Extracting the 3D spatial structure of soil cracks in the North China Plain from paraffin casting and exploring the developmental patterns of vertical cross-sectional morphology of the cracks.

• Use paraffin casting and 3D scanner to reconstruct the cracks in North China Plain. • The vertical profile of the crack is neither a V-shaped nor a cuboid-shaped, but an arc-shaped. • The normalized crack width versus depth obey power function relationship. • Soil moisture profile and the SSCC fail to predict the concave crack profile. Cracking from soil desiccation is a common phenomenon. Understanding the structure of cracks is highly important in the research of soil moisture and heat transportation, as well as the strength of soil structure within soil. There is limited research on the three-dimensional (3D) spatial characteristics of soil desiccation cracks. Furthermore, the absence of a dependable and precise in-situ technique for extracting and digitizing the three-dimensional crack structure means that the commonly employed simplified crack models (such as V-shaped and cuboid-shaped) have received limited testing. This study examines the feasibility of extracting the 3D structure of cracks by combining in-situ paraffin casting in sandy loam soil of North China Plain with a 3D laser scanner. Parameters (length (L), width (W), depth (D), depth to width ratio (C)), as well as the vertical cross-section profile (W ∼ D) of each crack, were used to quantitatively describe the structure of cracks. Our study demonstrated that digitalized 3D cracks with 5 μm spatial resolution can be obtained quickly by paraffin infusion combined with a 3D laser scanner. Our result based on 18 crack cross-section profiles also demonstrates that W ∼ D do not follow either the V-shaped crack model or the cuboid-shaped crack model. All of the 18 normalized vertical profiles of cracks obey the power function relationship (R2 = 0.850) rather than the linear relationship of the V-shaped model (R2 = 0.794). The shape of the soil moisture profile is convex, which contradicts the concave profile of the observed vertical cracks. Even after combining the Soil Shrinkage Characteristic Curve (SSCC) with the soil moisture profile, we still failed to replicate the observed concave crack profile. The observed crack morphology in experiments can be replicated by both the elliptical model and the parabolic model of fracture mechanics, albeit with less predictive accuracy when compared to the power function model. This suggests that the current understanding of soil desiccation cracking in soil science is deficient, as it lacks appropriate methods and theoretical mechanisms for investigating soil cracking. [ABSTRACT FROM AUTHOR]