flood inundation, damage and risk assessment - DHI
FLOOD HAZARD AND RISK ASSESSMENT OF HOANG LONG RIVER BASIN, VIETNAM VU Thanh Tu1, Tawatchai TINGSANCHALI2 1
Water Resources University, Assistant Professor, 175 Tay Son Street, Dong Da District, Hanoi, Vietnam 100000, e-mail: [email protected]
2
Nakhon Pathom Rajabhat University, Program of Civil and Environmental Engineering, Professor, Nakhon Pathom, Thailand 73000, e-mail:[email protected] Keywords Flooding depth, inundation map, flood damage assessment, flood hazard, risk Abstract Hoang Long is the largest river in Ninh Binh province, Vietnam. This province is frequently affected by flooding due to heavy rainfalls in the river basin. Rainfall frequency analysis was done to obtain a maximum 7-day rainfall for various return periods, and to construct design hyetographs corresponding to 5, 10, 20, 50, 100 and 200-year return periods. Design flood hydrographs corresponding to different return periods are obtained by using rainfall-runoff model (MIKE-NAM). These results are input to MIKE 11 hydrodynamic model to simulate flood flow in Hoang Long basin. The NAM and MIKE 11 models were calibrated and verified using past flood data. Using the maximum water levels computed by MIKE 11 model at different locations along the rivers and floodplains and digital elevation map (DEM), flood inundation maps were developed for Ninh Binh province. Flooding depth and inundation area corresponding to various return periods were determined. From the inundation maps and field survey data, damage curves for residential, agricultural and roads damages were developed. Flood hazard and flood risk maps were constructed typically for the flood of 20-year return period. For flood hazard and risk assessment, the combined effects of flood depth and flood duration were considered. The flood hazard and flood risk maps were determined for each community in Nho Quan and Gia Vien district. These maps are useful for flood risk zoning and flood mitigation planning for the Ninh Binh province. INTRODUCTION Located in northern Vietnam, Hoang Long river basin is a tributary of Day river. Hoang Long river basin is formed by merging of Boi river and Hoang Long river at Ben De and with Day river at Gian Khau (Figures 1 and 2). The total basin area is 1515 km2. The causes of big floods in Hoang Long are monsoons such as tropical monsoons from the North Indian Ocean, equator monsoons from the south and monsoon from the Pacific Ocean. The combinations of these monsoons and climate turbulence (such as front, typhoons and tropical storms) lead to heavy rain over the basin. The riverbank system of Hoang Long river was constructed in order to protect flooding for Ninh Binh town, 1A Highway and Hanoi capital. When the water level in Hoang Long
P033-1
river is higher than +4 m at Ben De and Gian Khau stations, the government require some parts of riverbank along Hoang Long river to be broken to reduce flood flow in Day river, to protect Ninh Binh city and other downstream areas. Thus, some communities of Ninh Binh province may be become flooded storage areas, with the flooding return period from 3 to 5 years. Flooding in 2007 inundated 11 of 26 communities of Nho Quan district and 4 of 20 communities of Gia Vien district (People’s Committee of Ninh Binh province, 2007). These flooding storage areas are separated into two parts, Duc Long – Gia Tuong and Lac Khoai (Figure. 2). The total submerged areas was around 12 km2, the number of building was submerged in water around 14 thousands of 18 thousands household , the water depth in the flooded areas varies from 2 to 4 m. Total losses/damages estimated about 15 million USD. Besides, flooding affected to human living which can not be estimated in terms of money. Local people are always worried about the damages of flooding, their life, and their properties. The overall objective of this study is to (i) simulate and determine the magnitudes of flood flow along the rivers and floodplains for different return periods; (ii) develop inundation maps in flooding areas; (iii) estimate tangible impact on resident, agriculture and infrastructure for various return period floods; and (iv) develop flood hazard maps and flood risk maps.
Ba Tha
N
N
Phu PhuLy Ly
Nam Dinh
10 Km
Day Sea Port
Figure 1: Hoang Long, Day, Boi and Dao Rivers and Flood Plains
P033-2
Upstream Boi River
Upstream Day River Ba Tha
Hung Thi Boi River
Day River
Ben De Upstream Hoang Nho Hoang Long Long R. Quan River Legend Hydrological station Meteorological station Hydromet station
Phu Ly
Gian Khau Dao River
Upstream Dao River Nam Dinh
Doc Bo
Nhu Tan Day Sea Port
Figure 2: Schematic diagram of rivers METHODOLOGY Daily rainfalls, flood flow data and river basin geometry were collected for model simulation. By frequency analysis, maximum daily rainfall data is analyzed to determine design maximum rainfalls for different return periods. From measured daily rainfall data at Hung Thi, Ba Tha and Nho Quan, the design rainfall corresponding to 200, 100, 50, 20, 10 and 5 year return periods at each station were estimated. In this study, Pearson type-3 distribution was used for rainfall frequency analysis. Choosing an actual (measured) hyetograph which has total rainfall is close to design rainfall. The design hyetograph was obtained by multiplying the ratio between actual rainfall in a period and design rainfall in the same period to actual hyetograph. Corresponding runoffs were calculated using NAM model. The computed runoffs were input into MIKE 11 hydrodynamic model (DHI, 2003) for flood flow simulation. Questionnaires on affected population, economic values of flooded properties, flood depths and duration and flood damages were distributed and data are collected. Flood damages for various return periods and vulnerability of the flooded areas are estimated. Flood hazards are estimated based on flood depth and duration. Flood risks are then computed as product of flood hazard and vulnerability for different locations in the flooded areas.
