Flood modeling in the Ba River basin using a coupled hydrodynamic model—MIKE FLOOD

The problem of flooding in Central Vietnam in general and the lower Ba River in particular is one of the natural disasters that frequently threatens people's lives and socioeconomic development in the region. Especially, climate change is becoming ever more prominent and hotter, making extreme natural disasters more unusual and unpredictable. In this research, MIKE‐FLOOD—a model that connects a 1‐dimensional (1‐D) MIKE 11 Hydrodynamics (HD) model with a 2‐dimensional (2‐D) MIKE 21 HD model—was used to set up. The model was calculated for three floods: (1) flood in October 1993, (2) flood in November 2003, and (3) flood in November 2007; these are floods with high frequency and relatively large magnitude. The results show that the 1993 flood rose and receded quickly. The flood peak inundated an area of 22,600 ha, accounting for 52% of the natural area. The flooded areas deeper than 1, 2, 3, 4, and 5 m were 16500, 11,000, 7000, 4200, and 2200 ha, respectively. In the center of Tuy Hoa city, the flooded area at the time of maximum water level was almost 100%.


| INTRODUCTION
Flooding in Central Vietnam especially in the lower Ba River is one of the natural disasters that frequently threaten people's lives and socioeconomic development in the region.Every year from mid-September to mid-January, the downstream region of the Ba River and its vicinity is seriously threatened by floods and storms, causing great damage to the properties and lives of the people in the area.The Ba River basin is a fairly large plain with nearly 25,000 ha of arable land.The elevation difference between the river bed and the field is not large (about 1.5 m).In particular, the downstream area is directly adjacent to the East Sea, so whenever there is a flood, this area is completely submerged in water, according to survey data in recent years, the flood situation downstream of Ba River, Ban Thach River and Ky Lo river become more serious.
It can be said that flooding is a big problem in the Central region and tends to be aggravated by the effects of climate change-From Figure 1, it can be seen that storms hitting Vietnam have increasingly high wind speeds, which is correlated with heavier rainfall, making the risk of floods even more serious.In the past, when modeling tools and computers were not developed, it was quite difficult to simulate flood events.The hydraulic calculation of the Ba river basin has been done with the VRSAP software [Dac, N.T. (2005)].However, Vietnam River System And Plains (VRSAP) is very difficult to use because of its complications.Moreover, it is not compatible with high-speed computers.In recent years with the strong development of computer technology as well as modeling tools, the application of simulation models of flood flows in rivers is quite popular, and this has also assisted researchers, management agencies, and planning.Van (2010) used a MIKE FLOOD model to assess the level of inundation in the lower Ba River.He established a set of parameters that give relatively accurate results in terms of quantity, flood type, and the timing of its occurrence.The model was calibrated using the historical flood of 1993 (corresponding to a frequency of 5%) with relatively accurate results, from which the model can be used to simulate future flood situations.Hoa (2012) used the Environmental Fluid Dynamics Code (EFDC) model to simulate the extent of inundation for the Ba River basin.The results obtained were quite good for calculating and simulating inundation area, inundation depth, and velocity field at locations in the lower Ba river basin.
Tuy Hoa City-Phu Yen Province, located in the downstream region of the Ba River-is one of the cities that is gradually developing service industries to become a famous beach tourism destination as well as a venue for major national and international events.Therefore, forecasting and assessing natural disasters such as climate, inundation, and saltwater intrusion in the downstream area of the Ba River are important to building Tuy Hoa into a smart and modern city.Therefore, it is urgent to assess the flood level in the Ba River downstream area and propose flood management measures.
To solve that problem, the Study will simulate flood events in the Ba River downstream using a 1-2-dimensional MikeFlood hydraulic model.Thereby assessing the extent of flooding in the downstream area of Ba River.

| RESEARCH DATA AND METHODS
As this paper aims to simulate field overflow and assess the level of flooding in an area, the requirement is to calculate and simulate the amount and area of inundation in the entire area.Therefore, it is necessary to use a model capable of connecting features of both 1-dimensional (1D) The wind speed of storms gradually increases over the years.
The study area: (a) Ba River Basin and (b) Ba River downstream.and 2-dimensional (2D) models to simulate both river flows and flood plains while the simulation time is not large (Figure 2).
While VRSAP model is a hydrological-hydraulic mathematical model of 1D flow and other models that cannot connect modules (connect modules hydrological, 1D hydraulic, and 2D hydraulic).Found that the MIKE-FLOOD model of the Danish Hydraulic Institute (DHI) can meet the requirements of assessing the level of flooding for a system, that has the feature of connecting 1D hydraulic model MIKE 11 HD and 2D hydraulic model MIKE 21 HD.Therefore, the authors used the MIKE-FLOOD model to determine the degree of flooding for the downstream Ba River, Phu Yen province.

