2025-12-01 JOURNAL OF HYDROLOGY 2025 663(卷), null(期), (null页)
The proliferation of deep-filling engineering projects in the Loess Plateau of China (LPC) has significantly modified the microstructure of intact loess, thereby altering its macro-scale hydraulic behavior. While previous studies have characterized microstructure variations and associated permeability changes through mercury intrusion porosimetry (MIP) and scanning electron microscopy (SEM), critical gaps persist in three-dimensional (3D) microstructural differentiation and its mechanistic linkage to hydraulic permeability. To address this issue, we employed high-resolution X-ray micro-computed tomography (micro-CT, 1 mu m resolution) to characterize the 3D pore structure of intact and compacted loess specimens with varying dry densities (1.50-1.80 g/cm3), complemented by MIP pore analysis. Laboratory-measured vertical saturated permeability coefficients Ksat were compared with predictions from three models derived from three different types of pore data, namely the MIP pore/throat data (MIP-throat), the 3D pore structure data (CT-pore), and the converted throat data from the 3D pore structure (CT-throat). Horizontal Ksat was additionally evaluated through the CT-pore prediction method. The findings demonstrate that intact loess with a dry density of 1.54 g/cm3 possesses superior vertical Ksat in comparison to the compacted specimens, exhibiting a newly identified low-porosity and high-permeability behavior, which can be attributed to the vertically aligned, well-connected large pores during loess formation. For the compacted specimens, compaction-induced increase in dry density promotes structural uniformity, consequently reducing Ksat anisotropy between vertical and horizontal orientations. Notably, pore-throat-based models (MIP-throat and CT-throat) systematically underestimate Ksat, whereas CT-pore predictions show excellent agreement with experimental data (RMSE = 0.056). This discrepancy suggests that despite serving as primary flow channels, pore throats are insufficient to fully capture flow dynamics, underscoring the critical role of pore size and shape in influencing permeability.