P033-3
DATA COLLECTION AND ANALYSIS There are four rainfall stations in the study area which measured hourly or daily rainfall at Ba Tha, Hung Thi, Phu Ly, Nho Quan. Water level measuring stations in the study area are: Hung Thi, Ba Tha, Ben De, Gian Khau, Phu Ly and Doc Bo. In addition, Ba Tha and Hung Thi stations also measured discharges. 135 cross sections along the river of Day, Boi, Hoang Long and Dao river were measured in 2001. Digital Elevation Map (DEM) of 90 m x 90 m for Hoang Long basin. Reports on flood losses/damages due to flooding in 2007 and answers to questionnaires on information and data about flood losses in 2007 and 2008 are collected from Nho Quan district people’s committee and Gia Vien district people’s committee (Ninh Binh Statistical Office, 2008). Based on the data collected, the average annual rainfall of Hoang Long basin is 1925 mm (Nho Quan station). 85% of rainfall occurs in the flooding season (June to October). Heavy rainfalls normally occur in August or September. However, there are several years that heavy rainfalls occurred in October. The period of heavy rainfall is normally from 3 to 5 days, sometime up to 7 days. Hoang Long is the biggest river of Ninh Binh province; its flood regime and drainage are influenced by Day river and Dao river. The time to peak at these rivers normally different from several hours to few days depend on the position along the river. From the data collected at Hung Thi station (Boi river – upstream of Hoanglong river) and Ba Tha (Day river), the time to peak in Hoang Long river is sooner than Day river by 1 to 3 days. Questionnaires were distributed to local people in Nho Quan and Gia Vien districts to get information of flood damages per household, agricultural damages and road damages in the study area caused by floods in 2007 and 2008. In addition, all flood damages reports of provinces, districts or communes related are collected also. The interview and responses are based on floods occurred in 2007, 2008. This study focuses on estimating residential damages, agricultural damages and roads damages as described by Dutta et al (2003).Following responses of affected people, the average flood depth which does not effect to building and their activities is found to be 0.24 m. If the flooding occurs, the average duration of flooding is 7 days. In general, if flooding depth is under 0.6 m, the level of flooding impact will be not much for buildings as well as human activities. The duration of flooding is generally under 3 days. Most of the buildings are slightly affected in range from 0.5 to 1.5 m of depth and from 3 to 7 days of flooding duration, while some of the other areas have higher impact for 1.5 to 3.0 m of depth and 7 to over 20 days duration. The area which has flooding above 3.0 m is agricultural land and flooding duration is normally more than 3 weeks. RESUTLS AND DISCUSSION Rainfall runoff simulation by NAM model For Ba Tha sub-basin, the model calibration and verification was done considering the observed data from 1-30 August 1996 and 1-30 August 1974
P033-4
respectively (Figure 3). The difference between the computed and observed peak discharges is 0.65% for calibration and 10% for verification. The time to peak of the simulated hydrograph is sooner than the observed one by 1 day in calibration while there is no difference in verification. Both of calibration and verification have correlation coefficient (R) higher than 0.94. Calibration NAM model Observed and simulated daily discharge hydrographs at Ba Tha station from7-Aug to 30-Aug-1996
8/29
8/27
8/25
8/23
8/21
8/19
8/17
8/15
8/13
8/9
8/11
8/7
400 350 300 250 Q (m3/s) 200 150 100 50 0
Verification NAM model Observed and simulated daily discharge hydrographs at Ba Tha station from7-Aug to 29-Aug-1974 140 120 100 80 Q (m3/s) 60 40 20 0 8/7 8/9 8/11 8/13 8/15 8/17 8/19 8/21 8/23 8/25 8/27 8/29
Time
Time Observed
Observed
Simulated
Simulated
Figure 3: Calibration and verification NAM model for Ba Tha sub-basin For Hung Thi sub-basin, the peak discharges of observed hydrograph are higher than simulated hydrograph in both calibration and verification with difference ranging from 5.6% to 10.5%. However the times to peaks match very well. In addition, the correlation coefficient is very good, above 0.9. Estimating design rainfall hyetographs and developing flood design hydrographs As explained earlier, the method for estimating design hyetograph is done by choosing an actual hyetograph which has total rainfall is similar to design rainfall. The typical distribution should provide highest peak discharge hydrographs in sub-basin. Then, corresponding design hydrographs are determined by using NAM model (Figure 4). 1400
4000
1200
3500 3000
1000
2500
800
Q (m3/s) 2000
Q (m3/s) 600
1500
400
1000
0 1Se p 3Se p 5Se p 7Se p 9Se p 11 -S ep 13 -S ep 15 -S ep 17 -S ep 19 -S ep 21 -S ep 23 -S ep 25 -S ep 27 -S ep 29 -S ep
500
0
1Se p 3Se p 5Se p 7Se p 9Se p 11 -S ep 13 -S ep 15 -S ep 17 -S ep 19 -S ep 21 -S ep 23 -S ep 25 -S ep 27 -S ep 29 -S ep
200
Time (day) T=5 years
T=10 years
T=100 years
T=200 years
Ba Tha
T=20 years
Time (day)
T= 50 years
T=5 years
T=10 years
T=100 years
T=200 years
T=20 years
T= 50 years
Hung Thi
Figure 4: Design Hydrographs corresponding to various return periods In Nho Quan sub-basin, there is no hydrological station. Hung Thi sub-basin has similar characteristics with Nho Quan sub-basin about soil condition,
P033-5
plants cover, landuse, geological and meteorological. Thus, the parameters of NAM model for Hung Thi sub-basin are used for Nho Quan sub-basin to calculate design flood hydrographs. Flood Flow Simulation To simulate flood flow in river network, MIKE 11 model, is calibrated and verified by using collected data in the study area. The hydrographs corresponding to different return periods at Hung Thi, Ba Tha, Nho Quan obtained from output of MIKE-NAM and water level at Nam Dinh station are input at upstream boundary conditions of MIKE 11. The water level at Nhu Tan station is the input at downstream boundary condition. The outputs of model are the water level and discharge at every cross-section. For modeling floodplain, the FP4 method (DHI, 2003) is chosen for simulating the flood flows. Floodplains are modeled separately using a quasi 2-D approach. Three large floodplain areas along Hoang Long river named Gia Tuong, Duc Long and Lac Khoai are modeled as FP4 method. The exchange of water between river and floodplain is simulated through weir structures. Only maximum water level are available in floodplains, so calibration in floodplains is done by comparing observed and simulated maximum water level in floodplains. The exchange of water between river and floodplain were considered as flow through weir structures. The measured weir lengths were adjusted by trial and error to match the calibration (Vu, 2009). The bed resistance coefficient (Manning n) is adjusted for the rivers from upstream to downstream. All hourly rainfall, discharge and water level data at upstream and downstream boundaries observed from 6-Aug to 30-Aug-1996 is used for calibration; and from 1-30 September 2000 is used for verification (Figure 5). These results show that, the parameters of MIKE 11 for flood simulation in Hoang Long river network is acceptable. The Manning (n) values are determined for each river, ranging from 0.012 to 0.030. Estimation of floods for various return periods Considering that the return period of rainfall in upstream and of water level at downstream are the same design floods in river networks were calculated. The period of flood simulation is one month, assuming that flood occurs from 1 to 30 September using 1-day time step. The design hydrographs at Hung Thi, Ba Tha and Nho Quan obtained from NAM and the design water level hydrograph at Nam Dinh station are input as upstream boundary conditions of MIKE 11. The design water level hydrograph at Nhu Tan station is input as downstream boundary condition. Design water level hydrograph at Nhu Tan station is determined by considering past water level hydrographs which have the worst situation for drainage (October, 1978). MIKE 11 computes water levels in rivers and flood plains for design floods at various stations as shown in Tables 1 and 2.
