Study on flood risk assessment in downstream area in ke go reservoir, ha tinh province

CS 7 Satellite image Image. Jpg June 2 nd 2013 4.3.2. Data analysis 4.3.2.1. Hydro-meteorological data a. Runoff data The hourly discharge flow into Ke Go reservoir was calculated from water level and release of Ke Go reservoir based on the water balance equation (shown Appendix 1). The results on hourly discharge flow into Ke Go reservoir of flood event in 2010 and 2013 are shown in Figure 4 - 4 and Figure 4 - 5. The calculated results are used to setup rainfall runoff model (UHM) to determine model parameters for Ke Go catchment and sub-basin of downstream area. 37 Figure 4 - 4: Inflow of Ke Go reservoir in October, 2010 Figure 4 - 5: Inflow of Ke Go reservoir in October, 2013 b. Water level All of water level measuring stations in the study is Cam Nhuong, Do Ho, Thach Dong which are affected by the tide and are the station measuring water level in short period from 1970 to present. Besides, there are local stations belong Ke Go 0 500 1000 1500 2000 2500 3000 3500 4000 4500 10/2 10/4 10/6 10/8 10/10 10/12 10/14 10/16 10/18 10/20 10/22 D is ch ar ge (m 3/ s) Time Inflow of Ke Go reservoir in October, 2010 Inflow 0 50 100 150 200 250 300 350 4000 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0 5 10 15 20 25 R ai nf al l (m m ) D is ch ar g e (m 3 /s ) Hour Measured rainfall and flood at Ke Go reservoir in Oct 16th 2013 Rainfall (mm) Discharge (m3/s) 38 irrigation company to control water level in downstream area when releasing flood of Ke Go reservoir. They are only measured when flood coming. This thesis collected water level data all of above station in October, 2010. The main stations are Phu and Hoi bridges are shown in Figure 4 - 6. They will be used to calibrate and verify numerical model in next part. Source: Ke Go Irrigation Company, 2010 Figure 4 - 6: Water level at Phu and Hoi Bridge station in October, 2010 4.3.2.2. Topography data A significant input for hydrodynamic modeling is the correct representation of terrain on which the model will work on. To develop DEM the following dataset were using: 1. Topographic maps of the study area in 1:10,000 scales (measured 2010). Projection is VN 2000 that was transformed to WGS 1984. 2. The total cross section data along the river of Rao Cai and Gia Hoi is 40 cross sections, in which Rao Cai has 25, Gia Hoi has 15. These cross sections were measured in 2011, which are collected from ―Emergency preparedness plans (EPP) 0 0.5 1 1.5 2 2.5 3 3.5 10/4/2010 0:00 10/8/2010 0:00 10/12/2010 0:00 10/16/2010 0:00 W a te r le v e l (m ) Time (hour) Measured water level at Cau Phu and Cau Hoi in October 2010 Phu Bridge Hoi Bridge 39 in emergency case of Ke Go Reservoir - Ha Tinh province‖ project of Thuyloi University. 40 CHAPTER 5: RESULTS AND DISSCUSIONS 5.1. The reasons cause the flooding in downstream area According to analysis of data about hydro-meteorology, statistic data analysis, flood and flooding occur more frequently in recently. In this part, thesis will research reasons caused flooding in downstream of Rao Cai river basin, especially in Ha Tinh city area. 5.1.1. Climate change impacts Along with the status of climate change (CC) on the global recently, Ha Tinh province also has effect of climate change lead to the great disaster suffered more major effect on the development of socio - economic, military and security. According to recent researches were done by the Ha Tinh Department of Natural Resources and Environment, the average temperature in the province increase from 0.1 – 0.2 0 C per decade, the average temperature period 2000-2010 increased more from 0.3 – 0.6 0 C than 10 - 30 years ago, especially, Huong Khe area increased from 0.7 - 1,4 0 C. Meanwhile, annual rainfall tends reduced greater with variation in terms of space, time and intensity. Although the rainfall reduces, the intensive of rainfall caused flooding or flash floods increases. Accordingly, the frequency and regularity of the storm is also changed, which will be described below. Normally, the rainy season in Ha Tinh province is from September to November and the storm season is from July to September. However, recently, the storm season was tends to change clearly, from August to December. (Hoai, 2009) Instability in rainfall would cause more severe floods in rainy season and droughts in dry season. • Increase in frequency and intensity of typhoons, storms would cause high floods & inundation, flash floods, landslide and erosion. • Increasing water shortage and growing water demand threaten water supply, water use conflicts. .To estimate climate change impact cause to flooding in Rai Cai river basin in recent year based on meteorological data in the study area and surrounding area from 1975 to 2005. Some annual rainfall changes are shown below: 41 Figure 5 - 1: Annual rainfall change on the Rao Cai river basin from 1975 – 2005 In the Figure 5 - 1, in general, the average annual rainfall change of Huong Khe and Ha Tinh have decrease trend. The decrease of annual rainfall of Ky Anh station is higher than Ha Tinh ones with 9.42 and 4.03 mm per year period from 1975 to 2005 respectively. This result is suitable with researches of DONRE of Ha Tinh province. However, rainfall intensive (mm/m, mm/day) have opposite trend. The results changing of 1 and 3 days maximum rainfalls are shown the following Figure 5 - 2. y = -4.0289x + 10713 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 1975 1980 1985 1990 1995 2000 2005 X (mm) Year Ha Tinh station y = -9.4216x + 21664 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 1975 1980 1985 1990 1995 2000 2005 X (mm) Year Ky Anh station 42 Figure 5 - 2: Changing of maximum rainfall of Ha Tinh station Based on rainfall of Ha Tinh meteorological station period from 1975 to 2005, daily maximum rainfall changes have stable trend or reduce but not much. The trend of daily maximum rainfall, which