Ca River is one of the largest rivers in Vietnam. The river provides water, electricity, and navigation for millions of people living along the banks. Besides these great benefits, the river also poses many potential risks for people. In the case of flooding, this river can cause terrible damage, especially in the case of dike breach. Therefore, this article, by combining a field survey and mathematical simulation, has been conducted to present the results of the dike breach for the La Giang dike of Ca river, in Ha Tinh province, Vietnam. Via the field survey, potential dike breach locations were specifically identified, which helps minimize the number of calculation scenarios. The mathematical model was calibrated and validated with large floods in the area. The results from the mathematical model show that the model is consistent with the observation data, with the Nash index at good to very good levels. Based on the data, a series of simulations was performed to assess the dike breach consequences. In each case, the study provided details on the inundation area, the number of affected residents and the length of flooded road for each inundation level by administrative unit. Based on the calculated results, the degree and scope of consequence varied depending on the locations of the dike breach. The results of the study are the foundation for the building of a local database of flood and storm control. This is very useful information for the decision-makers to establish different response plans for different emergency cases.

  • The study conducted a survey to identify the dike break potential locations, which reduces the scenarios.

  • The study established a model to simulate hydraulic regimes with high reliability.

  • The results show that breach time development plays an important role.

  • The study provides detailed quantitative dike breach results, which is important information for the development of a disaster prevention plan for the study area.

Floods are a common type of natural disaster in the world. Damage caused by floods is huge (The Central Steering Committee for Disaster Prevention and Control 2018, 2019, 2020, 2021). One of the current solutions to minimize the impact of floods is the construction of flood-resistant dikes. This measure is remarkably effective, but it increases the risk to people in the protected area. In case the dike breaks, the damage to the residents in the protected area will be beyond imagination (Apel et al. 2009). Although the consequences of a dike breach are massive, the risk of occurrence is very high and Vietnam is not an exception. Most dikes in major rivers in Vietnam have been built for a long time. In spite of regular supervision and repair, the deterioration of the dikes is inevitable. According to Ministry of Agriculture and Rural Development of Vietnam (2023), there are currently 288 key locations on the main dikes. Most of these locations are often under the influence of flow or are built on weak geological foundations. This leads to phenomena such as erosion, and subsidence, causing damage to the dike. It is urgent to have solutions to reduce the damage caused by floods, especially in the context of the increasingly obvious impacts of climate change.

The solution to this problem often lies in dike protection. More specifically, flood control solutions in general and dike protection solutions in particular are often classified into two categories: structural solutions and non-structural solutions. Each type of solution has its own advantages and disadvantages. Structural solutions are often effective immediately but normally have high cost. Moreover, under increasingly severe disaster conditions, to completely control natural disasters by using structural solutions is often not feasible. In contrast, non-structural solutions such as natural disaster warnings, development of emergency response plans, and raising awareness for the community are often adaptive. A great advantage of such solutions is that they are suitable for countries with limited economic conditions like Vietnam. Among these proposed solutions, developing emergency response plans is an effective solution that has been applied in many parts of the world. In order to develop an effective response plan that meets the actual needs of the local area, detailed information about the impacts and affected subjects is essential. As a result, assessing the consequences of floods, especially in the case of a dike breach, is an urgent need of the locals.

Flood has been studied extensively in Vietnam, and the Ca River is not an exception (Tran et al. 2014; Nguyen et al. 2020). However, studies of flood as a result of dike breaches are still limited. In fact, research on dike breaches is a challenge. First, a dike breach has a very complex mechanism. According to ASCE (2011) in the case of a dike breach, the water level outside the river will remain or decrease very little. Even when the outside and inside water levels are equal, the size of the breach continues to increase. Even when the outside and inside water levels are equal, the size of the breach continues to increase. Besides, the different locations of the breach in the dike will also cause different consequences (Bomers 2021). However, this was not mentioned in many previous studies when the breach location was often assumed to be at a specific location (Zolghadr et al. 2011; Viero et al. 2013). Therefore, the study of different consequences resulting from different breach locations on a dike is extremely vital, it provides important information for flood prevention and mitigation. However, it is not feasible to study all possibilities along a long dike. A more practical approach is required, which is a huge gap in research.

