Characterization of topographic surface and evaluation for flood hazard zonation in coastal lowland of danang city, vietnam

1

大阪市立大学大学院創造都市研究科

博士学位申請論文

CHARACTERIZATION OF TOPOGRAPHIC SURFACE AND

EVALUATION FOR FLOOD HAZARD ZONATION IN COASTAL

LOWLAND OF DANANG CITY, VIETNAM

2017 年 03 月

大阪市立大学大学院創造都市研究科

創造都市専攻創造都市研究領域

D13UD509

Tran, Thi An (トラン, テイ アン)

2

ACKNOWLEDGMENT

This thesis was completed under the kind assistances of number of people who I

am in debt. I would like to take this opportunity to express my gratitude to these people.

First of all, my deepest gratitude is sincerely to my supervisor - Prof. Venkatesh Raghavan

for his great supervision, support and continuous encouragement during my Ph.D course

at Osaka City University. His guidance always helped me in open-minded thinking and

motivated me to be able to pursue my research interests. I greatly appreciate him for

patiently listening to me, understanding my weakness and strengthening me when I have

any trouble in researching. I learned from him not only research methods but also how to

evaluate an issue, writing skills and also many moral guidance. His advises were valuable

not only for my study but also for my life in Japan. He was more than a supervisor for me.

My sincere gratitude goes to Prof. Shinji Masumoto, Graduate School of Science

for his continuous support for me from the first day I entered this Osaka City University.

His immense knowledge and guidance helped me during my research study and he gently

pointed me in the right direction when I had mistakes. Thanks to his great support, I could

pass over difficulties and finish my Ph.D course.

I would like to express my sincere thanks to Associate Prof. Go Yonezawa and

Associate Prof. Daisuke Yoshida for their great support for me in Geoinformatics Lab. I

would like to thank them for giving me useful comments during seminar over last four

years which were valuable for my research.

I wish to express my gratitude to Dr. Susumu Nonogaki, Geological Survey of

Japan for great support for me in DEM generation method. He has given me the guidance

in detail, the valuable comments and suggestions that helped me grow up in research. His

developed BS-Horzion program which was investigated in this study is one of the most

important tools for the completion of this thesis. Especially thanks to Prof. Kiyoji Shiono,

Japan Society of Geoinformatics for his kindly suggestions and corrections for the

evaluation of parameters in BS-Horizon DEM generation.

My sincere thanks especially goes to my colleagues in Geoinformatics Lab for

accompany with me during my course. The discussion with them were always helpful for

improving my study.

3

I also gratefully acknowledge Japanese Ministry of Education, Culture, Sport,

Science and Technology for granting me Monbukagakusho - MEXT scholarship to enable

me to study in Osaka City University. Also I thank to Graduate School for Creative Cities

for their facilities support for my research. Besides, I would also thank to Danang

Department of Natural Resource and Environment for providing field survey data which

was very important in this research.

I would also like to thank The University of Danang, my professors as well as my

colleagues in Faculty of Geography, University of Science and Education, The University

of Danang for their support for me to study abroad.

Thanks also goes to all of my friends in Japan. With memories of joy and help

from them, I will never forget the great times we spent together. Especially thanks to Ms.

Sachiko Raghavan for her kind help in everything during my life in Japan.

Last but not least, my unlimited thanks go to my beloved parents for their long￾distance support and encouragement in every moment of my life, even though with lot of

their difficulties. Especially thanks to my little family, husband and my son for their great

love and heartening me up in researching. I could not finish my Ph.D without

encouragement from my family. Whatever I achieved is only to make them happy and

proud.