P033-6
4.5 4.0 3.5 3.0 H 2.5 (m, MSL) 2.0
a) Phu Ly
1.5 1.0 0.5 0.0 6/8/1996
10/8/1996
14-8-1996 01:00:00
18-8-1996 01:00:00
22-8-1996 01:00:00
26-8-1996 01:00:00
Time (hour) Computed
H (m, MSL)
5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 6/8/1996
Observed
b) Gian Khau
10/8/1996 14-8-1996 18-8-1996 22-8-1996 26-8-1996 01:00:00 01:00:00 01:00:00 01:00:00 Time (hour) Computed
Observed
Figure 5: Computed and observed hourly water level hydrographs at a) Phu Ly and b) Gian Khau, 6 to 29 August 1996, Model Calibration Table 1: Maximum water level at various stations corresponding to different return periods of flood. Station Phu Ly Doc Bo Ben De Gian Khau
T=200 years 6.59 5.11 6.93 6.4
Maximum of water level (m, MSL) T=100 T=50 T=20 T=10 years years years years 6.20 5.69 4.86 4.30 4.80 4.50 3.97 3.78 6.33 5.78 5.23 4.95 6.07 5.76 4.81 4.21
T=5 years 3.99 3.48 4.71 3.94
P033-7
Table 2: Maximum water level at various flooded storages corresponding to different return periods of flood. Return periods T T T T T T
= = = = = =
5 years 10 years 20 years 50 years 100 years 200 years
Maximum water level (m, MSL) Gia Tuong Duc Long Lac Khoai 4.86 4.82 2.95 4.99 5.09 4.55 5.27 5.29 5.38 5.61 5.61 6.15 6.19 6.18 6.54 6.83 6.82 6.92
Constructing flood inundation maps The computed maximum water levels along the rivers and flood plains by MIKE 11 and digital elevation map (DEM 90m x 90m grid data for land elevation) were used for constructing inundation maps by MIKE-11 GIS (DHI, 2001). Based on the total rainfall observed at Hung Thi station and maximum water level at Ben De station, flood occurred in October 2007 is approximately has 20 year return period. The maximum of observed and simulated water levels at Ben De in Hoang Long river are respectively 5.17m and 5.19m. From maximum water depth data in floodplains in October 2007, the total flooding area in 2007 was estimated to be about 10,533.3 hectares. The difference of inundation area in 2007 and that of 20 year return period of flood is 6% (Figure 6). Overall, the result of comparison between the computed and observed maximum water depths, inundation areas in the floodplains are found to be close. The maximum of water level in flooding area can reach over 6.5 m when return period of flood T ≥ 100 years. The areas which have flooding depth over 6.0m are over 0.55% of total flooding area. While, the area which has flooding depth less than 4.5 m is about 80%, and 3.9% 12.8% from 4.5 m to 6.0m. For the case of T ≤ 50 years, most of flooding areas having flooding depth is less than 3.0 m while maximum of depth can reach 6.4 m. Flood damage estimation In general, inundation depth is the most important characteristic in damage estimation. Depth-damage functions are developed using information on land use, average assets values, inundation depth. The total flood damages in both Nho Quan and Gia Vien districts are calculated the sum of residential damages, agricultural damages and road damages. As shown in Table 3, the total flood damages sharply double when the return period increases from 50 to 100 years. The big difference in flooding is that in case 50-year return period, flooding occurs on only one side of Hoang Long river while in case of 100 year, flooding occurs on both sides of the river. The percentage of residential damage increases from 29.2% to 64.4% of total damages corresponding to 5-year return period to 200 year return period, respectively.
P033-8
20-year return period
100-year return period
Figure 6: Flood inundation maps for 20 and 100-year return periods Agricultural damage takes a high percentage of total damages; however the percentage of damage decreases from 54.4% to 28.8% of total damages. Both of residential and agricultural damages are over 80% of total damage, whereas roads damage is only below 20% of total in all cases. The total flood damages as a function of return period is shown in Figure 7.
P033-9
Table 3: Total damages due to flooding in Nho Quan and Gia Vien districts. Return period T (year) 5 10 20 50 100 200
Total damages (Mill. USD) 3.0882 7.2647 9.0000 13.9294 30.4882 33.9529
(%) 100.00 100.00 100.00 100.00 100.00 100.00
Residential damages (%) 29.2 42.6 42.5 52.6 61.7 64.4
Agricultural damages (%) 54.4 47.4 47.6 39.4 30.7 28.8
Road damages (%) 16.4 10.0 9.9 8.0 7.6 6.8
Figure 7: Relationship of total damages with probability of flood (Currency exchange rate: 1 USD = 17,000 Vietnamese Dong, VND) Flood Hazard and Flood Risk Assessment Flood hazard assessment depends on many parameters such as flooding depth, flooding duration, velocity of flood flow, timing and frequency of occurrence. Hazard index (HI) represents the level of flooding impacts. The hazard zone area is determined from hazard factor which represents the combination of all hazard parameters. To compute hazard factor and hazard zone for a community, hazard assessment is done for a flood of 20 years return period, which is approximately the flood occurred in October 2007. The inundation map for T = 20 years (Figure 6), the percentage of flooding area are 25%, 39%, 28% and 8% for depth categories under or equal 0.5m, 0.5 m to 1.5 m, 1.5 m to 3 m and above 3 m respectively. The result shows that the main part of flooding area under depth of 0.5 m to 3m. Table 4 shows the hazard index for flood depth. The flooding durations are determined based on the water depth in floodplain areas. The duration of flooding are checked satisfactory with the field survey results. The indices for flooding duration are shown in Table 5. Flooding depth and flooding duration are categorized based on the level of flooding impact. In general, if flooding depth is under 0.6 m, the level of
P033-10
flooding impact will be not much for buildings as well as human activities. Most of the buildings are slightly affected in range from 0.5 m to 1.5m of depth, while some of the other areas have higher impact for 1.5 m to 3.0m. The area which has flooding above 3.0 m is agricultural land. Table 4: Hazard index for flooding depth. Depth (m) 0.0 0.5 1.5 3.0
< < <
25 )
1 2 3 4
Hazard index Alternative 1 Alternative 2 0 0.5 1 1.0 2 1.5 3 2.0
