Tóm tắt: 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.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 2 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

1

INTRODUCTION

1 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 2 basins, namely: Nam Ly and Na Nhung in Ha Giang province

2 Study objective

Based on research and practice to develop flash flood warning methods for small mountainous basins and experimentally applying the methods to the 2 basins Nam Ly and Na Nhung in Ha Giang province

3 Objects and scope of the study

The scope of the thesis is in mountainous river basins, with case studies in the 2 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

• Mathematical modeling method: using hydrological mathematical model which

is developed for the study basins to determine the flow value at the outlets of basins, which acts as a basis to calculate and determine rainfall threshold of flash flood FFG in real time;

sub-• Remote sensing and GIS methods to identify basin characteristics from map data, digital elevation models, satellite images, etc

5 Scientific and practical significance

3

The study has applied flash flood warning method for mountainous basins in the

2 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 2 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

6 Structure of the thesis

In addition to the introduction, conclusion, and appendices, the summary of the thesis is organized into 3 chapters

CHAPTER 1 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 4 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 sediment-related 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 6 hours) The model uses the intermittent rainfall threshold in sub-basins, 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 sub-basin

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)

5

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

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 2 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 4 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

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 cross-section 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

7

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 2 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

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) Figure 2.1 Diagram of the method to determine the bankfull discharge

9

In addition, in some documents, it is also possible to rely on (7) discolored marks

or stains, sand on the rocks to determine the overflow water level The study conducted a survey on the study area Figure 2.2 depicts some common signs of overflow water level

2.1.2 Correlation between water level and discharge for survey sites

The study also establishes the relationship between water level and discharge for each survey site Using Manning Chow's formula (1959) [15] to construct the curve Q = f(Z)

𝑄 = 𝑛1𝐴𝑅2/3√𝑆 (0-1)

In which: A and R are hydraulic area and radius, respectively These values can

be determined through the shape of the measured cross-section; S is the hydraulic slope, which, in this study, is taken as the slope of the river bed measured at the survey site; n is the roughness coefficient, according to the guidance of Barnes (1969) [16], the roughness of each corresponding river section is determined From formula (2-1), flow values corresponding to water levels are determined Summarizing all these values, the Q = f(Z) curve of each site is identified The study also measured some values of the flow and water levels to test the accuracy

of the built curves

2.1.3 Building an empirical equation for the calculation of the bankfull discharge

2.1.3.1 Input variables in the empirical equation

According to Bent (2013) [13], the value of bankfull discharge is not only affected by the basin area but also by factors such as land use, hydro-Figure 2.2 Some common signs to identify the overflow water level

meteorological conditions, and other characteristics of the river system In this study, due to the small area, the difference in rainfall in the study area can be ignored Therefore, meteorological conditions are not considered in this study In the study, the author provides input variables which can be calculated from available digital data sources in Vietnam Input variables include: basin area (F), basin slope (Slv), average basin elevation (Ztb), main river length (Lc), main river slope (Ss), density of the river (n) which characterizes the topography, CN index to characterize the buffer element

2.1.3.2 Method of building empirical equation to determine the bankfull discharge for the study basins

Statistical methods are used in building the regression function to determine the bankfull discharge for the sub-basins in this study In most studies by Bent (2013), Fernandez (2017), and Lumina (2006) [13, 17, 18], exponential functions were used to describe the correlation between dependent and independent variables In this study, the exponential equation form is also applied in all regression functions, including univariate and multivariate regression Univariate regression is applied to the basin area, while multivariate regression is built for the combination of basin area and other factors The results of the empirical equation are presented in Chapter 3

2.2 Building a rainfall-flow mathematical model combined with real-time rainfall data to calculate and update the current conditions of the basin

For flash floods, it is important to identify the initial condition of the basin To

do this, building a hydrological mathematical model that can be integrated into the flash flood warning tool for the study area is necessary, from which the alarm threshold value FFG can be determined in real time

Currently, there are many models that have been developed in the world Some models are applied in practical problems in Vietnam, such as NAM, HEC HMS, and TANK The TANK Sugawara (1995) [19] and NAM Nielsen (1975) [20] models were based on the vertical tank to calculate the flow These models have the advantage as simple structure and good simulation of flow components HEC HMS is also a very commonly used model today with many calculation options for each flow component to suit each user's conditions However, these are still foreign models with input data as well as outputs in a format that is not suitable for the intended tool used for the study basins If the data are wished to be put in

11

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 3 main components, namely: basin component, river section component, and connection component Of the 3 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:

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

Figure 2.4 Diagram of the mathematical model structure in the study