4

TABLE OF CONTENTS

................................................................................................................................................ Page

ACKNOWLEDGMENTS

LIST OF FIGURES

LIST OF TABLES

ABSTRACT ......................................................................................................................... i

Chapter One: INTRODUCTION ....................................................................................... 1

1.1. Overview and motivation ................................................................................................ 2

1.2. Research objectives ......................................................................................................... 3

1.3. Flood situations in Central Vietnam and Danang area .................................................... 3

1.4. Review of related researches........................................................................................... 4

1.5. Thesis outline................................................................................................................... 6

Chapter 2: FUSION OF OPTICAL STEREO AND InSAR DERIVED

GLOBAL DEMs................................................................................................................... 7

2.1 Introduction ...................................................................................................................... 7

2.2. Study area ........................................................................................................................ 8

2.3. DEM datasets................................................................................................................... 9

2.4. Fusion of optical stereo and InSAR derived DEM data .................................................. 11

2.4.1. Pre-processing ................................................................................................... 11

2.4.2. DEM quality assessment ................................................................................... 13

2.4.3 Minimizing DEM bias effect ............................................................................. 14

2.4.4 DEM fusion algorithm ....................................................................................... 16

2.4.4.1. Weighted averaging .............................................................................. 16

2.4.4.2. Filtering the noises for fused DEM ....................................................... 18

2.5. Accuracy assessment for fused DEM.............................................................................. 19

2.6. Limitations of fused DEM............................................................................................... 21

Chapter Three: GENERATION OF HIGH RESOLUTION DEM USING

BS-HORIZON METHOD ................................................................................................... 22

3.1 Introduction ...................................................................................................................... 22

3.2. BS-Horizon theory........................................................................................................... 23

3.3. Data.................................................................................................................................. 25

3.4. Evaluating effects of parameter settings on the BS-Horizon DEM generation............... 26

5

................................................................................................................................................ Page

3.4.1. Equality and inequality constraints .................................................................... 26

3.4.2. Effect of M and α settings................................................................................... 28

3.4.2.1. M and α settings in case of using only equality constraints................. 28

3.4.2.2. Effects of M and α settings for equality-inequality constraint .............. 30

3.4.3. Surface characteristics for different inequality constrained intervals ................ 32

3.4.4. Evaluating effect of m1 and m2 settings.............................................................. 33

3.5. Discussion........................................................................................................................ 35

3.5.1. Comparing BS-Horizon DEM generation from equality and equality￾inequality constrained data ..................................................................................................... 35

3.5.2. Selection of appropriate parameters for BS-Horizon DEM generation .............. 36

3.6. BS-Horizon DEM assessment ......................................................................................... 37

Chapter Four: FLOOD HAZARD ZONATION USING MULTI-PARAMETRIC

ANALYTICAL HIERARCHY PROCESS (AHP)............................................................ 39

4.1. Introduction ..................................................................................................................... 39

4.2. Study area and data used ................................................................................................. 40

4.3. Methodology.................................................................................................................... 41

4.3.1. Data preparation ................................................................................................ 41

4.3.1.1. DEM generation for study area ............................................................. 41

4.3.1.2. Flood inundation mapping from satellite image.................................... 42

4.3.2. Analytical Hierarchy Process (AHP) method .................................................... 43

4.3.3. Causative parameters of flood ............................................................................ 45

4.3.3.1. Elevation based flood inundation (EFI) ................................................ 45

4.3.3.2. Distance from the river channel (DIST)................................................ 46

4.3.3.3. Topographic Wetness Index (TWI)....................................................... 46

4.3.3.4. Land use (LU) ....................................................................................... 47

4.3.3.5. Slope. ..................................................................................................... 47

4.4. Results ............................................................................................................................. 48

4.4.1. Determining the weights for parameters of flood hazard................................... 48

4.4.2. Flood hazard index (FHI) and flood hazard zonation ...................................... 49

Chapter Five: DISCUSSIONS AND CONCLUSIONS..................................................... 50

REFERENCES ..................................................................................................................... 54

6

LIST OF FIGURES

................................................................................................................................................ Page

Figure 2.1. Location of study area and topographic overview .............................................. 62

Figure 2.2. Flowchart of data processing .............................................................................. 62

Figure 2.3.Correlation between GDEM and Reference DEM before (left) and after (right)

filling voids ............................................................................................................................ 63

