Abstract :
[en] Air permeability and water conductivity are fundamental physical properties when it comes to the soil functions across the environment. The water conductivity and the air permeability as functions of the soil’s degree of saturation (K(θ) and ka(ɛ), respectively) are only discretely measurable, and the use of models is necessary to obtain continuous expressions of these functions. Most models however consider the soil pore network structure as a fitting parameter although it is public knowledge that K(θ) and ka(ɛ) depend mostly on the soil microstructure, which is, none the less, unique between samples with homogeneous texture. New ways of studying K(θ) and ka(ɛ) are needed.
The direct soil pore space visualization is a promising avenue to lead us to objectifying soil physics. The X-ray microtomographic technique (X-ray µCT) is now widely used by soil scientists and delivers 3D grayscale images of objects composed by materials of different densities. When dealing with a porous medium such as the natural soil, the X-ray µCT images need to be cautiously and expertly processed to obtain realistic feature quantification. A parallel, but however perquisite, objective of this dissertation is to statistically compare the effects of various image processing on the final X-ray µCT image features quantification. We simulated grayscale images to be processed to conclude about the image processing methodology we applied in our research.
The overall objective of this dissertation is to explore the relationships between one microscopic soil structure (the volume of the smallest visible pore is 0.0004 mm³) and its macroscopic functionalities, such as its water conductivity and air permeability. More specifically, we confirmed that the use of 3D X-ray µCT data enables a better estimation of the soil water retention curve near saturation through the identification of the largest soil pores. These are indeed often by-passed with pressure plate’s laboratory measurements because of various artefacts. We also identified microscopic pore space morphological parameters that explained the soil saturated hydraulic conductivity, and microscopic porosity distribution measures that explained the soil air permeability.
The final X-ray µCT image features quantification depends on the applied image processing, as stated, but also, clearly, on the image resolution. We concluded that working with a higher resolution would not necessarily lead to a higher degree of knowledge because resolution is sample-size dependent, and one pore size distribution could moreover be sufficiently visible at low resolution. We however observed that the pore network morphological and topological connectivity increases with resolution. Finally, we highlighted the imperfections of the capillary theory applied to soil through scanning the same soil samples at various water contents. As hypothesized, the pore network connectivity seems to play an important role in the pore accessibility to draining.
After having studied the effects of the soil pore network structure on the soil hydrodynamic properties, we turned the question around and evaluated the effects of the chemical soil composition (organic carbon and free forms of iron) on the very same soil pore network structure.
This dissertation therefore discusses the advantages and limitations of the use of X-ray microtomography to study soils for a more realistic understanding of the soil hydropedodynamic processes.
