Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.Nghiên cứu phương pháp cảnh báo lũ quét cho lưu vực nhỏ miển núi và áp dụng thử nghiệm cho 2 lưu vực Nậm Ly và Nà Nhùng, tỉnh Hà Giang.MINISTRY OF EDUCATION AND TRAINING MINISTRY OF AGRICULTURE AND RURAL DEVELOPMENT THUYLOI UNIVERSITY NGUYEN THE TOAN FLASH FLOOD WARNING METHOD FOR SMALL MOUNTAINOUS BASINS – PILOTING ON 2 BASINS, NAME.
MINISTRY OF EDUCATION AND TRAINING MINISTRY OF AGRICULTURE AND RURAL DEVELOPMENT THUYLOI UNIVERSITY NGUYEN THE TOAN FLASH FLOOD WARNING METHOD FOR SMALL MOUNTAINOUS BASINS – PILOTING ON BASINS, NAMELY: NAM LY AND NA NHUNG IN HA GIANG PROVINCE Major: Hydrology Code: 9440224 SUMMARY OF DOCTORAL THESIS IN TECHNICS HANOI, 2023 The thesis is at Thuyloi University Scientific supervisors: Assoc.Prof., Dr Tran Kim Chau Assoc.Prof., Dr Nguyen Ba Quy Referee 1: Prof., Dr Vu Minh Cat, Vietnam Union of Science and technology associations Referee 2: Prof., Dr Huynh Thi Lan Huong, Hanoi University of natural Resources and Environment Referee 3: Dr Trinh Quang Toan, Vietnam academy for water resources The thesis will be examined by Examination Board of At ……………………………………………………………………………… - The thesis can be found at: - National Librar - Thuyloi University Library INTRODUCTION The necessity of the study Flash flood is a type of natural disaster which is increasing in almost all mountainous areas and river basins in the world, especially in tropical and subtropical basins In Vietnam, according to the Vietnam Disaster Management Authority, there are about 10-15 flash floods per year on average Flash flood occurs frequently in four areas in Vietnam, namely Northern mountainous region, the Central region, the Central Highlands, and the Southeast region Most flash floods and landslides occur in remote, sparsely populated areas; however, there are also flash floods that have great destructive power and cause great damage to people's lives and property In recent years, the magnitude, frequency, and complexity of flash floods tend to increase and become more serious In the past 20 years, according to the Vietnam Disaster Management Authority, there are over 300 flash floods in the Northern mountainous provinces, with increasing scale and scope, causing heavy losses to people, property, and infrastructure From 2005 to now, there have been a number of significantly heavy flash floods causing great damage to people's lives and properties, such as in Lai Chau (2012, 2018), Yen Bai (2005, 2011), Lao Cai (2008), Bac Can (2009), Nghe An (2007, 2016), Dak Lak (2001), Kon Tum (2009), Hoa Binh (2011), and Ha Giang (2012 - 2020) Human loss caused by flash floods, far higher than that of other natural disasters, such as storms and floods, mainly occurs in far-reaching residential areas where the majority of people are ethnic minorities This reality raises necessary and urgent need to have early warning of flash flood to minimize its harmful effects by all means, creating a safer environment for residential communities and providing them with information about potential flash floods for proactive prevention Previously, flash flood warning and prediction is still based on static forecasting models (with pre-built coping scenarios); however, these models lack feasibility because there is a delay in the actual measured rainfall data or the models are run manually and lack continuity and automation Therefore, real-time warning method should be studied, given the cumulative effects over time and more realistic warnings at different times during the flood season Real-time flood warning has been applied in many countries around the world as well as in some projects in Vietnam However, there have not been studies focusing on small areas, especially mountainous river basins where there are few hydro-meteorological gauge stations Also, the approach has not taken the local characteristics into account to improve the accuracy of warning and forecast Based on these realities, the author has chosen to research flash flood warning methods for small mountainous basins – piloting on basins, namely: Nam Ly and Na Nhung in Ha Giang province Study objective Based on research and practice to develop flash flood warning methods for small mountainous basins and experimentally applying the methods to the basins Nam Ly and Na Nhung in Ha Giang province Objects and scope of the study The scope of the thesis is in mountainous river basins, with case studies in the basins Nam Ly and Na Nhung in Ha Giang province Study objects of the thesis are the bankfull discharge, rainfall threshold of flash flood FFG, and real-time flash flood warning methods and tools Methodology • Field survey method to build the relationship between the discharge and water level and to determine the overflow water level based on signals • Statistical