Hydraulic Conductivity Distribution from Multi-Level Slug Tests and Multivariate Facies Associations in a Conglomeratic Fluvial Aquifer, Boise Hydrogeophysical Research Site

Knowledge of the distribution of hydraulic conductivity (K) at high-resolution in heterogeneous aquifers is important for modeling fundamental hydrologic processes, investigation and remediation of groundwater contamination, and understanding petrophysical relations or multivariate associations. We examine K structure in the conglomeratic fluvial aquifer at the Boise Hydrogeophysical Research Site (BHRS) using high-resolution K data from multi-level slug tests modeled with updated wellbore-skin K. The K data are analyzed both alone and in combination with porosity (φ), capacitive conductivity (CC), and grain-size distribution (GSD) data. K population pdfs (probability density functions) by φ-CC-lithologic stratigraphic units follow the well-established BHRS stratigraphy only in the lower half of the section, but have different rank order of relative magnitude compared with φ and CC.MANOVA verifies the presence of seven multivariate K-facies including one repeated type, and t-tests for K alone recognize six univariate K-facies, with two of the seven multivariate K-facies (distinguished largely by CC differences) combined into one type of univariate K facies. Crossplots indicate K-facies exhibit multivariate parameter associations rather than correlations or petrophysical relations. Principal component analysis of K parameters with φ and CC (4-way), and also with GSD information (8-way), indicate nearly half the variance is related to φ and CC, with little influence from K. Significant fractions of the remaining variance are “flow related,” with K and φ both varying together and in opposition. Such K-φ “polarity” explains the stratigraphic division of K-facies into three lower facies with K-φ generally varying together, and three upper K-facies with K-φ generally varying in opposition. GSD information indicates K variation occurs by a range of combinations of φ with cobble and matrix fractions, representative grain size parameters, and sorting.