To select hazard index value, a trial and error method suggested by Islam and Sado (2000) is used. There are two alternative sets of hazard index established for depth and duration of flooding. Hazard index represents the level of flooding impacts. There are four case studies done for hazard assessment using alternative hazard index values given in Table 6. Weighted hazard index indicates the resultant effect of all hazard categories for each community. Various flooding depths and flooding durations could occur in the same community, so that it is necessary to estimate the weighted average of hazard index (WHI) for each community: Hazard factor HF presents the combination of all hazard parameters, i.e.: HF = a.WHID + b.WHIT
(1)
where WHID is weighted area average hazard index for flooding depth, WHIT is weighted area average hazard index for flooding duration, a and b are weighting factors for flooding depth and duration. Table 6: Four combination cases of hazard index. Hazard index Depth of flooding Duration of flooding Alternative 1 Alternative 1 Alternative 2 Alternative 1 Alternative 1 Alternative 2 Alternative 2 Alternative 2 and 2 are defined in Table 4 and Table 5
Case study Case 1: Case 2: Case 3: Case 4: * Alternative 1
P033-11
To determine weighting factors for flooding depth and duration, sensitivity analysis is done. There are three sets of weighting factors considered: (i) Set 1: a =0.5, b =0.5; (ii) Set 2: a = 0.7, b =0.3; (iii) Set 3: a= 0.3, b = 0.7. The three sets yield nearly the same result. For simplicity, the weighting factors are taken as 0.5 for both depth and duration of flooding. The hazard factor is normalized by dividing HF by maximum hazard factor HFmax, i.e., HFn = (HF/HFmax) x100%. Tingsanchali and Karim (2004) suggested five levels of hazard zone to represent the hazard magnitude. These levels are categorized as: Very low (0% < HF ≤ 20%), Low (20% < HF ≤ 40%), Medium (40% < HF ≤ 60%), High (60% < HF ≤ 80%), Very high (80% < HF ≤ 100%). Case 1 has the area under medium and high hazard levels equal to 63.7% (Figure 8) which is very close to the actual flooding area of medium and high depth of flooding which is 67% of total flooding area. In case 2, 3 and 4 the area under medium and high hazard levels are 40.2%, 40.2% and 19.9% respectively. Therefore, case 1 is selected for further analysis in this study. From field surveys, average flood damage/person of a land unit (1 km2) in each district is determined. By knowing average population of land unit in the district, the flood damage of a land unit for the district can be computed. The vulnerability of a land unit is equal to damage of the land unit divided by maximum damage of a land unit of all districts in the study area. The vulnerability factor VF is expressed on a scale from 0 (no damages) to 1 (totally damaged). The risk factor (RF) is determined as: Risk factor (RF) = Hazard factor (HFn) x Vulnerability factor (VF)
(2)
Risk factor is normalized as: RFn = (RFi/RFmax) x 100, where RFn is normalized risk factor, RFi = risk factor for land unit i and RFmax is maximum risk factor for all land units. The risk zone is determined from risk factor are used to represent the risk level. The level of risk is categorized as: Very low (0% < RFn ≤ 20%), Low (20% < RFn ≤ 40%), Medium (40% < RFn ≤ 60%), High (60% < RFn ≤ 80%), Very high (80% < RFn ≤ 100%). The result of risk zone and percentage of flooding area and number of affected household corresponding to levels of risk zone is shown in Table 7. Figure 9 shows the flood risk map for 20-year return period for Ninh Binh Province. Table 7: Percentage of flooding area and number of affected household under different risk levels. Risk zone Very Low Low Medium High Very high
Number of community 2 12 7 2 2
Number of household 532 5,067 4,629 1,138 7,22
Area (%) 10.1 49.5 31.6 5.9 3.0
P033-12
Figure 8: Flood hazard map for 20-year return period flood, Ninh Binh province
Figure 9: Flood risk map for 20-year return period flood, Ninh Binh province
P033-13
CONCLUSIONS AND RECOMMENDATIONS Flood inundation, damage and risk assessment corresponding to various return periods of flood can be useful for constructing flood control system, infrastructure, buildings improvement, and flood warning. In addition, they are useful for rapid flood damage estimation in Hoang Long basin. Maximum of water level corresponding to different return periods were determined at various locations along the rivers and floodplains by using the NAM and MIKE 11 models which were calibrated and verified in this study. Flood duration in rivers may occur in a week, while flooding duration in floodplains can be up to one month. Inundation maps for various return periods developed for Ninh Binh province are checked with observed flooding depths at some locations. Flooding extents are quite comparable to flooded area which achieved from field survey. Tangible damages due to flooding on household, agriculture and roads corresponding to various return period floods were determined based on questionnaires and field surveys. Damage curves for residential, agricultural and roads were also developed. For hazard and risk assessment, the combined effect of flood depth and flood duration is considered by weighting factors for both depth and duration of flooding. Flood hazard and flood risk maps were developed for various return periods. The levels of hazard and risk were presented for each community in Nho Quan and Gia Vien district. REFERENCES DHI (Danish Hydraulic Institute) (2001) MIKE 11 GIS – User Guide. MIKE Package Software DHI (Danish Hydraulic Institute) (2003) User guide to MIKE 11: A modeling system for rivers and channels. DHI Software Manual, 460p. Dutta, D., Herath, S., Musiake, K. (2003) A mathematical model for flood loss estimation: Journal of Hydrology, vol. 277, pp 24-29. Islam, M. M., Sado, K. (2000) Flood hazard assessment in Bangladesh using NOAA AVHRR data with geographical information system. Hydrological Processes. vol. 14, no.3, pp. 605-620. People’s Committee (2007) The first estimation of losses caused by floods in Ninh Binh province – October, Report Number 29. Ninh Binh Statistical Office (2008) Statistical yearbook: Statistical Publishing House, Vietnam. Tingsanchali, T., Karim, M.F. (2005) Flood hazard and risk analysis in the southwest region of Bangladesh: Hydrological Processes, vol. 19, no. 10, pp 2055-2069, June. Vu, T.T. (2009) Flood inundation, damage and risk assessment in Hoang Long Basin, Vietnam; Thesis. Asian Institute of Technology, Thailand.