analysis method to build an empirical equation to calculate the bankfull discharge; • Mathematical modeling method: using hydrological mathematical model which is developed for the study basins to determine the flow value at the outlets of subbasins, which acts as a basis to calculate and determine rainfall threshold of flash flood FFG in real time; • Remote sensing and GIS methods to identify basin characteristics from map data, digital elevation models, satellite images, etc Scientific and practical significance Scientific significance: The study has established scientific and practical basis in building an empirical equation to calculate the bankfull discharge for mountainous river basins as well as successfully applied this equation for the river basins Nam Ly and Na Nhung in Ha Giang province The study has applied flash flood warning method for mountainous basins in the basins Nam Ly and Na Nhung by integrating a self-developed hydrological model with building an empirical equation to determine bankfull discharge which acts as a basis to determine rainfall threshold of flash flood FFG Consequently, forecast and warning of flash floods for the above basins can be improved and at the same time, the method and the model can be extended to other mountainous river basins Practical significance: The results of the study are valuable reference and a supporting tool for the management of local departments and agencies responsible for prevention and mitigation of natural disasters Structure of the thesis In addition to the introduction, conclusion, and appendices, the summary of the thesis is organized into chapters CHAPTER RESEARCH OVERVIEW OF THE SITUATION ON FLASH FLOOD AND ITS WARNING AND FORECAST 1.1 Studies in the world Flash flood warning studies in the world as well as in Vietnam focus on the following directions: a Building a risk map of flash floods based on the analysis of the forming factors Many studies approach flash flood warning by zoning areas vulnerable to flash flood based on flash flood potential index (FFPI) The FFPI quantitatively describes the flash flood risk of a basin based on its inherent static characteristics such as slope, surface cover, land use, and soil type and composition This method was introduced by Jeffrey Zogg and Kevin Deitsch (2013) [1] and applied by many authors such as Greg Smith (2003) [2], Brewster (2009) [3], and Kruzdlo 2010 [4] A simple method of determining FFPI is to use GIS technology to build database GIS consists of basic layers in raster form, namely: slope, vegetation cover/land use, soil, forest/vegetation density, and arithmetic average b Based on rainfall thresholds for flash flood warning Forestieri (2016) [5] used the TOPDM model to assess the rainfall threshold to determine flash flood risk in the Sicilian basin with fixed initial conditions The use of the mathematical model yielded certain results However, the disadvantage of this approach is that it did not take into account the time variation of basin conditions In addition to the method of determining the flow/rainfall threshold of flash flood which are mentioned above, another method is Critical Line (CL) method, which was used in Japan in the “Guidelines to identify rainfall threshold for sedimentrelated disaster warning and evacuation" by the Japanese Ministry of Construction in 2005 to determine flood and flash flood warning thresholds This method is used to predict the occurrence of flash floods using rainfall indices (intensity and total rainfall) drawn from data on rainfall intensity and total rainfall collected from flash floods occurring in the study area c Based on rainfall threshold of flash flood FFG (flash flood guidance) and bankfull discharge (Qbf) The study of flash flood forecast based on rainfall threshold FFG has been researched and developed by many authors Timothy L.s et al (1992) [6] determined the threshold of flash flood based on the rainfall threshold FFG; Konstantine et al (2006), (2013), (2018) [7] [8] [9] developed the model of Timothy L.s et al (1992) [6] and applied on the area from 2,000 - 4,000 km2 to calculate the risk of flash floods according to the intermittent rainfall frequency (1, 2, 4, 5, and hours) The model uses the intermittent rainfall threshold in subbasins, in which the occurrence of flash flood in a sub-basin can be identified if the rainfall in the intermittent rainfall exceeds the tolerance threshold of the subbasin d Building rainfall-triggered monitoring systems and early warning of flash floods Flash Flood Warning System Alert - World Meteorological Organization (WMO): The ALERT system was originally developed in the 1970s to apply in California-Nevada River The system consists of hydrometeorological and meteorological sensors which automatically report events, communication