indicated various changes (including decrease and increase) in the first years from 1975 to 1995, ending the period with significantly increased. For the maximum rainfall during 3 days shows an increasing trend, this rainfall duration is main reason cause to flooding in the downstream area. To have an accurate estimate the data need be updated latest meteorological data. Interestingly, in recent year although daily maximum rainfall of Rao Cai area in 2010 (455.6mm) is higher than 5 time in 2013 (95mm) however peak flow of Ke Go reservoir are estimated as equal. Main reason is rainfall in 2013 occurred a few hours. In addition, maximum rainfall in 2010 and 2013 correspond to frequency 10% and 90%, however, the peak flow of Ke Go reservoir in both year are higher than design peak flow correspond about 0.5%. Rainfall occurred more frequently with higher density, especially in short duration. 0 100 200 300 400 500 600 700 1975 1980 1985 1990 1995 2000 2005 X (mm) Time Changing of maximum total rainfall in a day Rainfall Linear (Rainfall) 0 200 400 600 800 1000 1200 1975 1985 1995 2005 X (mm) Time Changing of maximum total rainfall in 3 days Rainfall Linear (Rainfall) 43 Figure 5 - 3: Inflow of Ke Go reservoir and actual rainfall 2013 These results revealed traditional method to calculate design flow for reservoir need to check again and consider hyetograph or climate change impacts to inflow. In conclusion, with climate change impact inflow into Ke Go reservoir and flow of sub-basin in downstream increase, this cause flooding when happening heavy rainfall. 5.1.2. Infrastructure impacts Further investigating and finding the main reasons cause this situation is due to the drainage system in the province of Ha Tinh was built long ago, lack of synchronized planning, many area seriously degraded, small drainage tunnel cannot get up when heavy rainfall occur during a long time. On other hand, in the process of urbanization, cities are focused on building many projects. Besides, the elevation of some area in HaTinh city is quite low, that might be flooded even when the small rainfall. The roads can be as a river dyke to keep flood in upstream of river, this make water level in upstream increased. This issue increase hazard for resident. 5.2. Flood hazard 5.2.1. Rainfall runoff modeling 5.2.1.1 Model setup Rainfall frequency analysis Determine statistical characteristics of frequency curve 0 100 200 300 4000 1000 2000 3000 4000 5000 0 5 10 15 20 25 X(mm) Q (m3/s) Hour Rainfall and flood at Ke Go reservoir in Oct 16th 2013 Rainfall (mm) Discharge (m3/s) 0 1000 2000 3000 4000 5000 6000 0 2 4 6 8 10 12 14 16 Q (m3/s) Hrs Design Hydrographs of Ke Go reservoir P = 0.5% P = 0.1% P = 0.01% 44 Based on the data of daily rainfall of the meteorological stations: Ha Tinh (1975- 2010), Ky Anh (1975- 2006) and Huong Khe (1975- 2006) in Rao Cai river basin and surrounding, using Pearson type III distribution, daily maximum rainfall corresponding to various return periods of 1000 and 200 years for each station are estimated (see Table 5-1). Table 5 - 1: Result of frequency analysis of maximum daily rainfall Station Avg. Rainfall (mm) Cv Cs Daily maximum rainfall corresponding to return periods (mm) 1000 years 200 years Ha Tinh 304.8 0.43 0.89 882.6 750.9 Ky Anh 292.6 0.36 0.60 711.6 623.6 Huong Khe 246.7 0.37 0.82 639.7 551.7 Frequency curve of maximum rainfall during 1 day and design rainfall hyetography of above these are indicated in Appendix 1 Determining design rainfalls The design rainfall of meteorological stations on Rao Cai river basin are estimated based on the ratio of actual daily rainfall to design daily rainfall responding to different frequencies and the actual time distribution of chosen days. Table 5 - 2: Value of design rainfall distribution of Ha Tinh, Ky Anh and Huong Khe stations during 1 day corresponding to difference frequency (daily rainfall) Stations Frequency Type Day Ratio 1st 2nd 3th H a T in h 0.10% Actual (15 - 17/10/2010) 132 610 147 1.5 Design 191 883 213 0.50% Actual (15 - 17/10/2010) 132 610 147 1.2 Design 163 751 182 H u o n g K h e 0.10% Actual (15 - 17/10/2010) 53 524 86 1.2 Design 65 640 105 45 Stations Frequency Type Day Ratio 1st 2nd 3th 0.50% Actual (15 - 17/10/2010) 53 524 86 1.1 Design 56 552 90 K y A n h 0.10% Actual (13 - 15/10/1984) 85 519 122 1.4 Design 117 712 167 0.50% Actual (13 - 15/10/1984) 85 519 122 1.2 Design 102 624 146 Definition of sub-basin of Rai Cai river basin The MIKE - UHM was used to calculate lateral inflow. In order to develop lateral inflow of sub-basins corresponding to rainfall frequencies, MIKE - UHM (SCS) model is used. The results of this model are input for MIKE 11 HD as point source in downstream area in river network. The sub-catchment delineations and their tributaries (i.e. areas, average surface elevation and slope) were initially calculated from the DEM using the spatial extension tools of ArcGIS software. The UHM model for sub-basin are considered land use, elevation, slopes, and drainage system, also included for schematization as shown in (Figure 5 - 4). Figure 5 - 4: Rao Cai watershed sub-basin schematizations and Thiessen polygon weighting computation of mean rainfall of sub-catchment in Rao Cai river basin Meteorology Station 46 The spatial distribution of rainfall for Ke Go reservoir catchment and sub-catchment in downstream area was calculated based on the Thiessen polygon concept. And the results have shown in Table 5 – 3. Table 5 - 3: Sub-catchment of Rao Cai river basin and weighting factors of meteorological station No Sub- catchment Area (km 2 ) Weighting factor of each station Description Ha Tinh Ky Anh Huong Khe 1. Sub 1 223 0.59 0.20 0.20 Ke Go. Reservoir 2. Sub 2 65.93 1 Downstream Ke Go Reservoir to junction of Ngan Mo 3. Sub 3 70.58 1 Ngan Mo to Cau Ho on Gia Hoi river 4. Sub 4 57.80 0.93 0.07 Quen Catchment 5. Sub 5 255.73 0.08 0.92 Rac Catchment 6. Sub 6 49.39 1 Ngan Mo to Cau Phuon Ngan Mo river 7. Sub 7 80.55 1 Bang river 8. Sub 8 39.66 1 Thach Dong estuary 9. Sub 9 58.55 1 Cam Cua Nhuong 5.2.1.2 Calibration and verification of MIKE_RR UHM model for Ke Go reservoir catchment Calibration The area of Ke Go reservoir catchment is 223 km 2 . In order to calibrate MIKE RR UHM for this catchment, the hourly rainfall and discharge at Ke Go reservoir from Oct 4 th to 11 th Oct 2010 were used to compare between measured and calculated data. The parameters and initial condition are determined (Figure 5 - 5) based on the criteria such as (Coefficient of efficiency (NASH) and volume error, peak time error). 