P033-14
Water Resources University, Assistant Professor, 175 Tay Son Street, Dong Da District, Hanoi, Vietnam 100000, e-mail: [email protected]
2
Nakhon Pathom Rajabhat University, Program of Civil and Environmental Engineering, Professor, Nakhon Pathom, Thailand 73000, e-mail:[email protected] Keywords Flooding depth, inundation map, flood damage assessment, flood hazard, risk Abstract Hoang Long is the largest river in Ninh Binh province, Vietnam. This province is frequently affected by flooding due to heavy rainfalls in the river basin. Rainfall frequency analysis was done to obtain a maximum 7-day rainfall for various return periods, and to construct design hyetographs corresponding to 5, 10, 20, 50, 100 and 200-year return periods. Design flood hydrographs corresponding to different return periods are obtained by using rainfall-runoff model (MIKE-NAM). These results are input to MIKE 11 hydrodynamic model to simulate flood flow in Hoang Long basin. The NAM and MIKE 11 models were calibrated and verified using past flood data. Using the maximum water levels computed by MIKE 11 model at different locations along the rivers and floodplains and digital elevation map (DEM), flood inundation maps were developed for Ninh Binh province. Flooding depth and inundation area corresponding to various return periods were determined. From the inundation maps and field survey data, damage curves for residential, agricultural and roads damages were developed. Flood hazard and flood risk maps were constructed typically for the flood of 20-year return period. For flood hazard and risk assessment, the combined effects of flood depth and flood duration were considered. The flood hazard and flood risk maps were determined for each community in Nho Quan and Gia Vien district. These maps are useful for flood risk zoning and flood mitigation planning for the Ninh Binh province. INTRODUCTION Located in northern Vietnam, Hoang Long river basin is a tributary of Day river. Hoang Long river basin is formed by merging of Boi river and Hoang Long river at Ben De and with Day river at Gian Khau (Figures 1 and 2). The total basin area is 1515 km2. The causes of big floods in Hoang Long are monsoons such as tropical monsoons from the North Indian Ocean, equator monsoons from the south and monsoon from the Pacific Ocean. The combinations of these monsoons and climate turbulence (such as front, typhoons and tropical storms) lead to heavy rain over the basin. The riverbank system of Hoang Long river was constructed in order to protect flooding for Ninh Binh town, 1A Highway and Hanoi capital. When the water level in Hoang Long
P033-1
river is higher than +4 m at Ben De and Gian Khau stations, the government require some parts of riverbank along Hoang Long river to be broken to reduce flood flow in Day river, to protect Ninh Binh city and other downstream areas. Thus, some communities of Ninh Binh province may be become flooded storage areas, with the flooding return period from 3 to 5 years. Flooding in 2007 inundated 11 of 26 communities of Nho Quan district and 4 of 20 communities of Gia Vien district (People’s Committee of Ninh Binh province, 2007). These flooding storage areas are separated into two parts, Duc Long – Gia Tuong and Lac Khoai (Figure. 2). The total submerged areas was around 12 km2, the number of building was submerged in water around 14 thousands of 18 thousands household , the water depth in the flooded areas varies from 2 to 4 m. Total losses/damages estimated about 15 million USD. Besides, flooding affected to human living which can not be estimated in terms of money. Local people are always worried about the damages of flooding, their life, and their properties. The overall objective of this study is to (i) simulate and determine the magnitudes of flood flow along the rivers and floodplains for different return periods; (ii) develop inundation maps in flooding areas; (iii) estimate tangible impact on resident, agriculture and infrastructure for various return period floods; and (iv) develop flood hazard maps and flood risk maps.
Ba Tha
N
N
Phu PhuLy Ly
Nam Dinh
10 Km
Day Sea Port
Figure 1: Hoang Long, Day, Boi and Dao Rivers and Flood Plains
P033-2
Upstream Boi River
Upstream Day River Ba Tha
Hung Thi Boi River
Day River
Ben De Upstream Hoang Nho Hoang Long Long R. Quan River Legend Hydrological station Meteorological station Hydromet station
Phu Ly
Gian Khau Dao River
Upstream Dao River Nam Dinh
Doc Bo
Nhu Tan Day Sea Port
Figure 2: Schematic diagram of rivers METHODOLOGY Daily rainfalls, flood flow data and river basin geometry were collected for model simulation. By frequency analysis, maximum daily rainfall data is analyzed to determine design maximum rainfalls for different return periods. From measured daily rainfall data at Hung Thi, Ba Tha and Nho Quan, the design rainfall corresponding to 200, 100, 50, 20, 10 and 5 year return periods at each station were estimated. In this study, Pearson type-3 distribution was used for rainfall frequency analysis. Choosing an actual (measured) hyetograph which has total rainfall is close to design rainfall. The design hyetograph was obtained by multiplying the ratio between actual rainfall in a period and design rainfall in the same period to actual hyetograph. Corresponding runoffs were calculated using NAM model. The computed runoffs were input into MIKE 11 hydrodynamic model (DHI, 2003) for flood flow simulation. Questionnaires on affected population, economic values of flooded properties, flood depths and duration and flood damages were distributed and data are collected. Flood damages for various return periods and vulnerability of the flooded areas are estimated. Flood hazards are estimated based on flood depth and duration. Flood risks are then computed as product of flood hazard and vulnerability for different locations in the flooded areas.
