沟道侵蚀是黄土高原水土流失的主要形式之一。伴随着沟道侵蚀，陡坎在黄土源区沟道和淤地坝溃口等处广泛发育，是一种重要的沟床地貌形态，在地貌演变过程中起着关键的作用。目前关于黄土区陡坎发育的物理机制和水沙调控作用的研究甚少；对于人工陡坎淤地坝拦沙失效现象及溃决后库区泥沙释放过程背后的机制尚不清楚。本文将自然与人工陡坎纳入统一物理力学框架，系统研究陡坎形貌特征、演化机制与水沙调控作用，对流水地貌学和流域泥沙动力学具有重要的科学意义，对黄土高原水土保持和流域治理具有指导价值。通过野外测量获得了黄土高原周期性陡坎的形貌数据。结合水跃型陡坎模型和跌水池模型的优势，建立了考虑跌水池机制的周期性陡坎理论模型。黄土、基岩与粘性沙床面的陡坎形貌数据分析和数值模拟结果表明，陡坎高度和长度随流量的增大而增加，陡坎长度与高度的比值随河床可侵蚀性的增加而增加；新建立的周期性陡坎模型能较好地复演陡坎长度、高度、长度与高度比值随坡度的变化趋势，但难以解释陡坎长度与高度比值的实测大范围波动。研究发现陡坎演变时间的差异是陡坎长度与高度比值大范围波动的重要原因。通过数值模拟探索陡坎的演化机制，发现由于黄土沙粒沉降作用弱，沟道床面形态的演化以对流机制为主导，扩散作用弱，故黄土高原陡坎在迁移过程中得以稳定存在。陡坎发育使得局部水头损失占主导，其能量损失比例基本大于0.8。有陡坎和无陡坎发育的平均河床下切速率的比值基本小于0.1，表明陡坎发育能显著抑制河床下切。野外观测和数值模拟结果一致表明，淤地坝溃决后只有少部分泥沙释放出库。泥沙出库比较小的原因是由于溃口演化过程中形成的陡坎，增加了水流形貌阻力，降低了输沙浓度。采用沟道非平衡输沙模型，计算了淤地坝对次洪过程的水沙调控作用。结果表明，无坝情形下计算域出口输沙模数与径流侵蚀功率的幂函数关系不能推广至有坝情形；对一般洪水至特大洪水，拦沙失效淤积比例在88.3%~93.6%之间，工程中的经验常数值仅适用于小洪水情形。

Gully erosion is the main component of the total soil erosion in the Loess Plateau. The step which widely exists in the loess gully and at the check dam breach is an important gully bedform, and play a key role in the landform evolution process. To our knowledge, study on the physics of step formation and the water-sediment regulating effects of the step is limited. The check dam, as an artificial step, has similar characteristics to natural steps. The physics behind the failure on sediment interception of the check dam and the sediment release process after dam failure are still unclear. This study incorporates natural and artificial steps into a unified physical framework, and systematically studies the morphological characteristics, evolution mechanisms and water-sediment regulating effects of the step in the Loess Plateau. The study is of great value to geomorphology and watershed sediment dynamics, and is useful for controlling soil erosion and managing watershed in the Loess Plateau.The study collected the morphological data of field cyclic steps in the Loess Plateau. By combining a theoretical model of cyclic steps based on hydraulic jump hypothesis and a conceptual model of plunge pool erosion, this paper establishes a theoretical model of cyclic steps that considers the mechanism of plunge pool erosion. The morphological dataanalyses of the steps from loess, bedrock and cohesive beds and the numerical simulations consistently show that the step length and the step height increase with flow rate, and the ratio of step length to height increases with bed erodibility. The proposed model is capable of describing how the step length, height and the ratio of step length to height change with average channel slope. But it fails to explain the large fluctuation of the ratio of step length to height in the measured data. It may due to the fact the posed theory relates to equilibrium conditions and thus cannot consider temporal adjustments in step geometry.The paper conducted numerical simulations to explore the evolution mechanisms of cyclic steps and found that the evolution of bed morphology is dominated by the convection due to the negligible loess deposition, and thus the step will not disappear during the upstream migration. The development of the step in the gully make local head loss a major part of the total head loss, which is manifested as the proportion of the local head loss in the total head loss exceeds 0.8. Moreover, the step also inhibits the undercutting of the gully bed with the ratio of total degradation rate, with and without steps, smaller than 0.1. The field measurement and numerical simulations of the check dam breach evolution consistently show that only a small proportion of stored sediments can be released after the dam break. This may be due to the fact that the step formed during the check dam breach evolution significantly increases the flow form resistance and reduce the sediment transport concentration.Through a non-equilibrium sediment transport model, the water-sediment regulating effects of the check dam during flooding are analyzed. The results show that, the power function relationship between the runoff erosion rate and the sediment transport modulus at the outlet of the study area with no check dam can’t be extended to cases with a check dam. The threshold siltation rate above which the check dam does not trap sediment during flooding changes from 88.3% to 93.6% for a catastrophic flood to a regular flood. This study shows that the widely used empirical constant threshold siltation rate is only applicable to small floods．