Approaches to dike breach are commonly divided into three categories, namely parametric models, mathematical models, and experimental and mathematical integrated models. In parametric models, studies are often simplified by calculating breach parameters using empirical formulas. For example, Nagy (2006) determined the breach length based on 2,200 records of the Carpathian basin. The research results presented a correlation equation between the length of the breach and the height of the dike. Meanwhile, Danka & Zhang (2015b) built a multivariate correlation equation to determine three parameters, namely the maximum discharge through the breach, breach length, and breach depth. The main advantage of empirical formulas is their simplicity, but the results of the formulas only give the maximum values and do not determine the flow process through the breach. The second approach is mathematical models, which build the development process of the breach via time or the rate of soil erosion. These statistics are defined as the input for mathematical models. In this approach, the physical processes are not modeled. Instead, the water flow through the breach is calculated using simple fluid dynamics equations, such as flow through the weir and orifice. They do not improve the accuracy compared to the parametric model approach, but the breach flow hydrograph can be determined. Some case studies following this approach can be mentioned (Huthoff et al. 2015; Tadesse & Fröhle 2020; Bomers 2021), which provide promising results. The third approach is to combine both mathematical and experimental models (Schmocker & Hager 2009; Tadesse et al. 2017; Liu et al. 2019). The main advantage of this approach is the increased accuracy. The breach development process in this approach takes several factors into account, such as erosion, sediment transport, and slope stability. Besides, the calculated results from the mathematical model are verified with physical experiments. However, studies in this approach often take a lot of time, and effort, and are not flexible in the calculation cases. Which approach to choose depends on the type of analysis, data conditions, and simulation scale. In the case of a large-scale river basin, the approach using mathematical models is proved to be most appropriate. It can be flexible to design many scenarios to meet the requirements of disaster prevention.

The research area is the downstream area of the Ca River. The main flow of the Ca River in the downstream borders Nghe An and Ha Tinh provinces. In the scope of the study, Ca River has one primary tributary La River, and two secondary tributaries Ngan Sau and Ngan Pho. The dike system is shown in Figure 1. In the Nghe An province, the remarkableness dike is Ta Lam. While in the Ha Tinh province, La Giang dike is the most important dike in the system. In this study, we focus on evaluating a series of La Giang dike breach scenarios. La Giang dike was built in 1934, going through many stages with various ways of construction, such as upgrading, and reinforcing, so the quality of the dike is not consistent. The dike mainly passes through densely populated areas. It means that if the dike breach occurs on this dike system, it will cause extremely heavy consequences for the people living in the area. According to Ha Tinh Department of Agriculture and Rural Development (2023), if La Giang dike is broken, it will affect about 301,653 people, 48,401 hectares of farmland in Duc Tho districts, Hong Linh town, Can Loc district and a part of Thach Ha and Loc Ha districts. The infrastructure will also be under the influence, such as National Highway 1A, 8A, North-South railway, trans-Vietnam fiber optic cable route, and 500-KV, 220-KV power lines.
Figure 1

Research area.

This paper will present the results of the study on floods caused by the breach in the research area. In this study, the field survey method was conducted in combination with mathematical models to simulate floods for a series of dike breach scenarios at different locations. The results of the study aim to find the most feasible dike breach scenarios in the La Giang dike. Based on the simulation results, the study has made quantitative assessments of the number of people as well as detailed statistics on the flooded area by different land users. The results of the study will help the authorities have a basis to build response scenarios in emergency cases.

The research combined the field survey and the simulation of the dike breach with the mathematical model. Figure 2 presents the flow chart of this study.
Figure 2

The flow chart of this study.

Figure 2

The flow chart of this study.

Close modal

The study started by conducting a field survey along the La Giang dike. During the survey, the study collected information on the current status of the dike as well as high-risk locations. The information includes the erosion status of the dike, elevation along the dike, as well as information on the geology of the dike. These are all very important information to determine the high-risk locations of dike breaches. In addition, the study collected information from local officials who directly manage the dike, which increases the reliability of identifying potential locations of dike breaches.

Based on the information collected, several locations with a high risk of dike breach were identified. For each high-risk location, the study determined the development process of the breach. The breach development process is usually determined through the breach parameters. These parameters usually include the occurrence time of the breach, the breach development time, and the maximum width, and depth of the breach. According to Danka & Zhang (2015a), the multivariate regression equation of Danka & Zhang (2015b) provides good results for calculating the width and depth of the breach. Therefore, the width and depth of the breach were calculated using the formulas in (1) and (2). Due to the limited availability of data, there are currently few empirical works to determine the breach development time. According to the research results by Tadesse & Fröhle (2020), the breach development time did not have much influence on the degree of the flood. However, Vorogushyn et al. (2011) had contrasting opinions on this issue. The breach development time was assumed from 1 to 24 h for sensitivity analysis. In order to calculate the safety bias, the time of the dike breach was assumed to be at the moment of maximum water level at the location of the breach:
(1)
(2)
where w is the dike width (m); h is the dike height (m); m is a material coefficient (0.38 for coarse-grained soils, 0.42 for fine-grained soils, and 0.35 for organic soils); and t is a dike-type coefficient (0.94 for composite dikes and 0 for earthen dikes); f is the failure mechanisms (f = 0.74 for overtopping; 1.21 for piping; 0.81 for slope failure or horizontal sliding, and 1.15 for other failure mechanisms).

At the same time as the survey, the study also collected data for the hydraulic simulation. In this study, 96 river cross-sections for the Ca River system were collected. This is the most important data to present the channel's current condition. The floodplain topography data were created from a topography map 1/10,000 scale which was published by the Ministry of Natural Resources and Environment. Hydrometeorological data from all stations, which are in the national monitoring system, in the system (Figure 1) were collected and studied. These data were not only used to calculate the boundary for the hydraulic model but also used for model calibration and verification. All data collected in the study have clear origins, ensuring the reliability of the model establishment.