Figure 2.4. Comparing stream networks of global DEMsandReference DEM before (up)

and after (down) shifting DEM: (a) GDEM; (b) SRTM ....................................................... 63

Figure 2.5. Comparing GDEM and SRTM to Reference DEM: (a) before re-interpolation

SRTM and shifting data; (b) after re-interpolation SRTM and shifting data ........................ 64

Figure 2.6. Correlation of GDEM and SRTM in flat (a) and mountainous (b) areas ........... 65

Figure 2.7. A profile of GDEM and SRTM compare to Reference DEM in flat area .......... 65

Figure 2.8. Difference elevation of GDEM and SRTM with respect to Reference DEM

from mountain to flat area ..................................................................................................... 66

Figure 2.9. Behaviour of GDEM and SRTM to Reference DEM in difference topographic

contexts. (a) Whole area; (b) A area; (c) B area; (d) C area; (e) D area ................................ 66

Figure 2.10. Landform classification map from SRTM ........................................................ 67

Figure 2.11. Weighted averaging used to fused global DEMs .............................................. 68

Figure 2.12. Result of denoising algorithm (Sun et al. 2007) on fused DEM ...................... 68

Figure 2.13. Correlation between fused DEM and Reference DEM ..................................... 68

Figure 2.14. Difference in elevation between fused DEM and Reference DEM .................. 69

Figure 2.15. Histogram from the difference elevation maps of SRTM, GDEM and Fused

DEM. (X axis: cell values in tens; Y axis: number of cells in thousands) ............................ 69

Figure 2.16: Slope (a), profile curvature (b) and tangential curvature (c) of fused DEM ..... 70

Figure 2.17: Normal vector of topographic surface (a) and the angular difference between

two normal vector (Hodgson and Gaile, 1999) ..................................................................... 71

Figure 2.18. Limitations of Fused DEM compared to reference elevation data .................... 71

Figure 3.1. Location of study area including field survey point elevation data (a) and

Satellite RapidEye imagery in 2014 (b) of corresponding area. ............................................ 72

Figure 3.2. Distribution of field survey point elevation in study area based on different

cases of M .............................................................................................................................. 73

Figure 3.3. Equality and inequality constraints used in surface estimations.......................... 73

7

Page

Figure 3.4. Calculations of R( f ), J( f ) and the resulting Q( f ) in different cases of M and

 when using equality constrained data ................................................................................. 74

Figure 3.5. DEMs generated from equality constrained data using different M and

 settings ............................................................................................................................... 75

Figure 3.6. Representation of R( f ), J( f ) and Q( f ) according to different M and α when

using equality-inequality data................................................................................................. 76

Figure 3.7. DEMs generated from equality-inequality constrained data using different M

and  settings.......................................................................................................................... 77

Figure 3.8. Surfaces generated from equality constraints and equality-inequality

constraints with R( f ) <= 0.25 ................................................................................................ 78

Figure 3.9(a). DEM generated from equality constraints in different m1 and m2 settings

(Parameter M = 200,  = 1.2×102

) ......................................................................................... 79

Figure 3.9(b). DEM generated from equality constraints in different m1 and m2 settings

with R( f ) <= 0.25 (Parameter M = 200,  are different)....................................................... 80

Figure 4.1. Location of study area in Vietnam. (Background: ALOS PALSAR 15th

September, 2007)................................................................................................................... 81

Figure 4.2. Flow chart of the flood hazard zonation .............................................................. 82

Figure 4.3. ALOS PALSAR on 31st October 2007 (a) and the flood inundation map

extracted from PALSAR data (b)........................................................................................... 83

Figure 4.4. Parameters used in AHP based flood hazard zonation ........................................ 84

Figure 4.5. Flood hazard zonation map of the study area ...................................................... 85

Figure 4.6. Correlation between estimated and recorded flood depth data in 2007............... 85

8

LIST OF TABLES

Page

Table 1.1. Disaster history in Danang, Vietnam from 1997 to 2009...................................... 86

Table 2.1. General information of global DEMs and reference DEM .................................. 86

