LAUSR

A deeper understanding of the rainfall–flow processes can improve the knowledge of the rain-driven erosional processes in coarse-textured agricultural soil. In this study, on the red soil slope farmland developed from weathered granite, a simulated rainfall experiment was conducted to study the characteristics of rainfall redistribution, the processes of surface–subsurface flow generation and prediction, and sediment production. Rainfalls with three intensities of 45, 90, and 135 mm h−1 with a duration of 90 min were applied to the weathered granite red soil with the slope gradient of 10°. Under 45 mm h−1 rainfall intensity, the output of rainwater was composed by subsurface flow and bottom penetration, accounting for 35.80% and 39.01% of total rainfall, respectively. When the rainfall intensities increased to 90 and 135 mm h−1, the surface flow became the main output of rainwater, accounting for 83.94% and 92.42%, respectively. Coarsened soil exhibited strong infiltration-promoting but poor water-storage capacities under light rainfalls. With an increased rainfall intensity, the surface flow coefficient increased from 19.87% to 92.42%, while the amount of subsurface flow and bottom penetration decreased by 1.3 and 6.2 L, respectively. For sediment production, the sediment concentration was raised from 1.39 to 7.70 g L−1, and D10, D50, and D90 increased by 1.50, 1.83, and 1.40 times, respectively. The content of coarse particles (>1 mm) in surface soil increased by 12%, while the content of fine particles (<0.5 mm) decreased by 9.6%. Under strong rainfalls, severe soil and water loss, coarsening soil surface, and large loss of fine particles became major problems. During rainfall, the subsurface flow and bottom penetration could be predicted well through quadratic equations of rainfall time, which transformed into time-dependent exponential functions after rainfall. The results provide a theoretical basis and data reference for soil erosion prevention and water management in coarse-textured agricultural lands.