Principal Component Analyses of Potential Repository Groundwaters
- PDF / 3,355,253 Bytes
- 8 Pages / 384.84 x 608.76 pts Page_size
- 29 Downloads / 243 Views
ABSTRACT The chemical composition of ambient groundwater for a geological, high level radioactive waste repository is of crucial significance to issues such as radioelement solubility limits, sorption, corrosion of the overpack, behavior of compacted clay buffers, and many other factors involved in repository safety assessment. At this time, there are no candidate repository sites established in Japan for the geological disposal of high-level radioactive waste, and only generic rock formations are under consideration. It is important that a small, but representative set of groundwater types be identified so that defensible models and data for generic repository performance assessment can be established. Over 15,000 separate analyses of Japanese groundwaters have been compiled into a data set for the purpose of evaluating the range of geochemical conditions for waste repositories in Japan. This paper demonstrates the use of a multivariate statistical analysis technique, principal component analysis (PCA), to derive a set of statistically based, representative groundwater categories from the multiple chemical components and temperature that characterize the deep Japanese groundwater analyses. PCA also can be used to guide the selection of groundwaters that could be used in scenario analyses of future geological events in Japan. ANALYTICAL APPROACH Binary plots of chemical variables (Xi vs. X) can he used to evaluate the deep Japanese groundwater data set. Such plots can provide useful visual correlations and trends. However, they are limited in defining bounding or representative groundwater compositions for two reasons. First, changes of groundwater chemistry evolve through multi-dimensional chemical space; i.e., there are simultaneous co-variations in the concentrations of all cations and anions related to complex rock-water interactions. Binary chemical plots provide accurate, but possibly misleading, two-dimensional projection of these much more convoluted multi-dimensional trends. Second, specific groundwater samples that define bounding values in one binary plot are not necessarily the same samples that define bounding values for some other binary plot. For example, it is not generally defensible to simply combine the minimum values for Na' and Clfrom one binary plot with the minimum values for Ca"+and HCOA- from another binary plot, and then assume that they define the representative composition of dilute groundwater. This approach is only defensible if the samples representing the minima for all components from the two plots are the same. Other commonly used diagrammatic techniques for the classification and interpretation of hydrochemical data, such as Piper diagrams [I] and Schoeller diagrams [2], also suffer from various drawbacks. For example, Piper diagrams use percentages of ions rather than actual concentrations while Schoeller diagrams are only suitable for a small number of samples. Multivariate data analysis techniques, such as factor analysis, have been shown to be more effective in graphically pre
Data Loading...