Table 2.2. SRTM before and after interpolation into 30m .................................................... 87

Table 2.3. Results of GDEM after filling artifacts and shifting ............................................ 87

Table 2.4. The mean errors of GDEM and SRTM according to land cover map ................. 87

Table 2.5. Mean of absolute error (MAE) from slope error maps of GDEM and SRTM on

each landform area ................................................................................................................. 87

Table 2.6. General statistics for the error of GDEM, SRTM and fused DEM ...................... 88

Table 2.7. Comparison of differences in some terrain parameters of GDEM, SRTM and

Fused DEM with respect to Reference DEM ........................................................................ 88

Table 2.8. Result of angular difference of unit NV between global DEMs, fused DEM and

Reference DEM ..................................................................................................................... 88

Table 3.1. Distribution of field survey elevation points in different cases of M ................... 89

Table 3.2. Statistics of DEMs from equality data in different M and  settings ................... 90

Table 3.3. DEMs from equality-inequality data in different M and  .................................. 91

Table 3.4. Statistics of the surfaces with R( f ) <= 0.25 ......................................................... 92

Table 3.5. Statistical results of 5m DEMs created from inequality constraints in different

intervals ................................................................................................................................. 92

Table 3.6(a). Statistical results of 5m DEMs in different m1 and m2 settings (Input:

Equality data, M = 200,  = 1.2×102

)..................................................................................... 93

Table 3.6(b). Statistical results of 5m DEMs in different m1 and m2 settings with R( f ) <=

0.25 ......................................................................................................................................... 94

Table 4.1. Data used in flood hazard zonation ...................................................................... 95

Table 4.2. Saaty scale for various elements comparison........................................................ 96

Table 4.3. Random index (RI) used to compute consistency ratios (CR) .............................. 96

Table 4.2. Pair-wise comparison (PC), normalized values (NV) and corresponding

weights for flood hazard parameters ...................................................................................... 97

Table 4.3. Parameters classes, ratings and corresponding weights........................................ 98

Table 4.4. Flood hazard zonation based on FHI..................................................................... 99

9

ABSTRACT

Flooding is one of the most frequent and damage causing natural disaster in Vietnam.

Flood occurrence becomes more exacerbated in the coastal areas of Central Vietnam.

These areas, especially Danang and Quang Nam provinces were experiencing several

flood events in the past. Due to the increasing frequency of flood events, flood hazard

zonation become becomes indispensable for evaluation the flood risk in this region. The

topography which directly related to flood hazard but such relationship is still poorly

understood.

Several methods have been developed for flood hazard zonation using various

approaches. Previous studies have been carried out in order to determine the flood

potential in Danang City, Vietnam based on hydro-geomorphological methods integrated

with remote sensing data (Ho et al., 2012; Do et al., 2014). However, those studies reveal

limitations in lack of high resolution digital elevation model (DEM) and near-flood events

satellite data which leads to generate inaccuracy results in developing a comprehensive

flood prevention plan.

This study aims to characterize the topographic surface and evaluation for flood

hazard zonation in coastal lowland of Danang City, Vietnam. Digital elevation model

(DEM) is one of the most important data source for accurately characterizing the flood

hazard potential. Therefore, in the first stage of this study, algorithms for high quality

DEM generation was investigated. Firstly, global free DEM data including Advanced

Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global DEM

(GDEM) and Shuttle Radar Topographic Mission (SRTM) DEM data were utilized to

generate DEM for Danang City. It is observed that the accuracy of GDEM and SRTM

varies depending upon the geomorphological characteristics of target area. Fusion

between two global DEMs using geomorphological approach is an appropriate solution to

enhance the quality of free DEMs for Danang City, Vietnam. The data fusion technique

was applied by weighted averaging of GDEM and SRTM based on the topographic

context. Fused DEM were compared with reference DEM to discuss about accuracy and

impact of terrain related parameters in variation on DEM quality. Results indicate that

fused DEM has improved accuracy than individual global DEM and most artifacts were

successfully eliminated.

