^{1,2}

^{3}

^{4}

^{1}

^{1}

^{1}

^{2}

^{3}

^{4}

This study models a dam-break flow over a bed by using a depth-averaged numerical model based on finite-volume method and computes the dam-break flow and bed morphology characteristics. The generalized shallow water equations considering the sediment transport and bed change on dam-break flow are adopted in the numerical model, and the vegetation effects on the flow and morphological changes are considered. The model is verified against three cases from the laboratory and field data documented in the literature. The numerical results are consistent with the measured results, which show that the model could accurately simulate the evolution of the dam-break flows and the morphology evolution of bed within a computational domain with complex plant distribution. The results show that the riparian vegetation in the waterway narrows the channel and reduces the conveyance capacity of river. The flood flow is diverted away from the vegetation community toward two sides and forms a weak flow region behind the vegetation domain. The resistance of plants markedly reduces the flow velocity, which directly alters the fluvial processes and influences the waterway morphology.

Increase in catastrophic flood events has attracted increasing attention on the prediction of their dynamics and bed change evolution, especially in relation to riverine plant effect. Usually, grasses, shrubs, and mangroves, growing in watercourses and floodplains, are key members of the water ecosystem. They play important roles in water purification, flood control, and maintenance of bank stability, but they also create nonpositive obstruction effect on flood propagation in waterways. Recently, interactions between flow, morphology, and aquatic plant have been studied in the laboratory and field-scale experiments. Manners et al. (2014) believed that rapid expansion of tamarisk led to a narrower channel, which pushed the Yampa River into a new equilibrium having altered fluvial processes [

During dam-break flow conditions, the flow velocity is large, the sediment concentration is high, and the bed varies so rapidly that their effects on the flow cannot be ignored [

In the present study, a depth-averaged hydrodynamic and sediment transport model, based on finite-volume method, is developed to simulate the flood waves and bed change due to vegetation effect. For improving simulation accuracy, local mesh at selected locations is refined by using triangular mesh. The proposed model is first used to calculate the dam-break flows over fixed bed and to compare the water levels and velocities with measured data. Then, the bed changes of dam-break flood are computed to investigate the temporal and spatial variability in the bed elevation. Finally, the hydrodynamic variation and the evolution of bed elevation through the rigid vegetation domain are investigated and discussed.

The depth-averaged shallow water equations were obtained by integrating the Navier–Stokes equations over the flow depth, consisting of continuity equation and momentum equations for depth-averaged free surface flows. The conservative and vector forms of the 2D shallow water equations are written in (

The sediment deposition

The bed deformation

The vegetation effect on the flow was included to the momentum equations as an internal source of resistant force per unit fluid mass. Therefore, the drag force exerted on vegetation per unit volume can be expressed as follows [

Equation (

A standard finite-volume method with Roe Riemann solver for wet and dry computation has been developed. The discretization of the governing equations was based on the finite-volume method, for which the unstructured triangular mesh is shown in Figure

Control volume as represented by triangular mesh.

In a 2D triangular grid system, the line integral term can be approximated and assessed as follows:

Riemann problems at each cell interface were solved by various Riemann approximations for assessing the interface fluxes. In particular, Roe’s approach was used in this paper. The interface flux of Roe’s solver was computed as follows:

The three distinct eigenvalues of

The sediment flux across the interface of two neighboring elements in (

A treatment technique for wet and dry boundaries was adopted to achieve the purpose of zero mass error [

Wet edge as shown in Figure

Partially wet edge (with flux) as shown in Figure

Partially wet edge (no flux): a wet cell (left) links to a dry cell on the right, and the water level of the wet cell is lower than that of the dry cell as shown in Figure

Dry edge as shown in Figure

Wet cell: all the edges of this cell consisted of wet edge or partially wet edge (with flux) and all the nodes of the cell were flooded.

Dry cell: all the edges of this cell consisted of dry edge or partially wet edge (no flux).

Partially wet cell: all other cells were not satisfying the criteria of wet and dry cell as defined above.

Schematic diagram of wet-dry fronts.

The numerical model has been validated against theoretical solutions (not the focus of the paper). In the following, a series of applications are illustrated to appreciate the model applicability to real cases for which experimental data are available in the literature. First study considered the failed Malpasset dam that was once located in a narrow gorge of the Reyran river valley in Southern France with a width 500 m. The dam site was located 12 km upstream of Frejus estuary [^{6} m^{3}. In the immediate downstream of the dam, the Reyran river valley was very narrow and had two consecutive sharp bends. Then the valley widened as it went downstream and eventually reached a flat plain. The dam failed in 1959 following exceptionally heavy rain. After the dam failure, a field survey was performed to obtain the maximum water level along the Reyran river valley. In addition, a physical model with a scale of 1 : 400 was built to study the maximum water level and the flood wave arrival time at nine points along the river valley in 1964. Because of its complex topography and availability of measured data, the Malpasset dam-break case was selected as a test example for the present dam-break model.

In this computation, 11800 grid cells were composed of the refined mesh sizes at the dam site, main river channel, and downstream valley. The initial water level in the reservoir was set at 100 m above sea level. The rest of the computational domain was considered as dry bed. Manning’s coefficient was 0.019 over the entire computational domain. The time interval

Calculated and measured water front arrival times at each measuring point.

Calculated and measured water levels of the maximum flood at each measuring point.

Water depth evolution at times

For the proposed dam-break model, an idealized test on dam-break flow over a mobile bed was numerically investigated to examine the capability of capturing the wet-dry boundary and bed change. The experiment performed by Capart and Young [

Comparison of the computed and measured water surface and bed levels.

An experiment of dam-break flow over mobile beds was adopted to test the capability of the present model by calculating the interactions on flow, sediment transport, and bed change [^{3}, and porosity of 0.47 and the Manning’s coefficient was 0.022. An initial water depth was 0.25 m at the upstream of the gate. As shown in Figure

Sketch of a dam-break flow experiment over a mobile bed.

Triangular mesh near enlarged channel.

Comparison between the observed and calculated water levels.

Comparison between the observed and calculated final bed levels for cross-sectional profiles.

In order to investigate the interaction of flood, bed morphology, and floodplain vegetation for dam-break flow during a dam-break event, the sudden enlargement case in the previous section was recalculated by adding vegetation in a circular zone (the center of the circle had

Vegetation community arrangement.

Comparison between the observed and calculated final bed levels for cross-sectional profiles.

Comparison between the calculated final bed levels with and without vegetation.

Hydrodynamic and morphology processes for a dam-break flow.

In this study, an accurate and efficient depth-averaged hydro-morphodynamic model was developed based on finite-volume method using the unstructured triangular grid. In solving the associated equations, a framework of the fully coupled procedure was deployed with flow, sediment, and bed change computed simultaneously in the entire computational domain. The intercell fluxes were evaluated based on Roe’s approximation of Riemann solver with second-order accuracy as obtained by employing a MUSCL reconstruction technique, which provides an accurate description of flow near the moving waterline for dry and wet boundaries. The hydrostatic pressure was put into the source term of momentum equations such that it successfully eliminated the numerical imbalance between the source and the flux terms [

The authors declare that there are no competing interests regarding the publication of this paper.

This work was supported by the National Natural Science Foundation of China (51579030), the Program for Liaoning Excellent Talents in University (LJQ2013077), the Liaoning Natural Science Foundation (2014020148), and the Open Fund of the State Key Laboratory of Hydraulics and Mountain River Engineering (SKHL1517).