The study used a coupled 1D–2D hydrodynamic model Mike Flood for flood simulation in the research area. This is a tool that allows integrated flood modeling through dynamic coupling for river and flood plain. There are numerous Vietnamese studies that have used this tool (Tran et al. 2014; Tran & Do 2017). The research area encompasses both river flow and overland flow on the floodplain. The river flow is primarily focused in the direction of the river, while the overland flow on the floodplain is a 2D flow that is dependent on the floodplain topography. In order to accurately simulate the flow regime in the research area using a 2D model, a detailed model's mesh is required. To save computational time and reduce the workload, an integrated model can be used to simulate the flow regime in the research area. In this model, the Mike 11HD model was used to simulate flow in the river The principal of the model is the system of equations for conservation of mass (3) and momentum (4) for unsteady flow (DHI 2020):
(3)
(4)
where Q is the discharge, [m3/s]; A is the flow area, [m2]; h is the flow depth, [m]; g is the acceleration of gravity,[m/s2]; x is the distance in the flow direction, [m]; t is the time, [s]; C is the Chezy coefficient, [m1/2/s]; R is the hydraulic radius, [m]; the Mike 21FM model was used to simulate overland flow on the floodplain. Similar to Mike 11HD, the conservation of mass and momentum equations are expressed in the following form (DHI 2016):
(5)
(6)
(7)
where h is the total water depth, [m]; d is the still water depth, [m]; ζ is the surface elevation, [m]; q, p is the flux densities in x and y directions [m3/s/m]; is the Coriolis parameter, [s−1]; fV is the wind friction factor; V is the windspeed, [m/s]; = components of effective shear stress
In the Mike Flood model, the crest elevation of dikes was simulated properly in accordance with the survey data. The breaking process was represented by the gradual change in the elevation of the dike crest. The two models were linked together through lateral links. The model has three upstream boundaries, which used discharge data at the three following hydrological stations: Yen Thuong on the Ca River, Son Diem on the Ngan Pho tributary, and Hoa Duyet on the Ngan Sau tributary. The downstream boundary of the model used tidal water data at Cua Hoi station. The amount of contributed flow in the simulation area was calculated by regression methods in accordance with the river tributaries. The model used observer data at Nam Dan, Linh Cam, and Cho Trang stations to calibrate and validate the model. The reliability of the model will be calculated through the Nash–Sutcliffe efficiency index (Nash & Sutcliffe 1970). The formula for calculating the Nash index – Sutcliffe efficiency is shown in formula (8):
(8)
where is the mean of observed discharges, and is calculated discharge. is observed discharge at time i.
To calibrate and validate the combined 1&2D hydraulic model, the study simulated the two major floods occurring in the past. The flood from August 30, 2019, to September 12, 2019, was used to calibrate the model to determine the roughness parameter set for the river system. This set of parameters was kept unchanged to conduct a simulation for the flood occurring from October 29, 2020, to November 6, 2020, to verify the model. Calculation results are shown in Figures 3 and 4. The Nash indexes in calibrated and validated processes at Nam Dan hydrological station were 0.72 and 0.79, respectively. While at Linh Cam hydrological station, where close to La Giang dike, the result was very good with Nash indexes were 0.89 and 0.74 in processes. The lowest result appeared in Cho Trang with Nash indexes up to 0.64. According to Moriasi et al. (2007)and Mccuen et al. (2006), the Nash values at the calculation stations during both calibration and verification were in the ‘good’ to ‘very good’ categories. This was proved by the agreement between the calculated curves and the measured values. The difference in flood peak water level at Nam Dan, Linh Cam, and Cho Trang stations was less than 0.12 m. From this, it can be concluded that the model is reliable enough to conduct calculation simulations in future scenarios.
Figure 3

The calculated and observed water levels in the flood event 2019: (a) Nam Dan; (b) Linh Cam; and (c) Cho Trang.

Figure 3

The calculated and observed water levels in the flood event 2019: (a) Nam Dan; (b) Linh Cam; and (c) Cho Trang.

Close modal
Figure 4

The calculated and observed water levels in the flood event 2020: (a) Nam Dan; (b) Linh Cam; and (c) Cho Trang.

Figure 4

The calculated and observed water levels in the flood event 2020: (a) Nam Dan; (b) Linh Cam; and (c) Cho Trang.

Close modal

Based on the calibrated and verified model, the study simulated flood scenarios. In fact, the existing dike can withstand the dike breach with the design frequency P = 0.6%. In order to test this hypothesis, the flood scenario of frequency P = 0.6% was simulated. This is considered as the based scenario to compare with other dike breach scenarios. Then, dike breach scenarios were simulated for each identified location. In this study, the authors also evaluated the impact of the breach development time on the maximum discharge through the breach and the inundation area caused by the dike breach. The assessment aimed to minimize the uncertainty of unwarranted factors such as the breach development time.