47 Figure 5 - 5: Parameters of UHM in MIKE RR model of Ke Go catchment Table 5 - 4: Parameters of UHM in MIKE RR model of Ke Go catchment Parameters Base flow: BF Loss Curve number: CN Initial AMC: IAMC Time lag: TL Indicators m 3 /s - mm Hours Value 10.3 60 2 0.5 Figure 5 - 6: Observation and simulation of hourly discharge of Ke Go reservoir from 2 Oct to 6 Oct – 2010 for calibration model 48 Table 5 - 5: Different in peaks of observed and simulated discharge for calibration mode at Ke Go reservoir Calibration Q Max (m 3 /s) ΔQ (m 3 /s) Time (hr) Δt (hrs) R 2 Observed 2020 -140 (7%) 22 AM 4/10/2010 0 0.75 Simulated 1880 22 AM 4/10/2010 Verification The discharges of Ke Go reservoir from 14th to 19th of October, 2010 were used to verify parameters of model, which are determined in the calibration step. The calculated and simulated hydrograph at Ke Go reservoir during flood event are illustrated in Figure 5 - 7. The different of peaks of observed and simulated discharge in verification model at Ke Go reservoir is shown in Table 5 - 6. Figure 5 - 7: Observed and simulated hourly discharge of Ke Go reservoir from 14 Oct to 19 Oct – 2010 – Verification Table 5 - 6: Different in peaks of observed and simulated discharge in verification at Ke Go reservoir 49 Calibration Peak of discharge (m 3 /s) Different: ΔQ (m 3 /s) Time of peak hrs Different Δt (hrs) R 2 Observed 3980 190 (5%) 16 AM 16/10/2010 0 0.76 Simulated 4170 16 AM 16/10/2010 In calibration and verification model step, the calculated and measured data are fit well as shown Figure 5 – 6 and Figure 5 – 7. Following Table 5 - 5 and Table 5 - 6, the different of measured and calculated data in the peak flow is trivial, 7% for calibration model and 5% for verification model. The time of peak in the measurement and calculation are coincident. The coefficient of efficient (NASH) of the calibration and verification step is higher than 0.7. Therefore, the determined parameters are acceptable for Ke Go reservoir catchment and surrounding area to simulate discharge. In addition, discharge flow into Ke Go reservoir in October 16 th 2013 is run with the verified rainfall runoff model for and the results are shown below. Figure 5 - 8: Flood at Ke Go reservoir in 16 October 2013 In general, the results of the simulation on the Ke Go reservoir are quite good. Although some time the observed and calculated data are not fit and water balance 50 error is still high accounting for 28.1%. The results might be affected by local rainfall, hence the pre-process of input data for Rainfall runoff model is necessary. Determining parameters of Rainfall runoff model for sub-basin of Rao Cai river basin. In the downstream area of Ke Go reservoir and its tributaries, there is no hydrological station for discharge data. Therefore, the lateral inflow corresponding to actual flood, design hydrograph was calculated based on the rainfall data UHM model, which rainfall data and parameters obtain from typical sub-basin. In the study area, the Ke Go catchment has similar characteristics with the sub-basins in term of about soil condition, plant cover, land use, geology and meteorology. Thus, the parameters of UHM-SCS model for Ke Go catchment are used to determining parameter for sub-basins in Rai Cao River by adjustment. Besides, the land use condition of each sub-basin is considered. Table 5 - 7: Parameters of UHM - SCS for Rao Cai’s sub-catchments No Sub-catchment Area (km 2 ) Parameter (CN) BF (m 3 /s) CN IAMC (mm) TL (hrs) 1. Sub 2 65.93 3 74 2 1 2. Sub 3 70.58 3 74 3 1 3. Sub 4 57.80 5 74 2 1 4. Sub 5 255.73 10.2 74 3 0.5 5. Sub 6 49.39 3 91 2 1 6. Sub 7 80.55 3 74 2 1 7. Sub 8 39.66 3 74 2 1 8. Sub 9 58.55 3 74 2 1 The discharge of sub-basin is calculated by MIKE – UHM model, are connected to main river based on their location of sub-basins. 5.2.2. Flood modeling 5.2.2.1. Setup MIKE 11 HD model To simulate flooding by MIKE 11 HD model, the following data are required: river network including profiles (shape, roughness, structure, etc.); Digital elevation 51 model (DEM) to determine floodplain; Hydraulic boundary and initial conditions; Measured water level and inflow of sub-catchment. River network Hydraulic diagram is established based on the river network documents containing Gia Hoi, the Rao Cai river behind Ke Go dam. Figure 5 - 9: Hydraulic calculation network in downstream of Ke Go reservoir Topography Cross sections Whole network has 70 cross sections. Cross-sections are specified by a number of x-z coordinates where x is the transverse distance from a fixed point (often left bank top) and z is the corresponding bed elevation. All of cross sections were obtained from Institute of Civil Engineering of Thuy Loi University which measured data in 2011. The missing data were interpolated from measured value river bed, DEM and topography in 10,000 scales. Floodplain Land elevation in the downstream range from 2.5 to 10 m, so when big flood and heavy rainfall, the river can’t stand for big discharge. This issues cause overland along the river. In floodplain during flood event, 1D model is not adequate to 52 describe the spatial variable of overland, water depth, velocity, etc. However, to solve these limitations of the MIKE 11 model, the floodplain via structure or sub- branch can be setup. In this