Mineralogical and Bulk-Rock Geochemical Signatures of Ringold and Hanford Formation Sediments

Xie, YuLong ; Murray, Christopher J. ; Last, George V. ; Mackley, Robert D.

Pacific Northwest National Laboratory (U.S.)

关键词: Hanford;    Sediments;    Mineralogy;    Classification;    Boreholes;   

DOI  :  10.2172/15010111
RP-ID  :  PNNL-14202
RP-ID  :  AC05-76RL01830
RP-ID  :  15010111

美国|英语

来源: UNT Digital Library

【 摘 要 】

Pacific Northwest National Laboratory (PNNL) conducted this study for the Groundwater Protection Project managed by Fluor Hanford, Inc., and the U.S. Department of Energy (DOE). The purpose of the study was to explore the possibility of using statistical methods to classify sediment samples belonging to the Hanford and Ringold formations using mineralogy and geochemistry data. Visual examinations of the sediment do not always conclusively identify the geologic formation to which a sample belongs. The Hanford and Ringold formations constitute the majority of the vadose zone sediments within DOE's Hanford Site. They are particularly important because of the control they exert on contaminant transport in the subsurface environment. Although much is known about the formations from geologists' studies, quantitative methods for distinguishing between them had not been developed. In this study, PNNL constructed a database consisting of existing mineralogy measurements made using electron microprobe (EM) and petrographic (Petro) techniques and geochemical compositions measured by x-ray fluorescence (XRF). Data were concentrated in the 200 West and 200 East Areas of the Hanford Site. Statistical techniques were used to examine the distributions of the variables. Principal component analysis was used to examine the multivariate structure of the data. In addition, several classification methods were used to identify classification functions that could be used to distinguish between the two formations. Based on the results of this study, we concluded that principal component analysis, discriminant function analysis, and machine learning methods were valuable additions to the standard toolbox in quantifying the explanatory value of mineral and chemical variables and determining which are most effective in distinguishing between the Ringold and Hanford stratigraphic units. Given the differences in chemical and mineral composition for the Hanford and Ringold formations identified in this study, and the control that compositional differences often exert on Kd's for different contaminants, it appears that routine measurement of chemical and mineralogical composition from new Hanford boreholes would make it easier to classify Hanford and Ringold sediments and also provide data that would be useful for predicting the Kd values for the sediments.