Based on the results of hydraulic calculations, inundation maps corresponding to the scenarios were built corresponding to each scenario. Using spatial analysis techniques in GIS, the inundation area as well as the number of affected households in each scenario were also determined. The study had detailed information for each flood level and each administrative unit. This will be important information for authorities in implementing rescue operations as well as in building evacuation plans in emergency conditions. Also, this is a premise for building a large database for further research such as artificial intelligence applications.

Dike breach scenarios

The field survey along La Giang dike shows that there are three locations with a high risk of dike breach. The first location is Tung Anh, located at K0 + 600 – K2 + 100. The embankment route is located close to the dike foot. When the water level is high, the flow hits the embankment directly, causing erosion at the embankment foot, and consequently leading to landslides at the embankment and dike in the riverside. As a result, this area is at high risk of erosion at the dike foot and dike breach. The most likely form of dike breach in this area is sliding. The next high-risk location is the Duc Dien effervescent area, covering from K12 + 200 to K14 + 100. This is a section of dike with a thick layer of coarse sand (thickness from 6 to 8.5 m), with a maximum depth of 12 m, and only a thin cover on the river and the field sides. In infiltration areas where the ground is unstable, when the flood water level rises above alarm II, water penetrates through the dike body, which can cause piping and break the dike. The last location exposed to a high risk of dike breach is the section of Nga Song from K16 + 213 to K19 + 149. In this area, when the flood encounters big winds, large waves crash into the dike body, leading to overtopping and consequently dike breach. The locations of the three vulnerable areas are shown in Figure 5. The representative pictures of high-risk locations are shown in Supplementary material, Appendix 1 to Appendix 3.
Figure 5

High-risk locations.

Figure 5

High-risk locations.

Close modal

Based on the identified locations, the study simulated a series of scenarios. The first scenario simulated a design flood of La Giang dike. This was a based scenario. Subsequent dike breach scenarios are compared with based scenarios to independently assess the consequences of a dike breach. For each dike breach location, the breach development time was changed from 1 to 24 h, respectively, to evaluate the sensitivity. 24 h is also the time to maintain a high flood level in the design flood.

In Table 1, the data on the dike size were determined from the actual measurement data in the area where the dike was located. The parameters m and t were determined based on the current condition of the dikes. For each condition of the specific dike, the parameter of breach form f was also determined via the field survey. With the assumption that the breach did not develop through the foundation of the dike, the study presents that if the calculated breach depth is greater than the dike height, the depth will be considered equal to the dike height. The calculation results of breach parameters are summarized in Table 1.

Table 1

The dike breach parameter

Locationh (m)w (m)mtfBreach length (m)Breach depth (m)Breach development time (h)
Tung Anh 7.0 35.5 0.42 0.81 111 7.0 1–24 
Duc Dien 6.6 37.5 0.42 1.21 149 6.6 1–24 
Nga Song 6.8 38.0 0.42 0.74 146 6.8 1–24 
Locationh (m)w (m)mtfBreach length (m)Breach depth (m)Breach development time (h)
Tung Anh 7.0 35.5 0.42 0.81 111 7.0 1–24 
Duc Dien 6.6 37.5 0.42 1.21 149 6.6 1–24 
Nga Song 6.8 38.0 0.42 0.74 146 6.8 1–24 

Sensitivity breach development time analysis

To validate the importance of breach development time, the study analyzed the sensitivity of the case of breach development time. The study conducted 15 calculation scenarios corresponding to the breach development time from 1 to 24 h at three locations, respectively. Figure 6(a) represents the maximum discharge through the breach, while Figure 6(b) represents the inundation areas caused by the dike breach in the area protected by La Giang dike. Each point represents a value corresponding to a simulation scenario. It was found that when the breach development time changed, the flow through the breach and inundation area also varied greatly in each location. However, the degree of change was dependent on the location. This shows the importance of determining the location of the breach. In general, when the breach time increases, the maximum discharge through the breach tends to decrease. However, the amount of reduction varies from location to location. The ratio value of the difference between the maximum flow discharges varied from 19% at the Duc Dien location to 55% at the Tung Anh location, with and as the maximum flow through the breach when the breach development time was 1 and 24 h, respectively. For the inundation area, the degree of variation depended on each location. At the Tung Anh location, there was a relative variation when comparing the inundation areas in different scenarios. However, the inundation areas were not much different at the Nga Song location (Figure 6(b)). Thus, the breach development time has a certain role in the flow through the breach and the inundation area. This conclusion is contrary to the opinion of Tadesse & Fröhle (2020) when the authors commented that the development time of the breach does not affect the inundation area.
Figure 6

Sensitivity breach development time analysis: (a) discharge and (b) inundation area.

Figure 6

Sensitivity breach development time analysis: (a) discharge and (b) inundation area.