study, floodplain was setup via weir side with storage capacity, the crest of weir was determined by overflow threshold. In downstream have 8 sub- catchments, and each sub-catchment has different overflow threshold. It depends on their topography. Each floodplain is determined by storage capacity, length of weir and overflow threshold. Figure 5 - 10: Storage capacity of floodplain in downstream Determining boundary conditions of the model - Upstream boundary are release of Ke Go reservoir corresponding to actual flood in 2010 and different design floods obtain from the Establishing Ke Go reservoir operation regular in 2012. - Downstream boundaries are hourly water level at the Thach Dong hydrological station (Cua Sot) and the Cam Nhuong hydrological station (Cua Nhuong) - Lateral inflows are calculated by rainfall-runoff models in above part. The downstream of Ke Go reservoir is divided into 8 sub-catchments, which are linked along the major river systems and tributaries. 0 2000 4000 6000 8000 10000 12000 14000 0 1 2 3 4 5 W ( h a) Z (m) Z ~ W Sub9 Sub8 Sub2 Sub3 Sub4 Sub5 Sub6 Sub7 53 Table 5 - 8. Runoff link of sub-catchments into river network in MIKE 11 model No Sub- catchment Area (km 2 ) Description Link 1. Sub 2 65.93 Downstream Ke Go Reservoir to junction of Ngan Mo Raocai (0m – 7000m) 2. Sub 3 70.58 Ngan Mo to Cau Ho on Gia Hoi river Giahoi (0m-14700m) 3. Sub 4 57.80 Quen Catchment (Thuong Tuy Cat) Giahoi (28500m) 4. Sub 5 255.73 Rac Catchment Giahoi (31231m) 5. Sub 6 49.39 Ngan Mo to Cau Phuon Ngan Mo river Raocai (7000m-24000m) 6. Sub 7 80.55 Bang river Raocai (24000m-36200m) 7. Sub 8 39.66 Thach Dong estuary Raocai (36200m-43210m) 8. Sub 9 58.55 Cam Cua Nhuong Giahoi (31231m-33210m) - Checking boundary: At the downstream area of Ke Go reservoir, during flooding season, the Ke Go irrigation company usually arranges water level measurement stations in the downstream area to monitoring the process of releasing flood through operating spillways to ensure minimizing the downstream inundation. The positions of the water level measurement at the downstream area of Ke Go reservoir: Vang Vang trough, Cau Ngan Mo, Rao Cai trough, Rao Na, Cau Phu, Cau Hoi... In the thesis, two measurement stations at the Cau Phu and the Hoi bridge are used as checking stations. The collected data of the water level at the downstream area of Ke Go reservoir at the above locations are not continuous. Therefore, the simulated data will be extracted corresponding to available measurement data to analysis. Table 5 - 9: Monitoring points for the calibrating and verifying hydraulic model No Measuring point River Chainage (m) 1 Cau Phu Rao Cai Raocai 23421 2 Cau Ho Gia Hoi Giahoi 13684 54 5.2.2.2 Calibration and verification of the MIKE 11 HD model Calibration The most sensitive parameter of the MIKE 11 HD model is roughness coefficient (n or M). It is a very important parameter and greatly influenced on the study results. In this thesis, the roughness coefficient was estimated based on the topographic documents associated with field investigation to calculate. The roughness coefficient is in the range from 0.022 to 0.031 with the natural river (see Appendix 3). The results of model adjustment are presented under in form of charts and the maximum value between the results of calculated and measured at the checking locations. The results of calculated and measured water level at the checking stations are presented in the Figure 5 - 11, Figure 5 - 12. (Blue line is calculated values and black points are measured values). Figure 5 - 11: Calculated and measured water level at Cau Phu (2 nd to 6 th October, 2010) 55 Figure 5 - 12: Calculated and measured water level at Cau Ho (2 nd to 6 th October, 2010) Table 5 - 10: Results of flood simulation form 2 nd Oct to 6 th Oct, 2010 for calibration of MIKE 11 HD model No Checking location River HMax (m) Difference (m) Calculated Measured 1 Cau Phu Rao Cai 2.6 2.73 0.13 2 Cau Ho Gia Hoi 2.39 2.40 0.01 The results of the simulation of the Rao Cai river and Gia Hoi river: corresponding to available measurement data: + The line of the water level process between the calculation and measurement is relatively similar in terms of fluctuation phase and the peak value; the line of process shown the fluctuation of the tidal effect. + The difference of the higher water level is not significant with the fluctuated value ranging from 0.01 to 0.13 m. The difference of peak flow at the checking stations is in admissible range. Verification model After calibration of model, the model was run for flood from12 October to 18 October – 2010 to verify 56 The results of model testing is also presented on the chart about the process of the properly measured water level at the checking stations in the network, combining with corresponding testing target. The results of the calculated and measured data at the checking stations are shown in the Figures 5 – 13. Figure 5 - 13: Calculated water level and measured water level at Cau Phu(12 Oct to 18 Oct- 2010) Figure 5 - 14: Calculated water level and measured water level at Hoi Bridge (12 Oct to 18 Oct- 2010) 57 The different of the peak flow shown in the following table 5 - 11: Table 5 - 11: Results of flood simulation from 12 Oct to 18 Oct- 2010 No. Checking location River HMax (m) Difference (m) Calculation Measurement 1 Cau Phu Rao Cai 3.25 3.10 0.14 2 Cau Ho Gia Hoi 2.90 3.02 0.12 - The simulation results were quite consistent with the real data in term of the flood events occurred in the recently. That is, the parameters used in the hydraulic model are acceptable flood simulation of the system river of Rao Cai river basin. The model might be used to conduct the hydraulic calculations under the different scenarios for the study area. 5.2.2.3. Simulations of various return periods of flood events To calculate design flood in the downstream of the Ke Go reservoirs, this thesis assumed that the frequency of