P033-3
DATA COLLECTION AND ANALYSIS There are four rainfall stations in the study area which measured hourly or daily rainfall at Ba Tha, Hung Thi, Phu Ly, Nho Quan. Water level measuring stations in the study area are: Hung Thi, Ba Tha, Ben De, Gian Khau, Phu Ly and Doc Bo. In addition, Ba Tha and Hung Thi stations also measured discharges. 135 cross sections along the river of Day, Boi, Hoang Long and Dao river were measured in 2001. Digital Elevation Map (DEM) of 90 m x 90 m for Hoang Long basin. Reports on flood losses/damages due to flooding in 2007 and answers to questionnaires on information and data about flood losses in 2007 and 2008 are collected from Nho Quan district people’s committee and Gia Vien district people’s committee (Ninh Binh Statistical Office, 2008). Based on the data collected, the average annual rainfall of Hoang Long basin is 1925 mm (Nho Quan station). 85% of rainfall occurs in the flooding season (June to October). Heavy rainfalls normally occur in August or September. However, there are several years that heavy rainfalls occurred in October. The period of heavy rainfall is normally from 3 to 5 days, sometime up to 7 days. Hoang Long is the biggest river of Ninh Binh province; its flood regime and drainage are influenced by Day river and Dao river. The time to peak at these rivers normally different from several hours to few days depend on the position along the river. From the data collected at Hung Thi station (Boi river – upstream of Hoanglong river) and Ba Tha (Day river), the time to peak in Hoang Long river is sooner than Day river by 1 to 3 days. Questionnaires were distributed to local people in Nho Quan and Gia Vien districts to get information of flood damages per household, agricultural damages and road damages in the study area caused by floods in 2007 and 2008. In addition, all flood damages reports of provinces, districts or communes related are collected also. The interview and responses are based on floods occurred in 2007, 2008. This study focuses on estimating residential damages, agricultural damages and roads damages as described by Dutta et al (2003).Following responses of affected people, the average flood depth which does not effect to building and their activities is found to be 0.24 m. If the flooding occurs, the average duration of flooding is 7 days. In general, if flooding depth is under 0.6 m, the level of flooding impact will be not much for buildings as well as human activities. The duration of flooding is generally under 3 days. Most of the buildings are slightly affected in range from 0.5 to 1.5 m of depth and from 3 to 7 days of flooding duration, while some of the other areas have higher impact for 1.5 to 3.0 m of depth and 7 to over 20 days duration. The area which has flooding above 3.0 m is agricultural land and flooding duration is normally more than 3 weeks. RESUTLS AND DISCUSSION Rainfall runoff simulation by NAM model For Ba Tha sub-basin, the model calibration and verification was done considering the observed data from 1-30 August 1996 and 1-30 August 1974
P033-4
respectively (Figure 3). The difference between the computed and observed peak discharges is 0.65% for calibration and 10% for verification. The time to peak of the simulated hydrograph is sooner than the observed one by 1 day in calibration while there is no difference in verification. Both of calibration and verification have correlation coefficient (R) higher than 0.94. Calibration NAM model Observed and simulated daily discharge hydrographs at Ba Tha station from7-Aug to 30-Aug-1996
8/29
8/27
8/25
8/23
8/21
8/19
8/17
8/15
8/13
8/9
8/11
8/7
400 350 300 250 Q (m3/s) 200 150 100 50 0
Verification NAM model Observed and simulated daily discharge hydrographs at Ba Tha station from7-Aug to 29-Aug-1974 140 120 100 80 Q (m3/s) 60 40 20 0 8/7 8/9 8/11 8/13 8/15 8/17 8/19 8/21 8/23 8/25 8/27 8/29
Time
Time Observed
Observed
Simulated
Simulated
Figure 3: Calibration and verification NAM model for Ba Tha sub-basin For Hung Thi sub-basin, the peak discharges of observed hydrograph are higher than simulated hydrograph in both calibration and verification with difference ranging from 5.6% to 10.5%. However the times to peaks match very well. In addition, the correlation coefficient is very good, above 0.9. Estimating design rainfall hyetographs and developing flood design hydrographs As explained earlier, the method for estimating design hyetograph is done by choosing an actual hyetograph which has total rainfall is similar to design rainfall. The typical distribution should provide highest peak discharge hydrographs in sub-basin. Then, corresponding design hydrographs are determined by using NAM model (Figure 4). 1400
4000
1200
3500 3000
1000
2500
800
Q (m3/s) 2000
Q (m3/s) 600
1500
400
1000
0 1Se p 3Se p 5Se p 7Se p 9Se p 11 -S ep 13 -S ep 15 -S ep 17 -S ep 19 -S ep 21 -S ep 23 -S ep 25 -S ep 27 -S ep 29 -S ep
500
0
1Se p 3Se p 5Se p 7Se p 9Se p 11 -S ep 13 -S ep 15 -S ep 17 -S ep 19 -S ep 21 -S ep 23 -S ep 25 -S ep 27 -S ep 29 -S ep
200
Time (day) T=5 years
T=10 years
T=100 years
T=200 years
Ba Tha
T=20 years
Time (day)
T= 50 years
T=5 years
T=10 years
T=100 years
T=200 years
T=20 years
T= 50 years
Hung Thi
Figure 4: Design Hydrographs corresponding to various return periods In Nho Quan sub-basin, there is no hydrological station. Hung Thi sub-basin has similar characteristics with Nho Quan sub-basin about soil condition,
P033-5
plants cover, landuse, geological and meteorological. Thus, the parameters of NAM model for Hung Thi sub-basin are used for Nho Quan sub-basin to calculate design flood hydrographs. Flood Flow Simulation To simulate flood flow in river network, MIKE 11 model, is calibrated and verified by using collected data in the study area. The hydrographs corresponding to different return periods at Hung Thi, Ba Tha, Nho Quan obtained from output of MIKE-NAM and water level at Nam Dinh station are input at upstream boundary conditions of MIKE 11. The water level at Nhu Tan station is the input at downstream boundary condition. The outputs of model are the water level and discharge at every cross-section. For modeling floodplain, the FP4 method (DHI, 2003) is chosen for simulating the flood flows. Floodplains are modeled separately using a quasi 2-D approach. Three large floodplain areas along Hoang Long river named Gia Tuong, Duc Long and Lac Khoai are modeled as FP4 method. The exchange of water between river and floodplain is simulated through weir structures. Only maximum water level are available in floodplains, so calibration in floodplains is done by comparing observed and simulated maximum water level in floodplains. The exchange of water between river and floodplain were considered as flow through weir structures. The measured weir lengths were adjusted by trial and error to match the calibration (Vu, 2009). The bed resistance coefficient (Manning n) is adjusted for the rivers from upstream to downstream. All hourly rainfall, discharge and water level data at upstream and downstream boundaries observed from 6-Aug to 30-Aug-1996 is used for calibration; and from 1-30 September 2000 is used for verification (Figure 5). These results show that, the parameters of MIKE 11 for flood simulation in Hoang Long river network is acceptable. The Manning (n) values are determined for each river, ranging from 0.012 to 0.030. Estimation of floods for various return periods Considering that the return period of rainfall in upstream and of water level at downstream are the same design floods in river networks were calculated. The period of flood simulation is one month, assuming that flood occurs from 1 to 30 September using 1-day time step. The design hydrographs at Hung Thi, Ba Tha and Nho Quan obtained from NAM and the design water level hydrograph at Nam Dinh station are input as upstream boundary conditions of MIKE 11. The design water level hydrograph at Nhu Tan station is input as downstream boundary condition. Design water level hydrograph at Nhu Tan station is determined by considering past water level hydrographs which have the worst situation for drainage (October, 1978). MIKE 11 computes water levels in rivers and flood plains for design floods at various stations as shown in Tables 1 and 2.