Close modal

Inundation mapping and consequence assessment

Based on the model that was calibrated and validated, the study simulated the flood with frequency P = 0.6%. In the first scenario, all dike locations were assumed safe. This was the base scenario to compare with others. The study then conducted an inundation simulation in the based scenario and estimated the inundation area corresponding to the simulation results. Figure 7 shows the inundation map corresponding to the based scenario. The inundation maps corresponding to dike breach scenarios are shown in Supplementary material, Appendix 13 to Appendix 15. Table 2 shows the inundation area by the administrative units on several levels. The detailed statistics (up to commune level) of affected objects are shown in Supplementary material, Appendix 4 to Appendix 12. It was found that, as calculated in the design, the water in the river did not exceed the La Giang and Ta Lam dikes. Therefore, the areas protected by these dikes such as Vinh City of the Nghe An Province and Nghi Xuan District, and Hong Linh District of the Ha Tinh Province did not have large inundation areas. In districts such as Duc Tho, Nam Dan, and Hung Nguyen, although the inundation areas were many, most of them were floodplains located between the two dikes. Meanwhile, since this was a big flood, the water overflowed some low-grade dikes and caused widespread flooding in the Huong Son and Duc Tho districts.
Table 2

The inundation area in the base scenario

ProvinceDistrictInundation area (ha)
< 0.5m0.5–1 m1–1.5 m1.5–2 m2–2.5 m2.5–3 m> 3 mSum(ha)
Ha Tinh Duc Tho 323 343 397 543 801 867 3,520 6,793 
Hong Linh 182 216 
Huong Son 802 1,000 1,022 1,105 1,229 1,047 1,741 7,946 
Nghi Loc 31 23 35 36 22  152 
Nghi Xuan 96 108 159 97 101 137 1,208 1,906 
Vu Quang 278 288 287 245 241 191 375 1,905 
Nghe An Hung Nguyen 43 43 45 55 146 382 1,590 2,304 
Nam Dan 230 296 441 698 906 973 3,094 6,638 
Thanh Chuong 26 22 18 17 20 23 206 332 
Vinh 65 43 39 41 46 35 292 561 
Sum (ha) 1,901 2,172 2,446 2,841 3,518 3,667 12,209 28,754 
ProvinceDistrictInundation area (ha)
< 0.5m0.5–1 m1–1.5 m1.5–2 m2–2.5 m2.5–3 m> 3 mSum(ha)
Ha Tinh Duc Tho 323 343 397 543 801 867 3,520 6,793 
Hong Linh 182 216 
Huong Son 802 1,000 1,022 1,105 1,229 1,047 1,741 7,946 
Nghi Loc 31 23 35 36 22  152 
Nghi Xuan 96 108 159 97 101 137 1,208 1,906 
Vu Quang 278 288 287 245 241 191 375 1,905 
Nghe An Hung Nguyen 43 43 45 55 146 382 1,590 2,304 
Nam Dan 230 296 441 698 906 973 3,094 6,638 
Thanh Chuong 26 22 18 17 20 23 206 332 
Vinh 65 43 39 41 46 35 292 561 
Sum (ha) 1,901 2,172 2,446 2,841 3,518 3,667 12,209 28,754 
Figure 7

The inundation map for the base scenario.

Figure 7

The inundation map for the base scenario.

Close modal
The study selected the most unfavorable scenarios corresponding to the breach development time of 1 h at three locations for comparison. Compared with the based scenario, the inundation areas caused by the dike the breach increased significantly, which mainly took place in the area protected by La Giang dike. Contrary to the findings of (Vorogushyn et al. 2010; De Bruijn et al. 2014; Bomers et al. 2019) we did not find any reduction in the maximum water level and discharge at downstream of the breach locations. It is important to highlight the fact that the locations of the dike breach are very close to the sea. Tide dominates the hydraulic regime in this area. That leads to a significant reduction in the impact of the dike breach. Figure 8 shows the increased inundation area due to dike breaches at different locations. In all cases of dike breach, the inundation area extended to Hong Linh town, which would not be affected if the dike was safe. The study then estimated inundation areas as well as affected populations caused by dike breaches. The results of the increased inundated area due to the dike breach and the additional number of people affected are shown in Table 3. In the case of the dike breach at the Tung Anh location, although it has the least impact, the total inundation area in this case is 7,748 ha, of which about 34% of the inundation area is less than 0.5 m deep. With a depth of less than 0.5 m, the affected residents can completely handle on the spot without having to relocate to other areas. In this case, the inundation area over 2 m deep is almost negligible. In the case of dike breach at Huu Dien and Nga Song locations, the flood level increased by 1.5 times. Although the total inundation areas in these two cases were similar, there was a significant difference in the flood level. When the flood level was higher than 2 m (higher than adult height), the inundation area in the case of the dike breach in Nga Song was double in the Huu Dien case (3,982 ha compared with 2,087 ha). Especially, when the flood level was higher than 3 m (average height of 01 floors), the inundation area caused by the dike breach in Nga Song was 41 ha, compared to 11 ha in Huu Dien.
Table 3

The inundation area, number of affected households, and length of flooded road by dike breach

