Flow-driven soil erosion processes and the size selectivity of eroded sediment on steep slopes using colluvial deposits in a permanent gully

- Erosion mechanism of colluvial deposit in overland flow were investigated.
- Sediment transport capacity (Sc) was more affected by flow rate than by slope.
- Sc can be estimated by mean flow velocity and unit stream power.
- Runoff, Sy and Sc can be forecast using unit flow discharge and slope gradient.
- Provided initial parameters for colluvial deposit erosion model in permanent gully.

Colluvial deposits with loose, coarse material are easily erodible in permanent gullies, but the mechanisms of erosion and sedimentation during overland flow remain obscure. Hence, the processes and mechanisms of the transportation of soil particles by overland flow were investigated in this study. Experiments were carried out in a 5.0 m long by 1.0 m wide flume using colluvial deposits. The slope gradient varied from 36 to 84%, and the flow rate ranged from 0.72 L m− 2 min− 1 to 2.88 L m− 2 min− 1. The runoff rate and sediment yield rapidly increased with increasing overland duration. Runoff and sediment were highly variable when the flume was treated with a high flow rate compared with a low flow rate, with the fluctuation of sediment concentration under the high flow rate usually reaching 500 g L− 1. The slope gradient and overland flow rate have strong impacts on sediment transport capacity. The mean flow velocity and the unit stream power can be an optimal composite force predictor for estimating sediment transport capacity. Experimental results also revealed that the percentage of gravel-sized particles increased with increasing flow rate and slope gradient, but silt and clay fractions observed opposite trend. The average enrichment ratio (ER) of gravel was usually < 1. However, the ER of silt and clay fractions decreased following the increase of the flow rate and slope gradient, and the values stabilized in the range of 1.1 to 1.4. The predicted values of runoff, sediment yield, and sediment concentrations were simulated by an empirical equation with measured values, and the regression coefficients were 0.983, 0.996 and 0.877, respectively. When the flow rate was > 2.16 L m− 2 min− 1, the bed load transport became an important mechanism; however, the simulation model overestimated these values.