design flood of Ke Go reservoir and frequency of rainfall are the same. Water level at the downstream will use actual data of 2 hydrological stations (Cam Nhuong and Thach Dong) from 16 Oct to 18 Oct – 2010. The simulation of flood flow using calibrated MIKE 11 HD mentioned above, are done. The period of flood simulation is five days, assuming that flood occur from 1 Oct to 5 Oct - 2010. The simulated water level at Cau Phu and Cau Hoi station corresponding to flood periods are in Table 5 - 12. Table 5 - 12: Maximum water level corresponding to design and checking flood of Ke Go reservoir (Unit: m) Station Chainge (m) Design flood Checking flood Cau Ho Giahoi 2001.42 3.52 4.15 Cau Phu Raocai 4988.87 3.73 4.25 58 Based on results of maximum water level at Cau Ho and Cau Phu, the water level in Cau Phu is higher than Cau Ho. The difference of the maximum water level between design and checking flood about 0.5 m. 5.2.3. Flood hazard maps There are many methods to carry out flood risk assessment based on different applied modeling. In this research, the author selected MIKE 11 GIS is a tool for the spatial presentation and analysis of one-dimensional (1D) flood model results for use in the flood risk assessment process. The MIKE 11 GIS system integrated the MIKE 11 with the spatial analysis capabilities of the ArcView Geographic Information System (GIS) gives us an advanced tool to assess flood risk. The outputs were developed from using MIKE 11 GIS are important inputs for a range of floodplain management undertakings including flood risk assessment, flood control, flood forecasting, floodplain preservation and restoration, etc., The software requirements for developing inundation maps based on using MIKE 11GIS include MIKE 11 HD and ArcGIS. Alternately, in the research, some steps in building up flood risk maps, ArcGIS is evolved to visualize more clearly the pictures of flood. To assess hazard of flooding, a hazard level scale is defined on the basis of the flooding depth. A grid including four levels (low, medium, high, and very high) of flooding depth would be created. Areas inundated under 3m or deeper water will be recognized as very high hazard area. An area with 1.5-3m, 0.5-1.5m, and lower 0.5m deep of flooding depth will get high, medium or low hazard area, respectively. As a result, thesis has four different hazard levels defined as in the follow table: Table 5 - 13. Designed flooding hazard level scale for the downstream of the Ke Go reservoir No Flood depth (m) Hazard zone 1 < 0.5 Low (1) 2 0.5 – 1.5 Medium (2) 3 1.5 – 3 High (3) 59 No Flood depth (m) Hazard zone 4 > 3 Very high (4) Flood hazard threshold are divided 4 levels, which were depended on impact thresholds of flooding depth to person. The flooding depth is less than 0.5 m correspondence to low hazard. This value approximates the height of leg of adult people so when the flood comes they adapt with them and can help people around them especially the child and the old. The flooding depth is less than 1.5 m equivalent to medium hazard level. This value approximates the height of body of adult people. Beside in the center part of Vietnam, resident often have flooding mitigation measure by building small flat far from the land around 1.5 m to avoid flood or hide livestock (big, chicken, dog, cat). And for flooding depth is 3 m, this value is as the same the height of first floor of building level 2 or 3 (such as house publics school, clinic, house’s resident). When flooding, people often hide in building level 2 or 3, if flood level can reach this value people can move to upstairs. a) Low hazard level b) Medium hazard level c) High hazard level d) Very high hazard level ~ 1.5 m 60 Figure 5 - 15. Some typical picture to determine flood hazard threshold 5.2.3.1. Hazard map of Flood event in 2010 Figure 5 - 16. Flood hazard map for historical flood event in 2010 With differences of water depth, inundation map is classified into four groups: very high, high, medium, and low. Most of inundated areas have water depth lower 2 m and 3 m. These groups occupy three-fourths of whole inundated extent. High and very high hazard area in the term of inundation distribute along main river (Rao Cai) and a part of Bang river. Table 5 - 14. Flood hazard areas for flood event in 2010 Hazard zone Design flood Area (km 2 ) Percentage (%) Low 24.8 6.6% Medium 94.8 25.1% 61 High 218.2 57.8% Very high 39.9 10.6% 5.2.3.2. Hazard map of 0.5% and 0.1% flood Figure 5 - 17. Flood hazard map of 0.5% design flood event 62 Figure 5 - 18. Flood hazard map of 0.1% design flood event Table 5 - 15. Flood hazard areas corresponding to design and checking flood Hazard zone Design flood Checking flood Area (km 2 ) Percentage (%) Area (km 2 ) Percentage (%) Low 31.3 7.5% 30.6 6.7% Medium 96.9 23.3% 97.3 21.3% High 232.4 56.0% 257.3 56.5% Very high 54.6 13.2% 70.5 15.5% 5.3. Flood vulnerability There are various methods to determine vulnerability. One method is to use a vulnerability scale which is defined via the consideration of the defined hazard scenarios and the information available regarding the damage to the population and environment. In this study, vulnerability map was built on the basis of population density map. 63 Figure 5 - 19: Frequency distribution of population of study area Figure 5 - 20: Frequency distribution of population density of study area Following Figure 5 - 19 and Figure 5 - 20, population number of communes in Rao Cai river basin range from 1905 to 9875 people. The difference about population number between communes is not much with population distribution in Rao Cao river basin is estimated based on quite frequency have normal distribution type. For population density is quite different because this value depends on area and population of commune. It normally middle and mountain zone have large area but a few populations, is opposite in delta or city center. So criteria to divide vulnerability