Frontiers in Environmental Science (Apr 2018)
Can The Pore Scale Geometry Explain Soil Sample Scale Hydrodynamic Properties?
Abstract
For decades, the development of new visualization techniques has brought incredible insights into our understanding of how soil structure affects soil function. X-ray microtomography is a technique often used by soil scientists but challenges remain with the implementation of the procedure, including how well the samples represent the uniqueness of the pore network and structure and the systemic compromise between sample size and resolution. We, therefore, chose to study soil samples from two perspectives: a macroscopic scale with hydrodynamic characterization and a microscopic scale with structural characterization through the use of X-ray microtomography (X-ray μCT) at a voxel size of 21.53 μm3 (resampled at 433 μm3). The objective of this paper is to unravel the relationships between macroscopic soil properties and microscopic soil structure. The 24 samples came from an agricultural field (Cutanic Luvisol) and the macroscopic hydrodynamic properties were determined using laboratory measurements of the saturated hydraulic conductivity (Ks), air permeability (ka), and retention curves (SWRC). The X-ray μCT images were segmented using a global method and multiple microscopic measurements were calculated. We used Bayesian statistics to report the credible correlation coefficients and linear regressions models between macro- and microscopic measurements. Due to the small voxel size, we observed unprecedented relationships, such as positive correlations between log(Ks) and a μCT global connectivity indicator, the fractal dimension of the μCT images or the μCT degree of anisotropy. The air permeability measured at a water matric potential of −70 kPa was correlated to the average coordination number and the X-ray μCT porosity, but was best explained by the average pore volume of the smallest pores. Continuous SWRC were better predicted near saturation when the pore-size distributions calculated on the X-ray μCT images were used as model input. We also showed a link between pores of different sizes. Identifying the key geometrical indicators that induce soil hydrodynamic behavior is of major interest for the generation of phenomenological pore network models. These models are useful to test physical equations of fluid transport that ultimately depend on a multitude of processes, and induce numerous biological processes.
Keywords