equipment, and computer hardware and software In its simplest form, the ALERT sensor transmits coded signals, usually via very high frequency (VHF) and ultra-high frequency (UHF) radios The flash flood warning system ALERT is recommended by WMO and has been successful in the US and some other countries Flash Flood Monitoring and Prediction system (FFMP) The US Flash Flood Monitoring and Prediction system (FFMP) (under National Water Management Plan - NWMP) is integrated with multi-sensors to detect, analyze, monitor rainfall, and provide quick warnings to support flash flood warnings The FFMP system is deployed throughout the United States The average basin rainfall, based on rainfall estimates from Doppler radar, is compared with FFG to determine the flash flood risk and severity 1.2 Studies in Vietnam La Thanh Ha (2009) [10] studied and developed maps of areas vulnerable to flash flood to serve the prevention of flash floods in Yen Bai province In 2009, the Institute of Meteorology, Hydrology and Environment implemented the project: "Investigation, survey, zoning, and warning of the risk of flash floods in mountainous areas of Vietnam - Phase 1" [11] The project has built maps of areas vulnerable to flash flood, scale 1:100,000 for 14 Northern mountainous provinces and maps for warning floods by rain that can cause flash floods in 37 river basins in the region The 14 provinces mentioned above are on a 1:5000 scale map To serve flash flood warning for small basins, this project used the Critical Line (CL) method 37 CL charts were built for 37 river basins The limitations of the project are that the map was on a small scale while flash flood locations were mostly at commune and village levels The map was mainly built on the basis of static data, not a dynamic map system integrated on GIS to serve flash flood warning In Vietnam, there have also been studies on flash flood warning using the FFG index, such as the project: “Investigating, surveying, and building a map of areas vulnerable to flash floods in the Central region, Central Highlands, and building a pilot system to warn the localities at high risk of flash floods for planning, directing, and operating the disaster prevention and adaptation to climate change”, which was implemented as Phase by the Institute of Meteorology, Hydrology and Climate Change from 2012 to 2017 [12] The limitation of this project is that it focused mainly on the Central and Central Highlands regions; as a result, the calculation and determination of Qbf in the North has not been considered, and in small mountainous basins without gauge stations, specific characteristics were not into consideration In Vietnam, a number of warning systems based on gauging equipment have been researched and piloted under the sponsorship of New Zealand in Ha Tinh province to warn flash floods in La river basin (Hoa Duyet, Son Diem, and Linh Cam areas) and in Ke Go basin The operating principle of the system is as follows: - Via stations measuring water level and rainfall using automatic equipment (at Chu Le, Hoa Duyet, Son Diem, Ke Go) and automatic radio communication devices, rainfall - flood information on rivers of Ngan Pho, Ngan Sau, and Ke Go are transmitted through a relay station to the Operations Center in Ha Tinh for processing and warning 1.3 Gaps in real-time flash flood forecast and warning research and research orientation of the thesis 1.3.1 Gaps in literature There are a large number of studies conducted by different authors on flash flood warning; however, there are still limitations in these studies Flash flood warning by flash flood risk zoning, which is conducted based on flash flood risk index FFPI, has the advantage of being simple and not requiring complicated data However, it is only an additional tool as its disadvantage is that it does not take into account immediate surface conditions (such as humidity and stream flows) Therefore, the results should only be considered as a valuable reference Many studies are based on rainfall thresholds to determine flash flood risk for basins with fixed initial conditions However, the disadvantage of this approach is that it does not take into account the change of basin conditions over time Therefore, the threshold value for rainfall is a fixed value, which is not reasonable as when the basin is dry, the water storage capacity is large, and a large amount of rain is needed to create flash flood However, when the basin is saturated with water, even light rain can cause flash flood The flash flood warning method based on the rainfall threshold FFG associated with the bankfull discharge Qbf is determined by direct measurement at the crosssection or by empirical formulas However, not all basins have sufficient survey data and observation stations The determination