levels to ensure area for different level are equal. Table 5 - 16. Criteria of vulnerability map derived from population density for the downstream of the Ke Go reservoir 64 No Population density (Person/km 2 ) Vulnerability level 1 < 100 Low (1) 2 100 – 500 Medium (2) 3 500 – 1500 High (3) 4 > 1500 Very high (4) Figure 5 - 21: Vulnerability map in Rao Cai river basin Results determine vulnerability zone, high zone located in Ha Tinh province and some medium zone located in around vulnerability. A few areas are very high zone, because there is high density population. Based on vulnerability map, the population often lives near in the city center and coastal area. 65 5.4. Flood risk in downstream area of the Ke Go reservoir To assess flood risk in the downstream of Ke Go reservoir, a flood risk matrix is derived from flood hazard level and flood vulnerability level. As a result of hazard and vulnerability assessment, flood risk scale will be created by four levels of hazard and four levels of vulnerability (low, medium, high and very high). The resulting risk level scale consists of 4x4 = 16 cells and includes four different levels of risk: low, medium, high and very high (see the follow figure). Figure 5 - 22. Designed risk level for the downstream of the Ke Go reservoir H az ar d l ev el Low (1) 1 2 3 4 Medium (2) 2 4 6 8 High (3) 3 6 9 12 Very high (4) 4 8 12 16 Low (1) Medium (2) High (3) Very high (4) Vulnerability level Table 5 - 17. Criteria of vulnerability map derived from population density for the downstream of the Ke Go reservoir No Criteria Risk zone 1 1 – 3 Low 2 4 – 6 Medium 3 8 - 9 High 4 12 - 16 Very high Flood risk scale is the basis to develop flood risk map. Flood risk map is the combination of hazard map and vulnerability map which can describe comprehensively risky statement of study area, where was recognized as a flood- heavily-impacted area. Table 5 - 18. Flood risk map for the downstream area of Ke Go catchment in flood in October, 2010 Risk zone Area (km 2 ) Percentage (%) 66 Risk zone Area (km 2 ) Percentage (%) Low 52.7 14.0% Medium 164.5 43.6% High 107.6 28.5% Very high 52.8 14.0% Total 377.6 100% The below map shows the potential risk which was caused by 2010 flood event. In general, risk areas are nearly medium and high level, with above 70% affected area while low risk area and high risk area were equal with 14% affected area. The medium risk area reached 164.5 Km 2 approximately 43.6% (the highest), where located almost along Rao Cai river near Ha Tinh city and surrounding area. Figure 5 - 23. Flood risk map for the downstream area of Ke Go river basin of flood in October, 2010 67 Flood risk map for design flood and checking flood with 200 year and 1000 year return period was built by intersecting hazard map and population density map to give an assumption how is risky if 0.1% and 0.5% flood occurs at the moment. It does not mean that this map will depict exactly risk situation in future due to changing of population distribute land use, demographic density and other aspects as well. Figure 5 - 24. Flood risk map for 0.5% design flood 68 Figure 5 - 25. Flood risk map for 0.1% checking flood Table 5 - 19. Statistic table of flood risk of different floods Hazard zone Design flood Checking flood Area (km 2 ) Percentage (%) Area (km 2 ) Percentage (%) Low 70.0 16.9% 64.4 14.1% Medium 164.4 39.6% 178.9 39.3% High 112.3 27.0% 123.5 27.1% Very high 68.5 16.5% 88.9 19.5% Total 415.1 455.7 Following Table 5 - 19 shows risk areas, at four different levels correspond to design and checking flood of Ke Go reservoir. Overall, the total risk area of checking flood case is higher than design flood case because flow of the former is higher than last one. Medium and high risk areas are nearly total affected area with above 70%. In both case, percentage of affected medium and high area were equal with 39 % and 27% respectively. Only having change in low area and very high area. 69 The statistical results of table 5-15, figure 5-17 and figure 5-18 show that the flooding depth in downstream area of Ke Go reservoir is quite severe. Because this is an important area, there is a need to prepare some effective measures to minimize damages to the Ha Tinh city and surrounding area. To flood risk management, we can use structural and non-structural measures. The structural measures can design dykes or build levees along Rao Cai and Gia Hoi rivers. At the present, a number of measures are applied all over the world to ensure safety for particularly important areas. The nonstructural measures are structure elevation or structure relocation by moving structure in low land to high land. Beside, awareness increase of resident about flood control and mitigation solution is very important in flood risk management. 70 CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS 6.1. Conclusions The Rao Cai is one of the largest rivers in Ha Tinh province. In recent years, floods have been increasing not only frequency but also intensity and affect to all aspects of lives. Climate change, improper management of natural resources, inadequate awareness on flood risk, etc., response for the question ―Why do flood become increasingly serious?