Breach locationDistrictInundation depth
< 0.5 m0.5–1 m1–1.5 m1.5–2 m2–2.5 m2.5–3 m> 3 mTotal
Inundation area (ha) 
Tung Anh Can Loc 302 361 140 807 
 Duc Tho 1,716 2,434 777 83 5,019 
Hong Linh 627 863 378 47 1,921 
Huu Dien Can Loc 561 794 643 718 340 20 3,077 
Duc Tho 446 1,021 1,703 2,219 987 47 6,427 
Hong Linh 171 275 566 933 520 161 2,634 
Nga Song Can Loc 503 718 746 693 548 120 3,330 
 Duc Tho 294 758 1,298 2,072 1,822 313 6,561 
Hong Linh 145 193 412 759 794 344 34 2,681 
Number of affected households 
Tung Anh Can Loc 554 374 15 943 
Duc Tho 4,790 3,124 636 94 11 8,656 
Hong Linh 991 1,086 168 19 2,268 
Huu Dien Can Loc 1,137 1,019 945 962 251 4,314 
Duc Tho 1,772 3,464 4,872 3,256 648 14,015 
Hong Linh 460 746 1,237 1,184 428 35 4,091 
Nga Song Can Loc 925 1,175 1,152 1,018 434 4,711 
Duc Tho 985 2,917 4,175 4,148 2,225 32 14,484 
Hong Linh 526 300 1,245 1,130 981 79 4,265 
Flooded road (km) 
Tung Anh Can Loc 25.15 28.34 9.96 0.05 63.51 
Duc Tho 113.14 133.03 32.96 7.98 4.57 1.60 0.15 293.43 
Hong Linh 58.87 75.23 29.11 2.94 1.65 0.55 0.17 168.52 
Huu Dien Can Loc 31.36 37.00 37.50 40.95 19.10 1.17 0.00 167.08 
Duc Tho 30.15 65.72 114.95 126.61 49.73 5.69 1.76 394.62 
Hong Linh 17.55 25.71 56.85 82.27 41.57 12.07 1.16 237.18 
Nga Song Can Loc 29.18 37.09 38.31 40.71 27.85 7.30 0.02 180.45 
 Duc Tho 17.66 53.45 86.83 124.17 101.17 16.15 2.10 401.53 
Hong Linh 18.29 16.75 42.50 68.79 67.02 27.41 3.09 243.85 
Breach locationDistrictInundation depth
< 0.5 m0.5–1 m1–1.5 m1.5–2 m2–2.5 m2.5–3 m> 3 mTotal
Inundation area (ha) 
Tung Anh Can Loc 302 361 140 807 
 Duc Tho 1,716 2,434 777 83 5,019 
Hong Linh 627 863 378 47 1,921 
Huu Dien Can Loc 561 794 643 718 340 20 3,077 
Duc Tho 446 1,021 1,703 2,219 987 47 6,427 
Hong Linh 171 275 566 933 520 161 2,634 
Nga Song Can Loc 503 718 746 693 548 120 3,330 
 Duc Tho 294 758 1,298 2,072 1,822 313 6,561 
Hong Linh 145 193 412 759 794 344 34 2,681 
Number of affected households 
Tung Anh Can Loc 554 374 15 943 
Duc Tho 4,790 3,124 636 94 11 8,656 
Hong Linh 991 1,086 168 19 2,268 
Huu Dien Can Loc 1,137 1,019 945 962 251 4,314 
Duc Tho 1,772 3,464 4,872 3,256 648 14,015 
Hong Linh 460 746 1,237 1,184 428 35 4,091 
Nga Song Can Loc 925 1,175 1,152 1,018 434 4,711 
Duc Tho 985 2,917 4,175 4,148 2,225 32 14,484 
Hong Linh 526 300 1,245 1,130 981 79 4,265 
Flooded road (km) 
Tung Anh Can Loc 25.15 28.34 9.96 0.05 63.51 
Duc Tho 113.14 133.03 32.96 7.98 4.57 1.60 0.15 293.43 
Hong Linh 58.87 75.23 29.11 2.94 1.65 0.55 0.17 168.52 
Huu Dien Can Loc 31.36 37.00 37.50 40.95 19.10 1.17 0.00 167.08 
Duc Tho 30.15 65.72 114.95 126.61 49.73 5.69 1.76 394.62 
Hong Linh 17.55 25.71 56.85 82.27 41.57 12.07 1.16 237.18 
Nga Song Can Loc 29.18 37.09 38.31 40.71 27.85 7.30 0.02 180.45 
 Duc Tho 17.66 53.45 86.83 124.17 101.17 16.15 2.10 401.53 
Hong Linh 18.29 16.75 42.50 68.79 67.02 27.41 3.09 243.85 
Figure 8

The inundation area by dike breach at different locations.

Figure 8

The inundation area by dike breach at different locations.