of Qbf for small areas is not well studied, and there is no integration of actual measurements to increase accuracy Flash flood warning systems have been researched and invested for construction in the world as well as in Vietnam However, the flash flood warning system has the disadvantage that the accuracy depends on the density of the observation network, which is difficult to meet with many countries, especially when flash floods often occur in mountainous areas where there are few observation stations and the accuracy of the data is not high 1.3.2 Research orientation of the thesis Given that flash flood forecasting and warning is currently facing many difficulties as well as there are still gaps in the literature on flash flood forecasting and warning, such as how to determine the flood threshold for mountainous river basins with few or no hydrometeorological gauge stations, it is necessary to build a real-time flash flood warning toolkit, which takes into consideration factors such as the current conditions of basin, to determine the rainfall threshold of flash flood FFG in any given time In this study, the author will approach by determining the flash flood warning method based on the FFG associated with the bankfull discharge, which is based on the empirical equation built for mountainous river basins The research process combines (1) field survey design to survey the values of bankfull discharge to build an empirical equation to determine the bankfull discharge for mountainous rivers; (2) the development of a rainfall-flow model combined with real-time rainfall data to assess the current conditions of the basin, from which the real-time FFG value is identified, and (3) the forecast rainfall to give appropriate flash flood warnings The study is applied to the Nam Ly and Na Nhung basins in Ha Giang province, Vietnam Approach diagram as shown in Figure 1.1 Figure 1.1 Research diagram CHAPTER DEVELOPMENT OF FLASH FLOOD WARNING METHODOLOGY FOR SMALL MOUNTAINOUS RIVER BASINS To develop a flash flood warning method for mountainous river basins, it is necessary to: (i) approach to determine the bankfull discharge for small mountainous basins with few gauge stations; (ii) build a hydrological mathematical model that allows to determine the initial conditions of the basin to determine the rainfall threshold of flash flood (FFG); and (iii) use forecast rainfall models combined with online rainfall data to warn flash floods in real time 2.1 Scientific basis for building empirical equation to calculate bankfull discharge for small mountainous basins From the overview study of methods to determine the bankfull discharge and from the natural conditions of the mountainous river basins in Vietnam, the author has decided to combine theory and experiment to determine the bankfull discharge Figure 2.1 below illustrates a summary of the method used in the study Figure 2.1 Diagram of the method to determine the bankfull discharge 2.1.1 Determining the threshold of overflow from the signals in the area According to the study of Bent (2013) and Blanton (2010) [13, 14], there are some signs to identify the overflow water level, listed as follows: (1) the level of the active floodplain, (2) the highest point in depositional features, (3) the level where riverbank slope changes, (4) the level where grain material changes, (5) the highest ground level of undercut on the riverbank, (6) the level where vegetation is changed (e.g., from a non-vegetated area to an area with vegetation) the model or the calculation results want to be integrated from the model into the tool, format conversion algorithms are required In addition, the lack of control over the programming block in the model makes it difficult to build a separate tool to warn flash flood in the study area Within the scope of this study, a separate hydrological mathematical model (CTM) is developed for the study area Java language is used to build the model 2.2.1 Structure of the mathematical model (CTM) The model structure consists of main components, namely: basin component, river section component, and connection component Of the components, the basin component has the function of converting rainfall to flow on the river basin The author uses two methods to calculate the flow from rainfall: (1) SCS-CN method and (2) method of linear reservoir (NAM) The river section component is to route the flow in the river, in which the routing can be done by some common hydrological routing methods such as linear reservoir and Muskingum The connector component is the simplest one that connects the other components together The model structure is shown in the diagram in Figure 2.4: Figure 2.4 Diagram of the mathematical model structure in the study 2.2.2 Introduction to programming languages In the study, Java