Leveraging high spatial resolution bulk measurements to understand and characterize redox zonation and processes in natural sediment profiles.

Abstract:

X-ray absorption spectroscopy (XAS) has been used extensively over the last 20 years to characterize environmental samples. Recent advances allow us to measure large numbers of samples, either in microprobe-based arrays or mapping, or using high resolution (sub-meter) resolution sampling along environmental gradients. Our analysis of these large datasets is currently limited by our understanding and assertions about mineralogy and the reference spectra used to compose them. In this work, we leverage large XAS spectral datasets of reference materials and samples from North American and Asian sedimentary environments using a suite of unsupervised and supervised chemometric approaches including principal components combined with unsupervised hierarchical clustering, random forest modeling, and other approaches. We hypothesize that this analysis will reveal otherwise obscured patterns and trends in mineralogical composition between samples, and in the processes that alter those sediments. Our accrued datasets reveal consistent variability between samples at a variety of scales that can be attributed to the river basin identification, sediment source or provenance, and chemical processes operating in those sediments. Many of these processes are not apparent using smaller datasets. This work has relevance to understanding flow paths more clearly, in identifying authigenic processes active in specific sediments, and in delineating sediment facies. Ultimately, our analysis relates to differences in aqueous composition that are not apparent using aqueous geochemical information alone. We suggest that more effort is taken to unify geological datasets to permit the community to more consistently leverage data for future applications.