Close modal

Most of the people affected in the case of dike breach in Tung Anh were at low flood levels. The number of households which got flooded below 0.5 m deep was nearly 6,335, accounting for 53.4% of the affected households. This number increased to 92,000 households if the figure getting flooded under 1 m is counted. The number of people who had to relocate to high areas is fewer than 1,000 households. Therefore, in this case, the plan will focus more on supporting people to respond on the spot. In the case of the Huu Dien and Nga Song dike breach, the total number of affected households increased dramatically, with 22.4 thousand households and 23.5 thousand households, respectively. Furthermore, the number of households which got flooded over 1 m has also increased significantly. The number of households was 13.8 thousand households when the dike broke at Huu Dien, and 16.6 thousand households when the dike broke at Nga Song. This shows that the impacted scope as well as the impact on people when the dike breaks depends greatly on the location of the dike breach. In these cases, the response plan needs to pay more attention to evacuation. As the number of people to evacuate increases, the scope of response plans also increases greatly. Pressure on evacuation locations, food supplies, electricity and water also increases. Therefore, detailed information from the study will be an important input to develop an effective response plan.

Besides assessing the inundation area and the number of affected households, the study also assessed the degree of flooding on the roads. This information is not only the basis for assessing the extent of damage but also, more importantly, provides visual information about which sections of the road can be deeply flooded, limiting mobility in the case of a flood. This is also extremely useful information for decision-makers as well as local people.

Although the approach has shown clear advantages in terms of reducing the number of computational scenarios and providing detailed information to build emergency response scenarios, a number of limitations should be mentioned. First, the breach parameters are identified based on empirical formulas which can have uncertain variables. In the future, sensitivity analyses of these variables should be performed to assess their impacts on the flood consequence assessment. In addition, although the survey has found the most critical locations, the risk of dike breach can occur at any location on the dike for many reasons. Therefore, raising awareness of the local community as well as proactively responding to emergency situations should be implemented to minimize the damages. In this study, the authors only studied houses and roads. Other subjects such as crop area, and facilities have not been mentioned in the study. This is also the limitation of the study. With the same approach, such factors can completely be considered in the next studies.

The paper investigated the possible impact of dike breach scenarios for the La Giang dike. Based on the survey results, Tung Anh, Duc Dien, and Nga Song were determined as three high-risk dike breach locations. This has reduced the number of scenarios that need to be considered. The Mike Flood model was used to simulate the flow dynamic regime on the Ca River. The model was calibrated and validated with observer data at three hydrological stations. The results were very good with the Nash index up to 0.89. In this study, empirical formulas were used to calculate the breach parameters at three selected locations, including the length and depth of the breach in each location. The importance of breach development time, maximum discharge through the breach, and inundation area was analyzed via sensitivity analysis. The results show that breach time plays an important role, in influencing the above-mentioned factors. Through the GIS, the inundation area, the number of affected people, and the length of flooded roads are also statistically studied. All of the above elements are significantly different between the calculated cases. It means that we need a different response plan for each case. The results of the study also provide detailed statistics by administrative units and visual maps. This is useful information for the authorities and local people to help them have specific plans to face natural disasters. It is also very important information for the development of response plans as well as the establishment of a database for disaster prevention.

This research is supported by Thuyloi University.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.