is used as one of the object-oriented programming languages It is used in the development of software, websites, or applications on mobile devices 2.3 Calculation of the rainfall threshold of flash flood FFG and index of flash flood threat (FFT) 2.3.1 Determination of rainfall threshold FFG 11 To determine the rainfall threshold FFG, trial and error method is used The diagram to determine the rainfall threshold FFG in real time is shown in Figure 2.13 below Figure 2.13 Diagram to determine the rainfall threshold FFG At each step, rainfall data is updated to the current time, and a series of rainfall data updated to the current time is obtained (X1, X2,…Xt) The updated rainfall data series is taken as input to the hydrological mathematical model, and the current flow of the basin is obtained Hypothesize the rainfall at the next time is FFG (Xt+1 = FFG) The hypothetical rainfall data FFG is taken as input to the hydrological mathematical model, and the hypothetical flow is obtained Accumulate the current flow and the hypothetical flow, and the flow at the basin outlet is obtained Compare the flood at the outlet of the basin If the peak flood reaches the bankfull discharge, the initially hypothetical FFG value is the value to be found and then move to the next time step If not, repeat the process 2.3.2 Determination of FFT index (Flash flood threat level) FFT (Flash Flood Threat) is the amount of rainfall in a certain period exceeding the corresponding FFG value The flash flood risk index (FFT) is based on the difference (or percentage) between the cumulative rainfall during the forecast period and the corresponding FFG value 12 To determine the FFT value, the forecast rainfall is applied In the case study for the study basins, the author uses the GEM model to determine the forecast rainfall value This forecast rainfall value in each period is compared with the corresponding FFG value The FFT value is calculated as the difference of the forecast rainfall and the FFG value If the FFT value of any sub-basin at any time is greater than zero, it indicates that there is a risk of flash flood 13 CHAPTER RESEARCH RESULTS FOR FLASH FLOOD WARNING METHODOLOGY FOR NAM LY AND NA NHUNG BASINS IN HA GIANG PROVINCE 3.1 Results of building the empirical equation to determine the bankfull discharge for the mountainous river basins Nam Ly and Na Nhung in Ha Giang province As analyzed in the overview, the bankfull discharge Qbf is determined from the empirical equation by combining the actual topographic survey with the construction of the correlation between the bankfull discharge and the basin characteristics The results of building the empirical equation to determine the bankfull discharge are presented as follows: 3.1.1 Survey to determine the overflow water level and the corresponding bankfull discharge From the methodology discussed in Chapter 2, the overflow water level is determined from signs From the overflow water level and the correlation curve of the water level and discharge for the study area, the corresponding bankfull discharge is also determined This process is carried out in all survey sites Based on the criteria described in the methodology section, 34 locations are surveyed Detailed locations and signs are shown in Table 3.1 The table also shows the overflow water level investigated and the bankfull discharge calculated 14 Table 3.1 Survey sites and signs of overflow No Name Longitude Latitude Sign 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q11 Q12 Q13 Q14 Q15 Q16 Q17 Q18 T1 T2 T3 T4 T5 T6 T7 T8 N1 104.580 104.582 104.581 104.578 104.579 104.569 104.574 104.561 104.552 104.569 104.534 104.536 104.593 104.585 104.574 104.576 104.555 104.745 104.742 104.749 104.761 104.752 104.759 104.767 104.783 104.729 22.627 22.616 22.605 22.593 22.584 22.587 22.567 22.548 22.538 22.568 22.576 22.564 22.583 22.576 22.566 22.573 22.574 22.688 22.682 22.672 22.664 22.695 22.696 22.694 22.683 22.716 (1,7) (1,7) (1,6) (1,7) (2,6) (6) (3) (3,6,7) (6) (6) (4,6) (6,7) (1) (6) (6,7) (1,6) (7) (1,6) (5) (6) (6) (6,7) (7) (6,7) (6) (3) Overflow Bankfull water level (m) discharge (m3/s) 817.44 653.30 532.65 464.60 378.75 623.50 367.73 290.50 279.40 385.65 757.63 603.80 510.56 470.75 366.60 410.80 830.75 970.80 1044.80 938.00 1350.50 737.86 827.80 980.73 1128.75 571.20 27.70 18.52 52.44 24.58 78.11 24.91 86.68 20.69 109.20 23.62 30.12 38.52 35.15 47.34 40.07 81.63 15.19 53.65 43.95 32.58 21.26 61.04 53.62 40.31 21.18 93.86 3.1.2 Building an empirical equation to determine the bankfull discharge To build the empirical equation for the study basins, regression equations are developed based on the exponential form From the calculation results evaluating the correlation of independent variables and dependent variable, only the basin area, river bed slope, and