurring beneath ocean ridge magmatic systems. Ocean ridges represent a critical setting for many geochemical fractionation processes involved in the chemical evolution of the silicate Earth, and are fundamental to the plate tectonic cycle. Because of the inherent inaccessibility, it remains an ongoing challenge to interpret the geochemistry of ocean ridge lavas and exposed mantle residues in order to extract information about the petrogenetic and geodynamic workings of ocean ridge magmatic systems. This endeavor continues to require a concerted effort, incorporating field work, laboratory experimentation and quantitative modeling, in which the identification of features in the spatial or statistical distribution of geochemical variability represents an important contribution. In the three main chapters of this thesis, I apply techniques of exploratory data analysis, computational statistics, and petrologic modeling to develop original ideas about the relationship between sampled major element variability and the effects of specific processes, both petrogenetic and scientific: crystallization, melt transport, and sampling. In Chapter 2, I use spatial patterns of mid-ocean ridge basalt (MORB) glass variability to test competing hypotheses about crystallization in the thermal boundary layer beneath ocean ridges. I develop the hypothesis that reactive crystallization (crystallization influenced by chemical exchange with surrounding peridotite) could result in a different geochemical evolution of crystallizing magmas than expected for fractional crystallization. According to this hypothesis, fractionation-corrected MORB variability could be caused largely by sample-to-sample variations in the relative extents of reactive versus fractional crystallization. I demonstrate that MORB major element variability observed within 30-km-scale spatial bins contains 40-70 percent of globally observed variability, consistent with the predicted effects of reactive crystallization, but inconsistent with mantle temperature variations. Chapter 3 considers the effect of spatially heterogeneous sampling on apparent variability in MORB glasses. I demonstrate that MORB variability, as represented by the PetDB MORB glass database, contains large variability in sampling density, leading to significant artifacts in the estimated relative frequency of different MORB compositions. I introduce a method for removing these artifacts, and show that the increase in MORB data availability over the past decades has not been sufficient to increase significantly the resolution with which major element variability systematics can be studied at global or regional length scales, at least in comparison to early syntheses of global MORB data. Chapter 4 examines statistical variability within spatially defined volumes of mantle residue exposed in the Oman ophiolite. I provide a preliminary map of intermediate-scale compositional variability within the southernmost Oman ophiolite massif, in which multiple, spatially coherent, compositionally distinctive, 20-100 sq. km regions are resolved, representing the first mapping of compositional mantle domains at this length scale anywhere in the world. I interpret the observations as the consequence of regionally distinctive internal proportions of different mantle lithologies (e.g., dunite versus harzburgite), in turn reflecting the organization of focused melt transport at mid-ocean ridges into channel-rich and channel-poor zones.
Item views:

In Partnership with the Center for Digital Research and Scholarship at Columbia University Libraries/Information Services.