Flume experiments examining the effects of a flexible mattress on bed scour around spur dikes were conducted. Results showed that scour induced subsidence damage to the flexible mattress. If the width of flexible mattress, $B$ , was insufficient, then failure of the spur dike may occur. Results also revealed that (1) scour depth decreased as the area of the protective domain around the spur dike increased (e.g., it reduced 30%, from 20 to 14 cm, as the width $B$ increased from 0 to 0.35 m), and (2) higher approaching velocities resulted in deeper scour holes. On the basis of the experimental data and dimensional regression analysis, empirical formulas were proposed for calculating scour depth and the extent of the scour hole. The formulas were then used to obtain the optimum flexible mattress width.

Flash flooding is one of the most severe natural hazards and commonly occurs in mountainous and hilly areas. Due to the rapid onset of flash floods, early warnings are critical for disaster mitigation and adaptation. In this paper, a flash flood warning scheme is proposed based on hydrodynamic modelling and critical rainfall. Hydrodynamic modelling considers different rainfall and initial soil moisture conditions. The critical rainfall is calculated from the critical hazard, which is based on the flood flow depth and velocity. After the critical rainfall is calculated for each cell in the catchment, a critical rainfall database is built for flash flood warning. Finally, a case study is presented to show the operating procedure of the new flash flood warning scheme.

In small mountain catchments, the spatial and temporal resolution of rainfall can vary significantly across the catchment. However, rainfall gauging stations can be sparse in these regions, and collected data may not reflect the real rainfall distribution across the catchment. When modelling flash floods, finding a suitable approach to estimate the actual rainfall distribution is nontrivial. In this study, the effectiveness of different methods for obtaining a spatial and temporal rainfall distribution for use in numerical modelling of flash floods was investigated using a full two‐dimensional depth‐averaged shallow‐water hydrodynamic model. It was demonstrated that the Thiessen polygon method and the inverse distance weighted interpolation method (IDW), with appropriate empirical coefficients, produce results in agreement with observed stage and discharge hydrographs. We show that the uniform distribution method cannot be used to represent realistic spatial and temporal variability of rainfall for flash flood events in small mountain catchments. By combining available data with the common IDW method, missing rainfall timeseries data in a small catchment can be estimated, even for short‐duration time scales, such as a single flash flood event.

Dike breaching is the main component of flood defending system. To temporally enhance the capacity of the river, small levees will be build on the dike crown to increase the elevation of the dike crest. Also, the dike top usually be paved with concrete as transportation road. Under these two condition, break mechanism are different from those without complex measures on the crown, which has not been investigated yet. Large-scale experiment has been carried to investigate the breaching mechanism. Results show that with small levees, the flow forms a little fall at the downstream edge of the levee top and a much larger fall on the downstream face of the dike. The “headcut” backward retreat is the main breaching mechanism in the early stage of breaching. During the rapid development stage of breaching, the vertical erosion, lateral erosion and gravity collapse are the breaching mechanism. The existence of the external small levee protected the top of the dike from erosion for a long time, which largely delayed the breaching processes. As the top was paved, the breaching processes likes that of dike with small levees. Two falls forms at the edge of the road and at the downstream face respectively. When the backward retreat of “headcut” at the downstream face of the dike reaches the base of the dike, the underneath soil is washed away and lead to concrete of the road collapse. Once the road is collapsed completely, two falls merged into one, thereafter the breaching processes likes dikes without complex measures on the crown. The paved road also delayed the breaching processes. This study can provide scientific support to dike breaching emergency management.