Structural heterogeneity of soil clods: Correlating Weibull parameters to fracture surface topography

Abstract

Planes of weakness (i.e. cracks) existing in the soil matrix propagate, creating fracture surfaces and fragmenting the clod into smaller clods. The cracking process defines the fracture topography (i.e., roughness) that control many processes from flow exchange between macropore and matrix to local carbon storage and turnover. To date, useful relations between the surface topography of fractures and hydro-mechanical soil properties are missing because of technical and conceptual limitations. The objective was to test the assumption that the Weibull modulus of clod tensile strength can be related to the surface topography of fractures that resulted from the breakage of standardized soil clods. For carrying out standardized indirect tensile strength tests, a tension testing machine was constructed to prompt and monitor aggregate rupture. A total of 280 clods of 4 size-classes from two contrasting soils were considered. Clods were from the finer-textured Bt horizon of a Haplic Luvisol and the coarser-textured C horizon with lower amount of organic compounds of a Haplic Regosol. The natural clods were standardized in a cylindrical shape, in order to capture differences in the structural heterogeneity in terms of particle binding forces and porosity quantified by the Weibull modulus (m) of tensile strength. The stronger hierarchical organisation of the structure was quantified by smaller values of m (Bt: 3.1 vs C: 3.6) and the mass fractal dimension (Bt: 2.93 vs C: 2.98), and by higher values of the friability (m−1) (Bt: 0.14 vs C: 0.06). From these results, the random forest ensemble learning algorithm showed that the spatial heterogeneity of the organo-mineral pore coating (O-M) (shape factor of frequency distribution Bt: 0.58 vs C: 0.69), and not the organic carbon content (OC) (Bt: 0.48 vs C: 0.1 %), was the most important variable for predicting tensile strength (importance O-M: 0.21 vs OC: 0.09). Furthermore, the greater the spatial heterogeneity of the clod (m values ca. 1.4 for Bt) the lower the fracture surface roughness (ca. 1.8 µm2 µm−2) when compared to a more homogeneous clod (m values ca. 3.6, fracture surface roughness ca. 2.7 µm2 µm−2 for C). According to the random forest ensemble learning algorithm, porosity was the most important variable (variable importance 0.95) for predicting the rupture surface roughness followed by the spatial heterogeneity of the organo-mineral pore coatings, water content, and soil texture. Fracture surface roughness decreased when increasing porosity or decreasing the shape factor of the frequency distribution of organo-mineral coatings. Results confirmed that the Weibull modulus of clod tensile strength can be related with clod fracture surface roughness resulting from standardized clod breakage tests. This sheds light on the Weibull modulus as a distinct macro property for linking soil hydraulic and mechanical properties. This is of special importance in the parameterization of coupled models that combine the spatial heterogeneity of soil physical properties with the effects of hydrostatic forces on the soil structure.