Apel
H.
,
Merz
B.
&
Thieken
A. H.
2009
Influence of dike breaches on flood frequency estimation
.
Computers and Geosciences
35
(
5
),
907
923
.
https://doi.org/10.1016/j.cageo.2007.11.003.
ASCE/EWRI Task Committee on Dam/Levee Breaching
2011
Earthen embankment breaching
.
Journal of Hydraulic Engineering
137
(
12
),
1549
1564
.
https://doi.org/10.1061/(asce)hy.1943-7900.0000498
.
Bomers
A.
,
Schielen
R. M. J.
&
Hulscher
S. J. M. H.
2019
Consequences of dike breaches and dike overflow in a bifurcating river system
.
Natural Hazards
97
(
1
),
309
334
.
https://doi.org/10.1007/s11069-019-03643-y
.
Danka
J.
&
Zhang
L.
2015a
Evaluation of empirical methods for estimating breaching parameters of dikes
. In:
Geotechnical Safety and Risk V
, pp.
841
846
.
https://doi.org/10.3233/978-1-61499-580-7-841
.
Danka
J.
&
Zhang
L. M.
2015b
Dike failure mechanisms and breaching parameters
.
Journal of Geotechnical and Geoenvironmental Engineering
141
(
9
).
https://doi.org/10.1061/(asce)gt.1943-5606.0001335
.
De Bruijn
K. M.
,
Diermanse
F. L. M.
&
Beckers
J. V. L.
2014
An advanced method for flood risk analysis in river deltas, applied to societal flood fatality risk in The Netherlands
.
Natural Hazards and Earth System Sciences
14
(
10
),
2767
2781
.
https://doi.org/10.5194/nhess-14-2767-2014
.
DHI
2016
MIKE 21 Flow Model User Manual
.
DHI
2020
MIKE 11 User Manual
.
Ha Tinh Department of Agriculture and Rural Development
2023
Assess the Current Status of Dyke Works, Identify key Locations and Develop Dike Protection Plans in 2023
.
Huthoff
F.
,
Remo
J. W. F.
&
Pinter
N.
2015
Improving flood preparedness using hydrodynamic levee-breach and inundation modelling: Middle Mississippi River, USA
.
Journal of Flood Risk Management
8
(
1
),
2
18
.
https://doi.org/10.1111/jfr3.12066
.
Liu
X.
,
Chen
D.
,
Qu
G.
&
Zhang
H.
2019
Preliminary study on the erosion impact of breach floods based on numerical simulation and physical model test
.
IOP Conference Series: Earth and Environmental Science
376
(
1
).
https://doi.org/10.1088/1755-1315/376/1/012061
.
Mccuen
R. H.
,
Knight
Z.
&
Cutter
A. G.
2006
Evaluation of the Nash–Sutcliffe efficiency index
.
Journal of Hydrologic Engineering
11
(
6
).
https://doi.org/10.1061/ASCE1084-0699200611:6597
.
Ministry of Agriculture and Rural Development of Vietnam
2023
Dispatch No. 3276/BNN-ĐĐ on Implement the Plan to Protect key Points on Main Dykes in 2023
.
Moriasi
D. N.
,
Arnold
J. G.
,
Van Liew
M. W.
,
Bingner
R. L.
,
Harmel
R. D.
&
Veith
T. L.
2007
Model evaluation guidelines for systematic quantification of accuracy in watershed simulations
.
Transactions of the ASABE
50
(
3
),
885
900
.
Nagy
L.
2006
Estimating dike breach length from historical data
.
Periodica Polytechnica Civil Engineering
50
(
2
),
125
139
.
Nash
J. E.
&
Sutcliffe
J. V.
1970
Ireland's water budget – model validation and a greenhouse experiment
.
Journal of Hydrology
10
,
282
290
.
https://doi.org/10.1080/00750770109555783
.
Nguyen
T. P. T.
,
Tran
T. L.
,
Nguyen
T. T. H.
,
Pham
Q. V.
&
Nguyen
T. L.
2020
Mapping flood inundation areas over the lower part of the con river basin using sentinel 1A imagery
.
Vietnam Journal of Earth Sciences
42
(
3
),
288
297
.
https://doi.org/10.15625/0866-7187/42/3/15453
.
Schmocker
L.
&
Hager
W. H.
2009
Modelling dike breaching due to overtopping
.
Journal of Hydraulic Research
47
(
5
),
585
597
.
https://doi.org/10.3826/jhr.2009.3586
.
Tadesse
Y. B.
&
Fröhle
P.
2020
Modelling of flood inundation due to levee breaches: Sensitivity of flood inundation against breach process parameters
.
Water (Switzerland)
12
(
12
).
https://doi.org/10.3390/w12123566
.
Tadesse
B.
,
Birhanu Tadesse
Y.
&
Fröhle
P.
2017
Modelling of river dike breach type physical test flow with Telemac-2D. 24th Telemac-Mascaret User Conference, 17, 107–112. Available from: https://hdl.handle.net/20.500.11970/104491
.
The Central Steering Committee for Disaster Prevention and Control
2018
Vietnam Natural Disaster 2018
.
The Central Steering Committee for Disaster Prevention and Control
2019
Vietnam Natural Disaster 2019
.
The Central Steering Committee for Disaster Prevention and Control
2020
Vietnam Natural Disaster 2020
.
The Central Steering Committee for Disaster Prevention and Control
2021
Vietnam Natural Disaster 2021
.
Tran
C. K.
&
Do
K. X.
2017
Study on water balance in Sesan river basin in drought year 2015/2016
.
Journal of Meteorological
678
,
44
53
.
Tran
N. A.
,
Dang
D. K.
,
Nguyen
T. S.
&
Dang
D. D.
2014
Hydraulic modelling for flood vulnerability assessment, case study in river basins in North Central Vietnam
. In
Conference on Integrated Water Resource Management, Management Policy and Decision Making Supports
. .
Viero
D. P.
,
D'Alpaos
A.
,
Carniello
L.
&
Defina
A.
2013
Mathematical modeling of flooding due to river bank failure
.
Advances in Water Resources
59
,
82
94
.
https://doi.org/10.1016/j.advwatres.2013.05.011
.
Vorogushyn
S.
,
Merz
B.
,
Lindenschmidt
K. E.
&
Apel
H.
2010
A new methodology for flood hazard assessment considering dike breaches
.
Water Resources Research
46
(
8
).
https://doi.org/10.1029/2009WR008475
.
Vorogushyn
S.
,
Apel
H.
&
Merz
B.
2011
The impact of the uncertainty of dike breach development time on flood hazard
.
Physics and Chemistry of the Earth
36
(
7–8
),
319
323
.
https://doi.org/10.1016/j.pce.2011.01.005
.
Zolghadr
M.
,
Hashemi
M. R.
&
Zomorodian
S. M. A.
2011
Assessment of mike21 model in dam and dike-break simulation
.
Transactions of Civil Engineering
35
(
C2
),
247
262
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).

Supplementary data