average elevation of the basin are statistically 15 significant in the correlation equations to determine the bankfull discharge Therefore, research is conducted to build the correlation equations that are a combination of these three variables, as shown in Equations III, IV, and V in Table 3.2 Table 3.2 shows multivariable correlation equations and evaluation criteria corresponding to the above cases Table 3.2 Multivariable correlation equations to determine bankfull discharge based on basin characteristics No Equation I Qbf = 12.450F0.530 II Qbf = 7.591F 0.211 0.350 Lc Ztb 0.255 0.202 Slv0.438Ss- CN-0.326 Standard Correlation Deviation (m3/s) coefficient (%) 13.2 94.8 12.2 92.3 III Qbf = 10.749F0.493Ss-0.121 12.8 95.3 IV Qbf = 6.318F0.528Ztb0.097 13.2 94.9 V Qbf = 3.255F0.482Ss-0.150Ztb0.165 12.5 95.6 3.1.3 Comment on the results The results show a close correlation between the basin area and the bankfull discharge, with correlation coefficient = 94.8 % This is completely appropriate because the basin area is the area catching rainwater which is then concentrated and form the flow at the outlet Therefore, the basin area will determine the discharge value at the outlet The next factor affecting the bankfull discharge is the slope of the river bed The greater the slope is, the greater the value of the bankfull discharge is When comparing the statistical values of Equation I and Equation IV, there is no significant change It means that the impact of the basin average elevation is not clear Similarly, the length of the main river, the slope of the basin, or the CN index are not significant in the equations These factors sometimes even cause interference when the correlation coefficient of Equation II (the equation including these variables) is smallest (correlation coefficient = 92.3%) Of the above equations, Equation V: Qbf = 3.255F0.482Ss-0.150Ztb0.165 is the one with the highest correlation coefficient (Correlation coefficient = 95.6%) This is also the equation proposed by the author to use for the study area Only with topographic data, such as available topographic DEM, the basin characteristics can be determined quickly Furthermore, this also gives the user 16 more flexibility in determining the bankfull discharge at all desired sites in the basin 3.2 Results of building hydrological mathematical model for the study basins The author has built a model to calculate flood flow to be applied in this study: (1) The model is capable of taking input data by importing from a file or copying from excel or text format Table 3.4 Calculation methods in CTM model Component Absorption Surface flow Underground flow SCS-CN method Fixed Monthly CN SCS Note Tank type Have various SCS curve types Surface tank Snyder Fixed Monthly Underground tank Have many options reduction Flow routing in Muskingum the river Linear reservoir (2) The model has the ability to drop/move objects This feature makes it convenient for users to simulate the system of the basin (3) The model has a wide range of choices for the calculation Table 3.4 above describes the methods in calculating different components The model is built with main components including basin, river section, and connection The model simulates well, even with complicated cases which include many connected basins and river sections The model interface is presented in Figure 3.9 below 17 Figure 3.9 Interface of CTM model 3.3 Evaluation of the GEM model to determine the forecast rainfall for the study basins Currently, many global forecasting models are applied to forecast rainfall, providing input for computational research In this study, the author uses the GEM model to forecast rainfall, which acts as a basis for comparison with the rainfall threshold FFG to provide information for warning Before applying the GEM model, the author will evaluate the suitability of the GEM model for the two study basins Nam Ly and Na Nhung in Ha Giang province The GEM model is tested using error indicators These values, calculated from the average rainfall values in the GEM model, are compared with the actual rainfall data from and automatic rain stations for Nam Ly and Na Nhung basins, respectively Results are obtained from heavy rain events: (1) June 114, 2020; (2) July 4-7, 2020, and (3) July 19-22, 2020 For Nam Ly basin, the GEM model obtained the values of RMSE (12mm/3 hours), Bias (0.7mm/3 hours), r (0.12), and HSS (0.51); for Na Nhung basin, Bias (- 3.5 mm/3 hours), RMSE (13.5 mm/3 hours), r (0.22) and HSS (0.43) values were calculated In short, from these results, it can be seen that the GEM model relatively recorded the significant change of rainfall from day to day in Nam Ly basin Meanwhile, for Na Nhung basin, the GEM model could capture quite accurately the rainfall in the basin, as shown in Figure 3.15 18