Estimation of Soil Shear Strength Indicators Using Soil Physical Properties of Paddy Soils in the Plastic State

Saturated soil shear strength is a primary factor that reflects the driving resistance of agricultural machinery in paddy soils. The determination of soil shear strength indicators, such as cohesion and internal frictional angle, is crucial to improve the walking efficiency of agricultural machinery in paddy soils. However, the measurement of these indicators is often costly and time-consuming. Soil moisture content, density, and clay content are crucial factors that affect the cohesion and internal friction angle, while very limited studies have been performed to assess the interactive effects of the three factors on soil shear characteristics, especially on paddy soils. In this study, eight soil samples were taken from eight paddy fields in Southeastern China, and the central composition rotatable design was used to classify the soil samples into five levels based on different clay content (X1), moisture content (X2), and density (X3). The direct shear tests were carried out indoors on the remolded paddy soil using a self-made shear characteristic measuring device. Then, both individual and interactive effects of X1, X2, and X3 on soil cohesion and internal friction angles on paddy soils were systematically investigated and analyzed using the regression analysis method in the data processing software Design-Expert. Our results indicated that the effects of the three environmental factors on soil cohesion were in the order of X1 > X2 > X3, while the order was X2 > X3 > X1 for the impact on internal friction angle. The interactive effects were in the order of X1X2 > X1X3 > X2X3 for cohesion and X1X2 > X2X3 > X1X3 for internal friction angle. Two prediction models were successfully established to quantify the soil cohesion and internal friction angle as affected by soil physical properties, and the coefficient of determination (R2) was 0.91 and 0.89 for the two equations, respectively. The model validations using new soil samples suggested that the models were capable of predicting the shear characteristic parameters under different physical parameters effectively, with errors between predicted and measured soil shear strength indicators within 15% and relative root mean square error less than 11%.