P033-6
4.5 4.0 3.5 3.0 H 2.5 (m, MSL) 2.0
a) Phu Ly
1.5 1.0 0.5 0.0 6/8/1996
10/8/1996
14-8-1996 01:00:00
18-8-1996 01:00:00
22-8-1996 01:00:00
26-8-1996 01:00:00
Time (hour) Computed
H (m, MSL)
5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 6/8/1996
Observed
b) Gian Khau
10/8/1996 14-8-1996 18-8-1996 22-8-1996 26-8-1996 01:00:00 01:00:00 01:00:00 01:00:00 Time (hour) Computed
Observed
Figure 5: Computed and observed hourly water level hydrographs at a) Phu Ly and b) Gian Khau, 6 to 29 August 1996, Model Calibration Table 1: Maximum water level at various stations corresponding to different return periods of flood. Station Phu Ly Doc Bo Ben De Gian Khau
T=200 years 6.59 5.11 6.93 6.4
Maximum of water level (m, MSL) T=100 T=50 T=20 T=10 years years years years 6.20 5.69 4.86 4.30 4.80 4.50 3.97 3.78 6.33 5.78 5.23 4.95 6.07 5.76 4.81 4.21
T=5 years 3.99 3.48 4.71 3.94
P033-7
Table 2: Maximum water level at various flooded storages corresponding to different return periods of flood. Return periods T T T T T T
= = = = = =
5 years 10 years 20 years 50 years 100 years 200 years
Maximum water level (m, MSL) Gia Tuong Duc Long Lac Khoai 4.86 4.82 2.95 4.99 5.09 4.55 5.27 5.29 5.38 5.61 5.61 6.15 6.19 6.18 6.54 6.83 6.82 6.92
Constructing flood inundation maps The computed maximum water levels along the rivers and flood plains by MIKE 11 and digital elevation map (DEM 90m x 90m grid data for land elevation) were used for constructing inundation maps by MIKE-11 GIS (DHI, 2001). Based on the total rainfall observed at Hung Thi station and maximum water level at Ben De station, flood occurred in October 2007 is approximately has 20 year return period. The maximum of observed and simulated water levels at Ben De in Hoang Long river are respectively 5.17m and 5.19m. From maximum water depth data in floodplains in October 2007, the total flooding area in 2007 was estimated to be about 10,533.3 hectares. The difference of inundation area in 2007 and that of 20 year return period of flood is 6% (Figure 6). Overall, the result of comparison between the computed and observed maximum water depths, inundation areas in the floodplains are found to be close. The maximum of water level in flooding area can reach over 6.5 m when return period of flood T ≥ 100 years. The areas which have flooding depth over 6.0m are over 0.55% of total flooding area. While, the area which has flooding depth less than 4.5 m is about 80%, and 3.9% 12.8% from 4.5 m to 6.0m. For the case of T ≤ 50 years, most of flooding areas having flooding depth is less than 3.0 m while maximum of depth can reach 6.4 m. Flood damage estimation In general, inundation depth is the most important characteristic in damage estimation. Depth-damage functions are developed using information on land use, average assets values, inundation depth. The total flood damages in both Nho Quan and Gia Vien districts are calculated the sum of residential damages, agricultural damages and road damages. As shown in Table 3, the total flood damages sharply double when the return period increases from 50 to 100 years. The big difference in flooding is that in case 50-year return period, flooding occurs on only one side of Hoang Long river while in case of 100 year, flooding occurs on both sides of the river. The percentage of residential damage increases from 29.2% to 64.4% of total damages corresponding to 5-year return period to 200 year return period, respectively.
P033-8
20-year return period
100-year return period
Figure 6: Flood inundation maps for 20 and 100-year return periods Agricultural damage takes a high percentage of total damages; however the percentage of damage decreases from 54.4% to 28.8% of total damages. Both of residential and agricultural damages are over 80% of total damage, whereas roads damage is only below 20% of total in all cases. The total flood damages as a function of return period is shown in Figure 7.
P033-9
Table 3: Total damages due to flooding in Nho Quan and Gia Vien districts. Return period T (year) 5 10 20 50 100 200
Total damages (Mill. USD) 3.0882 7.2647 9.0000 13.9294 30.4882 33.9529
(%) 100.00 100.00 100.00 100.00 100.00 100.00
Residential damages (%) 29.2 42.6 42.5 52.6 61.7 64.4
Agricultural damages (%) 54.4 47.4 47.6 39.4 30.7 28.8
Road damages (%) 16.4 10.0 9.9 8.0 7.6 6.8
Figure 7: Relationship of total damages with probability of flood (Currency exchange rate: 1 USD = 17,000 Vietnamese Dong, VND) Flood Hazard and Flood Risk Assessment Flood hazard assessment depends on many parameters such as flooding depth, flooding duration, velocity of flood flow, timing and frequency of occurrence. Hazard index (HI) represents the level of flooding impacts. The hazard zone area is determined from hazard factor which represents the combination of all hazard parameters. To compute hazard factor and hazard zone for a community, hazard assessment is done for a flood of 20 years return period, which is approximately the flood occurred in October 2007. The inundation map for T = 20 years (Figure 6), the percentage of flooding area are 25%, 39%, 28% and 8% for depth categories under or equal 0.5m, 0.5 m to 1.5 m, 1.5 m to 3 m and above 3 m respectively. The result shows that the main part of flooding area under depth of 0.5 m to 3m. Table 4 shows the hazard index for flood depth. The flooding durations are determined based on the water depth in floodplain areas. The duration of flooding are checked satisfactory with the field survey results. The indices for flooding duration are shown in Table 5. Flooding depth and flooding duration are categorized based on the level of flooding impact. In general, if flooding depth is under 0.6 m, the level of
P033-10
flooding impact will be not much for buildings as well as human activities. Most of the buildings are slightly affected in range from 0.5 m to 1.5m of depth, while some of the other areas have higher impact for 1.5 m to 3.0m. The area which has flooding above 3.0 m is agricultural land. Table 4: Hazard index for flooding depth. Depth (m) 0.0 0.5 1.5 3.0
< < <
25 )
1 2 3 4
Hazard index Alternative 1 Alternative 2 0 0.5 1 1.0 2 1.5 3 2.0
