Abstract:
Abstract: Headcut erosion constitutes more than 60% of hillslope soil loss. Quantitative research on rill headcut erosion processes provides fundamental information for process-based erosion modeling. Due to the complicated headcut morphologies and flow regimes near a headcut, it is hard to accurately predict the erosion rate of a headcut in some soil erosion prediction models. Current knowledge on the impacts of headcut height, headcut advancing rate and secondary headcuts developed on well-formed rill channel on soil loss is limited. Thus, simulation experiments with pre-made initial headcuts (5 cm high) were designed to investigate the effects of inflow rate, slope gradient, headcut height and headcut number on rill head advancing process. Soil boxes (2.0 m long, 0.3 m wide and 0.5 m deep) with 2 slope gradients (15° and 20°) and 4 inflow rates (1.0, 2.0, 3.0 and 4.0 L/min) were subjected to upland concentrated flow. At slope lengths of 70 and 120 cm, 2 cameras were mounted 1.5 m over the soil box and were controlled by an infrared remote control to photograph simultaneously. High precision DEMs (digital elevation models) obtained by photogrammetry were used to detect changes of hillslope morphology and headcut advancing process. The results showed that sediment delivery and headcut advancing rate increased while initial headcut height and secondary headcut number did not strictly increase with the increase of inflow rate and slope gradient. Sediment delivery increased by 0.59-5.34 times and 14.0%-89.7% when inflow rate increased by 1 L/min and slope gradient increased from 15° to 20°, respectively. When inflow rate was equal to or smaller than 2 L/min, sediment delivery increased fast at the beginning and kept stable later. When inflow rate was greater than 2 L/min, sediment delivery kept increasing during the whole experiment. So, 2 L/min was a threshold of inflow rate that may cause rill head advancing rate to increase significantly. Compared to the treatments at 1 L/min inflow rate, the duration for a headcut retreating by a certain length (100 cm) was shortened by more than 12 min when inflow rate was greater than 2 L/min. The effects of slope gradient on rill head advancing decreased with the increase of inflow rate. Linear equation which included unit flow rate and slope gradient was used to predict the rill length time series. Relative errors between predicted values and observed values were smaller than 16% and both the R2 and the Nash coefficient values were greater than 0.95. Soil loss on the hillslope dominated by rill head advance was determined by headcut advancing rate, headcut height and the number of secondary headcuts developed below the initial headcut. Soil loss increased with the increase of headcut advancing rate or headcut height in a power function while showed a linear correlation with the number of secondary headcuts. Soil loss can be modeled with a non-linear regression equation with a determination coefficient of 0.932. Results of this study provide new knowledge on rill headcut modeling. It is recommended that upslope runoff should be intercepted and concentrated rill flow velocity should be reduced when soil and water conservation practices are designed on steep loessial hillslope.