Scientific Framework For Subsurface Characterization and Evaluation of Grain-Size Analysis Methods

Abstract

Listen

Translate article

Subsurface formations are highly variable with depth and space, which poses challenges in their characterization at a catchment scale. This study demonstrates a scientifically sound approach for lithology exploration and the relative performance of salient grain-size analysis (GSA) methods in estimating hydraulic conductivity of subsurface formations. Test drillings were carried out at eight locations, and geologic samples were collected at a regular interval of 3.05 m (10 feet). All the geologic samples were subjected to sieve analysis. Thereafter, grain-size distribution curves (GSDC) were prepared for individual geologic samples. The physical characteristics of subsurface formations such as porosity, uniformity coefficient, and effective grain diameter were computed using GSDC and empirical formulae. Hydraulic conductivity of subsurface formations was estimated by using four popular GSA methods, and their relative performance was evaluated. The analysis of GSDC revealed that the geologic samples at deeper depths (≥40 m) have S-shape curves compared to the geologic formations present at shallow depths (<40 m). The values of uniformity coefficient range from 2.43 to 16.94 for shallow-depth geologic samples, whereas they vary from 2.24 to 11.40 for deeper depth geologic samples. The lithologic analysis indicated that shallow aquifer layers exist at depths ranging from 9 to 40 m with thickness varying from 3 to 2626 m. Besides shallow aquifers, deep aquifer layers of 3 to 34 m thickness exist at 40 to 79 m depths. The average values of hydraulic conductivities (K) based on the four GSA methods range from 2.52 to 208.38 m/day for the deeper aquifer layers and 0.98 to 285.5 m/day for the shallow aquifer layers, thereby suggesting considerable aquifer heterogeneity. Among the four GSA methods, the Hazen and Kozeny methods consistently overestimate the mean K with higher standard deviation at all the sites, while the Slitcher method provides the least mean K values with a less standard deviation. These estimates are useful for a preliminary assessment of subsurface K in the absence of field-based estimates.