‖. To minimize negative impacts of flood on local people, it is necessary to make an efficient flood risk management. Flood risk assessment based on flood modeling has an important contribution to this progress. To assess flood risk, in this study, the MIKE package including MIKE UHM, MIKE 11 HD and MIKE 11 GIS, was used to simulate the previous flood events in 2010 and other floods as 0.5% and 0.1%. The main objective of this study is to develop inundation maps corresponding to the floods as mention above. That is the foundation to assess the flood risk by overlapping vulnerability and hazard layer. To assess risk of the flood events, the following conclusions could be depicted as elements of research procedure: + Main reasons and causes of flood in study area are affected by heavy rainfall with high intensity and more frequency. Those are impact of climate change. + Digital Elevation Model (DEM), which represents the floodplain’s surface, is the input information for simulating and estimating the flooding extend and depth. The DEM in the research was derived from interpolation of elevation spots and contour lines of different spatial data such as, contour maps with variable scales, transportation system maps, dyke system maps, and hydrological maps. The DEM was built from topography with scale of 1:10,000. + For the, in order to simulate the lateral inflow of sub-catchments in downstream of Ke Go reservoir, the regionalization method was applied by using the model parameters of Ke Go catchment; 71 + The combination of flooding depth (hazard factor) and population density (vulnerability factor) with the weighing factors for both of them, flood risk assessment was done and mapped. The level of hazard and risk were determined for each community in Cam Xuyen, Thach Ha and Ha Tinh city. These maps can be used for flood management. 6.2. Recommendations Due to limited time and knowledge, this thesis have focused on simple hazard and vulnerability such as flooding depth and population density, to have valuable result we have to considering to other aspect such as: velocity, during flooded (hazard factor) and vulnerability about agriculture, infrastructure (road, railway). Besides to improve result flood simulation, inundation map need good data, in study area are almost data from local supply (water level at upstream, downstream, release flow of Ke Go reservoir, which can be reedited for an another reason (political, ) From this study, the following issues are strongly recommended: - Technical issues: Before using the recorded data of a station, it is necessary to firstly check the data for continuity and consistency. - Research issues: flood risk assessment is difficult but necessary for flood risk management and the results of hazard and risk assessment should be represented to the local and regional decision makers. Recommendations for future research: The results of flood modeling and flood characteristics can be improved if better resolution and quality DEM is used for flood simulation. Research was based on the topography in 10,000 scales. The usage of better resolution and quality DEM and considering infrastructure such as road, railway, and building or using satellite image and classify image of past flood to improve quality of hydrodynamic model are topic for future research. Besides, it is undeniable that awareness of resident about flood it is very important in flood control and mitigation and it is effect to result flood risk. 72 The study of flood hazard should be more considering with others scenarios to obtain the higher results of risk assessment for the region in order to quantify the expected damage for different return periods. Recommendations for the Local Administrative Authorities and local communities: It should be improved the risk perception of local community regarding to flood process and its hazardous effect on population and their property. New hazard maps should be represented to the population and they should be informed about probable results. According to of European Union, the risk maps should be rebuilt after 5 years. It should be prohibited to use dikes for cattle pasture. Dikes should be built up to protect population along rivers Construction of small channels and rising vines with ground will decrease the water level on cultivated land and protect it from inundation. Raising the awareness of people about flood, hazard and mitigation ways 73 REFERENCES Alkema, D. (2007). Simulating floods: on the application of a 2D hydraulic model for flood hazard and risk assessment. ITC Dissertation; 147, ITC: 198. Asian Disaster Reduction Center (ADRC). (2005). Total disaster risk management- Good practices 2005. Baas, S., Ramasamy, S., DePryck, J. D., Battista, F. (2008). Disaster Risk Management Systems Analysis. Environment, Climate Change and Bioenergy Division, FAO, Rome, Italy, 68 pp. Bell, F. (1999). Geological hazards: their assessment, avoidance and mitigation. London: E & FN SPON. Bruijn, K. M. D., Klijn, F. (2005). Resilient flood risk management strategies. Delft University of Technology WL|Delft Hydraulics. Central committee for flood and storm control (CCFSC). (2013). Statistics of flood and tropical in Vietnam in 2013. Commonwealth Scientific and Industrial organisation (CSIRO). (2000). Annual report. In: Jennifer North , k. R. (ed.) Campbell: CSIRO. Chien, N. (2010). Dam Safety in major flood conditions, Thuy Loi University Chow, V. T., Maidment, D. R., Larry W. Mays (1964), Applied Hydrology, McGraw-Hill, Delft. (2014). Deltares enabling Delta life. Delft: Delft Hydraulics. DHI. (2011). MIKE 11 User Guide. Denmark: DHI. Federal emergency management agency (FEMA). (2010). Flood Hazard Mapping Web Site. Retrieved 3 2014, from from 74 Geohazards. (2009). Applied earth Sciences:Geo Hazards,Process Modelling Multi Hazard risk . ITC. Hawkins, R.H.; Jiang, R.; Woodward, D.E.; Hjelmfelt, A.T.; Van Mullem, J.A. (2002). "Runoff Curve Number Method: Examination of the Initial Abstraction Ratio". Proceedings of the Second Federal Interagency Hydrologic Modeling Conference, Las Vegas, Nevada (U.S. Geological Survey). doi:10.1111/j.1752-1688.2006.tb04481.x. Retrieved 24 November 2013 Ha Tinh. (2014, 6). Ha Tinh genneral infomation. Retrieved 2014, from HCFSCS, 2011, Annual damage report of Ha Tinh province in 2010, Ha Tinh. Hec-Ras. (2010). HEC-RAS v4.1 User Manual.. US Army Corps of Engnieers, USA Hieu, T. H. (2012). Ha Tinh city "flooded" after the rainfall. Retrieved July 2014, from con-mua-keo-dai-48428.html Hoai. A. T. (2009). Biến đổi khí hậu ở Hà Tĩnh - Thực trạng và giải pháp. Retrieved January 2014, from khi-hau-o-ha-tinh-thuc-trang-va-giai-phap/54003 Hoai, N. T. (2013). 