| Simulation model basis
MIKE FLOOD is a tool for coupling 1D and 2D models, aiming at studying the combination of 1D and 2D flow.In the study, MIKE FLOOD is used to combine the MIKE 11 HD river network model and the MIKE 21 HD rectangular grid 2D model.The 1D model is used to simulate the river system while the 2D model is used to simulate flows on reservoirs and lagoons.
MIKE FLOOD can simulate detailed physical conditions of the system.The 2D model will simulate the flow in shallow water areas such as reservoirs, wetlands, and lagoon systems.MIKE FLOOD is used where both 2D description in some areas (MIKE 21) and 1D modeling (MIKE 11) are required.For flexible connectivity, the model allows for many types of connections.In this study, the authors used a bilateral connection type; this connection allows each sequence of MIKE 21 cells to be joined on either side, creating a channel segment in MIKE 11 or a segment of a tributary or the entire tributary.A construction equation or a flow relation with the water level (Q~H) calculates the flow through the connection between two sides.This type of connection is particularly suitable for simulating overflow from a river into an inland field area where the flow across the dyke is calculated using a spillway equation (Figure 3).

| Database and modeling
The model set used the 1993 flood for calibration and the 2003 and 2007 floods for validation.The 1993 is a historical event due to the influence of historic storms in late September and early October, causing torrential rainfall in the entire Ba River downstream.The rainfall was 628.9 mm on October 3 at Tuy Hoa, and 579.0, 502.0, and 632.9 mm on October 4 at Son Hoa, Son Thanh, and Song Hinh, respectively.The water level at Cung Son was 39.9 m which was 6.4 m higher than the 3rd alarm water level.In addition, the water level at Phu Lam was 5.21 m which is 2.01 m higher than the 3rd alarm water level.The 1993 flood was selected for model calibration because it was the largest historical flood ever in the Ba River basin; The flood in 2003: Due to the influence of the cold continental high margin in the North combined with the tropical convergence band in the South, causing heavy rain; especially, local heavy rain from 7:00 p.m., November 12, 2003 to 7:00 p.m., November 13, 2003 in the Hinh river was 715.4 mm.In this flood, the water level in Cung Son was 36.44,2.94 m higher than the alarm water level III, and the water level in Phu Lam was 3.62, 0.42 m higher than the alarm water level III; The flood in 2007: The influence of tropical depression and cold air caused heavy to very heavy rain, the amount of rain measured from 7:00 p.m., November 2, 2007 to 7:00 p.m., November 3, 2007 measured rainfall in Tuy Hoa was 301, 1 mm, the largest daily rainfall in Son Thanh is 376.1 mm, causing major floods in rivers, in Cung Son the water level is 35.32,1.82 m above alarm water level III, in Phu Lam it is 4, 29 m is 1.09 m larger than an alarm water level III (Figure 4).

| Mike11 model
The simulation scope is the main tributary of the Ba River downstream from Cung Son to Da Rang estuary.The entire main river branch from Cung Son to Da Rang estuary has a total length of 50,688 km.The upper boundary is the flow process line or the actual water level process line measured at Cung Son.The lower boundary is the tidal boundary calculated from the Quy Nhon tidal measuring station.
The cross-section document for the model includes 24 river cross-sections from Cung Son station to Phu Lam Bridge by the Institute of Planning, measured and adjusted in 1997, and three cross-sections from Phu Lam Bridge to the sea mouth by Planning Irrigation measured in 2003.

| Mike21 model
The MIKE 21 2-D model extent is a two-dimensional domain measuring 14.4 × 10.0 km.The vertical from upstream (the cross section is 39,500 m from the source-Cung Son) to the downstream is Da Rang estuary (Figure 3).
A rather important part of the 2-D model setting is the construction of the Bathymetry digital elevation map.From the DEM map with a 40 × 40 m resolution, interpolate to build a Bathymetry digital elevation map for the study area.Bathymetry is set up with a grid step of 40 ×40 m, on an area of size 14.4 × 10.0 km with a grid of 400 × 250.The calculated domain is rotated at an angle of 650 to the meridian.The elevation of the soil point (the point where there is no overflow during the entire simulation period) is chosen to be 20.0 m.Local areas such as estuaries, dikes, and roads are locally modified to suit the actual topography.In particular, the surrounding grid points (except for the estuary area) are assigned as land points.

