Splash erosion of clay–sand mixtures and its relationship with soil physical properties: The effects of particle size distribution on soil structure

• Soil structure of engineering accumulations in different textures was investigated.
• Soil compactness was determined by the fabric and rearrangement of PSD.
• Cohesion was better quantified by particle fractal dimension than clay content.
• Splash erosion temporal variation was related to cohesion and non-capillary porosity.

Large quantities of disturbed soils and residues derived from engineering constructions have provided material resources for soil erosion and geological disasters, which is becoming a serious problem in China. To date, studies on the erosion mechanism of these soils in a wide particle size distribution (PSD) are limited, and the relationships between soil PSD and physical characteristics are not explicit. Here, we analyzed the effects of PSD on structural and mechanical parameters and attempted to quantify their relationships by blending silty clay soil and engineering sand as clay–sand mixtures with various sand mass proportions ranging from 0 to 90% at a 10% increment in a wet and a dry mixing type (MT). In addition, the effect of PSD on splash erosion was measured under simulated rainfall. Changes of bulk density and porosity of soil mixtures (for clay contents more than 13%) with clay content displayed apparent differences separately in the wet and dry mixing types mainly due to the deformation and reorganization of soil grains. Internal friction angle φ and cohesion c of saturated soils showed a parabolic trend and an exponential increase with clay content respectively, while cohesion c was better quantified by particle fractal dimension D than clay content. Shear strength indexes in the wet mixing type were generally larger than those in the dry mixing type when clay content was larger than 13%. Splash erosion temporal variation was significantly affected by PSD, MT, rainfall duration (RD) and their interaction effects (PSD × MT, PSD × RD), among which MT was the most prominent source of variation (F = 99.97, p < 0.001). Stepwise multiple regression analysis indicated that splash erosion rate was negatively correlated with cohesion (p < 0.001) and the coefficient of variations (CVs) of splash erosion rate were positively related with non-capillary porosity (p < 0.01). Simple equations were proposed for predicting physical parameters (bulk density ρ, porosity e, φ, c, splash erosion rate) as a function of soil PSD characteristics. These obtained results are useful to supplement the erosion knowledge of disturbed soils, facilitate soil erosion prediction and provide technical guidelines for soil and water conservation in engineering construction areas.