2013: 264 people dead and missing due to floods. Retrieved January 2014, from nguoi-chet-va-mat-tich-do-bao-lu-c17a141981.html International Strategy for Disaster Reduction (ISDR) (2004). Living with Risk - A global review of disaster reduction initiatives. International Strategy for Disaster Reduction. New York, UN/ISDR. 2: 429. Ke Go Irrigation Company (KGIC) . (2012). Operation process of Ke Go reservoir, 2012 75 Kumpulainen, S. (2006). "Vulnerability concepts in hazard and risk assessment. Natural and technological hazards and risks affecting the spatial development of European regions." Geological Survey of Finland Retrieved Special Paper, from Klijn, F., M. van Buuren, et al. (2009). "Flood-risk Management Strategies for an Uncertain Future: Living with Rhine River Floods in The Netherlands?" AMBIO: A Journal of the Human Environment 33(3): 141-147. Merz, et al. (2007). Flood Risk Mapping At The Local Scale: Concepts and Challenges. Flood Risk Management in Europe. Springer Netherlands, 25: 231-251. Plate, E. J. (2002). Flood risk and flood management. Journal of Hydrology 267(1- 2): 2-11. Pender, G. and S. Néelz (2007). "Use of computer models of flood inundation to facilitate communication in flood risk management." Environmental Hazards 7(2): 106-114. Thai, N. C., Lan, P. T. H., Huan. T. N. (2011). Study effects of dam breaking at Ke Go reservoir – Ha Tinh in downstream area. Water Resources and Environment Journal, Thuy loi University, Vol. 11, pp 18 - 25. Thong, L. Q., (2014). Study on effect of wave to Ke Go emergency dam. Master thesis of Thuyloi university, Vietnam. Twigg, J. (2004), Disaster risk reduction: Mitigation and Preparedness in development and emergency programming. Humanitarian Practice Network. London Kingdom. 76 Tamar Tsamalashvili. (2010). Flood risk assessment and mitigation measure for Rioni River, University of Twente; Tu, V. T . (2009). Flood inundation, damage and risk assessment in Hoang Long basin, Vietnam, AIT Dissertation, Asian Institute of Technology United States Department of Agriculture (USDA) (1986). Urban hydrology for small watersheds. Technical Release 55 (TR-55) (Second Edition ed.). Natural Resources Conservation Service, Conservation Engineering Division. USGS. (2013). Satellite image of Ha Tinh area USGS. (2011). DEM of Ha Tinh area, Vi, D. N. (2013). Nhìn lại 8 trận lụt kinh hoàng tại Việt Nam. Retrieved January 2014 from nam.2.456814.htm Viet, T. Q. (2009). Flood risk assessment for the Thach Han river basin, Quang Tri province, Vietnam. Cologne University of applied science. Vietnam Water Resources Assistance Project (VWRAP). (2003). Feasibility study for modernization of Ke Go project, Ha Noi, Haskoning. Vnexpress. (2010). Miền Trung vật lộn với lũ dữ nhất 10 năm. Retrieved January 2014 from nhat-10-nam-2896392.html Vnexpress. (2013). The flooding in Oct, 2013 in central Vietnam. Retrieved January 2014 from https://www.google.com/maps/d/u/0/viewer?mid=zGR_qj798D2c.kLUgSscZ XoJk&msa=0&iwloc=0004e8ea5f790918f950c 77 APPENDIX Appendix 1: Frequency curve of maximum rainfall during 1 day of stations Hyetograph of design rainfall of Ha Tinh stations corresponding to 1000 years return period Hyetograph of design rainfall of Ha Tinh stations corresponding to 200 years return period 0 100 200 300 400 500 600 700 800 900 1000 1st 2nd 3th R ai n fa ll (m m ) Time (day) Actual Design 0 100 200 300 400 500 600 700 800 1st 2nd 3th R ai n fa ll (m m ) Time (day) Actual Design 78 Hyetograph of design rainfall of Ky Anh stations corresponding to 1000 years return period Hyetograph of design rainfall of Ky Anh stations corresponding to 200 years return period 0 100 200 300 400 500 600 700 800 1st 2nd 3th R ai n fa ll (m m ) Time (day) Actual Design 0 100 200 300 400 500 600 700 1st 2nd 3th R ai n fa ll (m m ) Time (day) Actual Design 79 Hyetograph of design rainfall of Huong Khe stations corresponding to 1000 years return period Hyetograph of design rainfall of Huong Khe stations corresponding to 200 years return period 0 100 200 300 400 500 600 700 1st 2nd 3th R ai n fa ll (m m ) Time (day) Actual Design 0 100 200 300 400 500 600 1st 2nd 3th R ai n fa ll (m m ) Time (day) Actual Design 80 FFC 2008 © Nghiem Tien Lam 30 130 230 330 430 530 630 730 830 930 1030 1130 0.01 0.1 1 10 20 30 40 50 60 70 80 90 99 99.9 99.99 FREQUENCY CURVE OF MAXIMUM RAINFALL DURING 1 DAY - HA TINH STATION R a in fa ll, X (m m ) Frequency, P(%) Maximum rainfall during 1 day TB=304.81, Cv=0.41, Cs=0.86 Pearson Type III Distribution TB=304.81, Cv=0.43, Cs=0.89 © FFC 2008 FFC 2008 © Nghiem Tien Lam 20 120 220 320 420 520 620 720 820 920 0.01 0.1 1 10 20 30 40 50 60 70 80 90 99 99.9 99.99 FREQUENCY CURVE OF MAXIMUM RAINFALL DURING 1 DAY OF KY ANH R a in fa ll (m m ) Frequency, P(%) Ky Anh_Maximum rainfall during 1 day TB=292.62, Cv=0.36, Cs=0.60 Frequency curve of maximum rainfall during 1 day of Ky Anh TB=292.62, Cv=0.36, Cs=0.60 © FFC 2008 FFC 2008 © Nghiem Tien Lam 40 90 140 190 240 290 340 390 440 490 540 590 640 690 740 790 0.01 0.1 1 10 20 30 40 50 60 70 80 90 99 99.9 99.99 FREQUENCY CURVE OF MAXIMUM RAINFALL DURING 1 DAY OF HUONG KHE R a in fa ll (m m ) Frequency, P(%) maximum rainfall during 1 day of Huong Khe TB=246.72, Cv=0.37, Cs=0.62 frequency curve TB=246.72, Cv=0.37, Cs=0.82 © FFC 2008 81 Appendix 2: Roughness coefficient River Chainage (m) Manning (n) Raocai 0 0.03 Raocai 1382 0.031 Raocai 2676 0.031 Raocai 3593 0.029 Raocai 4558 0.029 Raocai 4638 0.029 Raocai 5588 0.029 Raocai 6576 0.029 Raocai 7951 0.029 Raocai 9651 0.024 Raocai 11438 0.024 Raocai 15969 0.024 Raocai 26702 0.024 Raocai 36140 0.024 Raocai 41238 0.024 Raocai 43098 0.024 Raocai 43210 0.024 GiaHoi 0 0.028 GiaHoi 1010 0.028 GiaHoi 3006 0.028 GiaHoi 3061 0.028 GiaHoi 5429 0.028 GiaHoi 7888 0.024 GiaHoi 17624 0.024 GiaHoi 20869 0.024 GiaHoi 23294 0.024 GiaHoi 27864 0.022 GiaHoi 32229 0.022 GiaHoi 33131 0.022 GiaHoi 33210 0.022