| Mike Flood model
After building the hydraulic network in MIKE 11 and MIKE 21, running through and preliminarily correcting both of the above hydraulic networks, we proceed to connect (couping) these two networks in MIKE FLOOD, and the connection is shown in Figure 5.The first step of model correction is to check and handle the unsuitable points when we combine the river network in MIKE 11 with the base of MIKE 21.This correction step makes the main river line in the model.MIKE 11 must coincide with the river line on the MIKE 21 platform (Figures 7 and 8).
The flood simulation of October 1993 has achieved quite good results at hydrological measurement stations.At Cung Son and Phu Lam stations, the water level process between calculated and measured is quite consistent in shape, and the criteria for evaluating the error between measured and calculated are at a good level or higher (Table 1).Therefore, it can be concluded that the set of hydraulic parameters we have chosen is reliable enough to calculate future flood plans.
F I G U R E 5 Methodology flowchart.

| Calculation results of the 2003 flood
This is a medium-sized flood in the lower third of the river.Floods formed by the influence of the cold continental high margin in the North combined with the tropical convergence band in the South caused heavy rain, especially local heavy rain in the Hinh River from 7:00 p.m. November 12, 1993 to 7:00 p.m. November 13, 1993, 715.4 mm.In the flood on November 12-14, the water level at Cung Son was 36.44,2.94 m higher than alarm III, and the water level at Phu Lam was 3.62, 0.42 m higher than alarm level III.
Specifically, in Cung Son, the flood peak appeared at 4:00 p.m. on November 13, 2003, and at 8:00 p.m. the same day, the flood began to subside.At Phu Lam, the flood peaked at 8:00 p.m. on November 13, 2003, and the flood peak was maintained for about 5 h.The flood transmission time from Cung Son to Phu Lam takes about 5 h for a distance of more than 46.6 km-the flow speed is much lower than the 1993 flood.The parameter set of the model after the correction step is changed to the upstream water level boundary, the water level boundary is also replaced by the tidal boundary, and the joining boundary is also replaced by the corresponding time rainfall.
The test results of the November 2003 flood, show that the criteria for assessing the error between actually measured and calculated at Phu Lam Station are good (Table 2).The difference in flood peak water level at Phu Lam station between calculation and actual measurement is not significant, at Phu Lam the difference in flood peak is 0.2 m.In addition to the error of flood peak water level, the flood shape between measured and calculated also has a suitable correlation.

| Calculation results of the XI/2007 flood
Due to the influence of tropical depressions and cold air, the rainfall was measured from 19:00 p.m. November 02, 2007 to 1:00 p.m. November 03, 2007, the amount of rain measured in Tuy Hoa was 301.1 mm, the amount of rain in Tuy Hoa was 301.1 mm.The heaviest rain a day in Son Thanh is 376.1 mm, causing major floods in rivers, in Cung Son the water level is 35.32,1.82 m higher than alarm III, and in Phu Lam, the water level is 4.29, 1.09 m higher than alarm III.
Flood transmission time and water level gradient are at typical levels for the area.The flood peaked at Cung Son at 2:00 p.m. on November 4 and remained for about 5:00 a.m.before it began to subside.At Phu Lam, the flood peaked 5 h later than at Cung Son and the flood peak also remained for 5 h.Through the results of calculating the flood-affected area, it is seen that, although the 2003 Flood and the 2007 Flood are two floods that frequently occur in the basin, corresponding to a frequency of 20% of the inundation area when H max appears at points in the study area is not much different from the particularly large flood in 1993 corresponding to a frequency of 5%, only different in the value of each point.
During this flood, the water level at Phu Lam was monitored and used to evaluate the test results.The time step of the test simulation is also selected as 3 s.Then the model also calculated and warned the number Cr < 1 to satisfy the model's conditions of stability and convergence.With a simulation time step of 3 s to simulate and test the flood from 1 a.m. on November 2, 2007, to 1 p.m. on November 9, 2007, the model must perform 216,000 calculation steps.The parameter set of the model after the correction step continues to be changed to the discharge margin upstream, the water level boundary is also replaced by the tidal boundary, and the joining boundary is also replaced by the corresponding time rainfall (Figures 11 and 12).The results of flood testing in November 2007 show that the test results are quite good, the Nash index is 84% and the correlation coefficient R2 is also above 0.90.At the same time, the difference in flood peak water level at Phu Lam station between calculated and measured is not significant, the difference is only 0.07 m (Actually measured is 4.29 m and calculated is 4.36 m).