To select hazard index value, a trial and error method suggested by Islam and Sado (2000) is used. There are two alternative sets of hazard index established for depth and duration of flooding. Hazard index represents the level of flooding impacts. There are four case studies done for hazard assessment using alternative hazard index values given in Table 6. Weighted hazard index indicates the resultant effect of all hazard categories for each community. Various flooding depths and flooding durations could occur in the same community, so that it is necessary to estimate the weighted average of hazard index (WHI) for each community: Hazard factor HF presents the combination of all hazard parameters, i.e.: HF = a.WHID + b.WHIT
(1)
where WHID is weighted area average hazard index for flooding depth, WHIT is weighted area average hazard index for flooding duration, a and b are weighting factors for flooding depth and duration. Table 6: Four combination cases of hazard index. Hazard index Depth of flooding Duration of flooding Alternative 1 Alternative 1 Alternative 2 Alternative 1 Alternative 1 Alternative 2 Alternative 2 Alternative 2 and 2 are defined in Table 4 and Table 5
Case study Case 1: Case 2: Case 3: Case 4: * Alternative 1
P033-11
To determine weighting factors for flooding depth and duration, sensitivity analysis is done. There are three sets of weighting factors considered: (i) Set 1: a =0.5, b =0.5; (ii) Set 2: a = 0.7, b =0.3; (iii) Set 3: a= 0.3, b = 0.7. The three sets yield nearly the same result. For simplicity, the weighting factors are taken as 0.5 for both depth and duration of flooding. The hazard factor is normalized by dividing HF by maximum hazard factor HFmax, i.e., HFn = (HF/HFmax) x100%. Tingsanchali and Karim (2004) suggested five levels of hazard zone to represent the hazard magnitude. These levels are categorized as: Very low (0% < HF ≤ 20%), Low (20% < HF ≤ 40%), Medium (40% < HF ≤ 60%), High (60% < HF ≤ 80%), Very high (80% < HF ≤ 100%). Case 1 has the area under medium and high hazard levels equal to 63.7% (Figure 8) which is very close to the actual flooding area of medium and high depth of flooding which is 67% of total flooding area. In case 2, 3 and 4 the area under medium and high hazard levels are 40.2%, 40.2% and 19.9% respectively. Therefore, case 1 is selected for further analysis in this study. From field surveys, average flood damage/person of a land unit (1 km2) in each district is determined. By knowing average population of land unit in the district, the flood damage of a land unit for the district can be computed. The vulnerability of a land unit is equal to damage of the land unit divided by maximum damage of a land unit of all districts in the study area. The vulnerability factor VF is expressed on a scale from 0 (no damages) to 1 (totally damaged). The risk factor (RF) is determined as: Risk factor (RF) = Hazard factor (HFn) x Vulnerability factor (VF)
(2)
Risk factor is normalized as: RFn = (RFi/RFmax) x 100, where RFn is normalized risk factor, RFi = risk factor for land unit i and RFmax is maximum risk factor for all land units. The risk zone is determined from risk factor are used to represent the risk level. The level of risk is categorized as: Very low (0% < RFn ≤ 20%), Low (20% < RFn ≤ 40%), Medium (40% < RFn ≤ 60%), High (60% < RFn ≤ 80%), Very high (80% < RFn ≤ 100%). The result of risk zone and percentage of flooding area and number of affected household corresponding to levels of risk zone is shown in Table 7. Figure 9 shows the flood risk map for 20-year return period for Ninh Binh Province. Table 7: Percentage of flooding area and number of affected household under different risk levels. Risk zone Very Low Low Medium High Very high
Number of community 2 12 7 2 2
Number of household 532 5,067 4,629 1,138 7,22
Area (%) 10.1 49.5 31.6 5.9 3.0
P033-12
Figure 8: Flood hazard map for 20-year return period flood, Ninh Binh province
Figure 9: Flood risk map for 20-year return period flood, Ninh Binh province
P033-13
CONCLUSIONS AND RECOMMENDATIONS Flood inundation, damage and risk assessment corresponding to various return periods of flood can be useful for constructing flood control system, infrastructure, buildings improvement, and flood warning. In addition, they are useful for rapid flood damage estimation in Hoang Long basin. Maximum of water level corresponding to different return periods were determined at various locations along the rivers and floodplains by using the NAM and MIKE 11 models which were calibrated and verified in this study. Flood duration in rivers may occur in a week, while flooding duration in floodplains can be up to one month. Inundation maps for various return periods developed for Ninh Binh province are checked with observed flooding depths at some locations. Flooding extents are quite comparable to flooded area which achieved from field survey. Tangible damages due to flooding on household, agriculture and roads corresponding to various return period floods were determined based on questionnaires and field surveys. Damage curves for residential, agricultural and roads were also developed. For hazard and risk assessment, the combined effect of flood depth and flood duration is considered by weighting factors for both depth and duration of flooding. Flood hazard and flood risk maps were developed for various return periods. The levels of hazard and risk were presented for each community in Nho Quan and Gia Vien district. REFERENCES DHI (Danish Hydraulic Institute) (2001) MIKE 11 GIS – User Guide. MIKE Package Software DHI (Danish Hydraulic Institute) (2003) User guide to MIKE 11: A modeling system for rivers and channels. DHI Software Manual, 460p. Dutta, D., Herath, S., Musiake, K. (2003) A mathematical model for flood loss estimation: Journal of Hydrology, vol. 277, pp 24-29. Islam, M. M., Sado, K. (2000) Flood hazard assessment in Bangladesh using NOAA AVHRR data with geographical information system. Hydrological Processes. vol. 14, no.3, pp. 605-620. People’s Committee (2007) The first estimation of losses caused by floods in Ninh Binh province – October, Report Number 29. Ninh Binh Statistical Office (2008) Statistical yearbook: Statistical Publishing House, Vietnam. Tingsanchali, T., Karim, M.F. (2005) Flood hazard and risk analysis in the southwest region of Bangladesh: Hydrological Processes, vol. 19, no. 10, pp 2055-2069, June. Vu, T.T. (2009) Flood inundation, damage and risk assessment in Hoang Long Basin, Vietnam; Thesis. Asian Institute of Technology, Thailand.
P033-14