| CONCLUSION
The study used the combined 1D/2D model-Mike Flood model to establish and simulate flooding for the lower Ba River.With the results of adjustment and verification as shown in Tables 1-3, we see that the set of parameters has been simulated for three floods October 1993, November 2003, and November 2007 corresponding to the frequency of large and relatively large floods then the simulation results are completely reliable.However, to get the most suitable set of parameters for all types of floods and all floods in both dry seasons and high flood seasons, we need to calibrate and simulate many different types of floods.
Through the results of calculation and analysis of the flooded area compared to the inundation time of the With the set of parameters set in this report, it gives us relatively accurate results in terms of quantity, type of flood, and time of occurrence.Therefore, it can be used to simulate and predict floods for the Ba River downstream.

F
I G U R E 3 Connection diagram of two sides.[DHI,2006] F I G U R E 4 Scope diagram of the study (MikeFlood connects the Mike11 and Mike21 models together).

3
| RESULTS AND DISCUSSION 3.1 | Calculation results of the 1993 flood The simulated flood is extracted from the October 1993 flood.In this flood, the upper boundary is the actual discharge measured at Cung Son, while the lower boundary is the tidal boundary and the amount joining the middle zone from MIKE NAM.Water level data at Phu Lam and flood trace survey sites are also monitored and used for correction during simulation.The grid step in the bathymetry digital elevation map has been set with the distance Δx = 40 m and Δy = 40 m, so the selected time step for simulation is 3 s.Then the model calculated the number Cr = 0.83 < 1 to satisfy the model's stability and convergence conditions.With a time step of 3 s, the model must perform 115,200 calculation steps (Figure 6).

F
I G U R E 6 Connect the network of Mike 11 rivers with the Bathymetry of Mike 21.F I G U R E 7 Chart of actual water level measured and calculated for the October 1993 flood at Phu Lam.TUAN ET AL. | 203 The time step of the test simulation is also selected as 3 s.With a simulation time step of 3 s to simulate and test the flood from 01:00 a.m. on November 12, 2003 to 7:00 p.m. on November 20, 2003, the model must perform 252,000 calculation steps (Figures 9 and 10).

F
I G U R E 8 H max in the flood October 1993.(a) Results cut from the model, (b) results are processed on GIS software.T A B L E 1 Indicator to evaluate the error between actual measurement and calculation at Phu Lam station.The floodError of flood peak (

F
I G U R E 9 Chart of actual water level measured and calculated for the 11/2003 flood at Phu Lam.F I G U R E 10 H max in the flood 11/2003.(a) Results cut from the model, (b) results are processed on GIS software.T A B L E 2 Indicator to evaluate the error between actual measurement and calculation at Phu Lam station.1993 flood, it shows that: when the flood water level reaches the maximum value, the total area affected by flooding corresponds to the water level.The flood area is 22,600 ha, accounting for 52% of the natural area, up to 16,500 ha are inundated with a depth of more than 1 m, 11,000 ha are inundated with a depth of more than 2 m, 7000 ha are inundated with a depth of more than 3 m, 4200 ha are inundated with a depth of more than 4 m and the flooded area more than 5 m deep is 2200 ha.Although the 2003 flood and the 2007 flood are two frequent floods occurring in the basin, with a frequency of 20%, the flooded area when H max occurs at points in the study area is not.much different from the exceptionally large flood of 1993 with a frequency of 5%, differing only in the value of F I G U R E 11 Chart of actual water level measured and calculated for the 11/2007 flood at Phu Lam.F I G U R E 12 H max in the flood 11/2007.(a) Results cut from the model, (b) results are processed on GIS software.T A B L E 3 Indicator to evaluate the error between actual measurement and calculation at Phu Lam station.Especially in the center of Tuy Hoa city, the flooded area at the time of H max is almost 100%.Compared with other hydro-hydraulic mathematical models, the DHI Mike model set is quite complete and diverse because many modules can be combined, in the article, Module MikeFlood was used to connect model 1-D Mike11 and 2-D model Mike 21.Saves computational resources.The flexible connections in the MIKE FLOOD Module have made the model quite complete for a basin including hills, plains, and coastal estuaries.The model is capable of simulating even complex systems.With the research method set out is to use the modules in the DHI MIKE model set including MIKE 11, MIKE 21, and MIKE FLOOD.As a result, flood flows in the Ba River Basin have been calculated.