10

Fusion of global free DEMs supports the effective utilization for global or regional

scale study, especially where high quality elevation data are not available. However, in the

small lowland area the topography as well as flood related conditions need to be

characterized in detail, fused DEM is insufficient. In order to characterize the flood hazard

potential of the coastal lowland area in Danang City, it is necessary to generate high

resolution DEM. In this study, BS-Horizon surface generation method (Nonogaki et al.,

2012) based on bi-cubic spline function and exterior penalty function was investigated for

generation of 5m resolution DEM. Elevation data extracted from high density field survey

points were used as equality elevation constraints. Further, the inequality constraints

which are the inter-contour height information or the elevation relationships (such as

“above” or “below”) were also taken into consideration for surface generation. This study

is the first time that the parameters involved in BS-Horizon method have been fully

evaluated. The findings facilitate the selection of suitable parameters for DEM generation

using the BS-Horizon method. The generated DEM can be used as a reliable input data for

further extraction of geomorphology and flood hazard zonation of this study area.

In the third part of this research, a hierarchical model for flood hazard zonation for

coastal lowland in Central Vietnam was applied using the Analytical Hierarchy Process

(AHP). DEM was used for extraction of geomorphological parameters such as slope,

topographic wetness index. ALOS PALSAR satellite imagery in 2007 was utilized for

mapping flood inundation which is useful in flood hazard zonation. Five factors including

elevation based flood inundation (EFI), distance from the channel (DIST), land use, slope

and Topographic Wetness Index (TWI) have been taken into consideration for analyzing

flood hazard probabilities. Flood Hazard Index (FHI) has been integrated for the purpose

of flood hazard zonation. FHI aims to identify areas with flood risk based on the

quantitative determination for weighting each parameter. AHP method was applied to

assign weights according to the pair-wise comparisons of the parameters. FHI then was

reclassified into four levels of flood hazard potential. Field survey flood pillars and flood

inundation areas extracted from ALOS PALSAR in 2007 was used to evaluate the flood

hazard zonation map. As a result, nearly 80% of inundated areas in ALOS PALSAR

image in 2007 belong to high and very high flood hazard potential. The comparison with

field flood pillars also indicated good correspondence with demarcated zones and

estimated depth. The proposed method demonstrates the effectiveness of applying AHP

method in flood hazard assessment for the areas with larger extents.

11

Chapter One

INTRODUCTION

1.1. Overview and motivation

Flood is a natural phenomenon that results in the temporary submerging with

water of a land that does not occur under normal conditions (European Commission,

2007). Flood can be caused by a number of factors including current weather patterns

associated with geomorphologic and hydrologic conditions, natural changes in the

environment and recent development in the community. Flood is one of the most

devastating natural hazards which lead to the loss of lives, properties and resources

(Forkuo, 2011). Flood occurrence becomes more serious in the coastal areas where large

proportion of population, industry, agriculture and infrastructure are concentrated, since

these areas are directly suffer from storms, heavy rains and water easily accumulates.

Coastal flooding has become the critical concern in many countries.

Located in southeast Asia and tropical monsoon climate region, Vietnam has

elongated shape with the coastline stretching across 3260 km in length and the Truong

Son Mountain Range in the West (Ho, 2013). The Central Vietnam which are mainly

coastal plains are characterized by narrow shape with short and steep river. Flood

occurrence in Central Vietnam becomes more exacerbate due to the extreme weather

phenomena (especially typhoon, tropical cyclones and heavy rainfall), associated with

topographical characteristics. The low coastal zone in Central Vietnam has suffered from

flood disaster almost every year.

The coastal lowland area in the south of Danang City is a part of Vu Gia - Thu

Bon river basin, Vietnam where is affected by serious and repetitive flood hazard. This

area is characterized with two seasons in a year: a rainy season from August to December

and a dry season from January to July, with rainfall mainly concentrated from September

to December (70% - 80% of the total annual rainfall). On average, this area is directly or

indirectly influenced by 1-2 typhoons and 1-2 great flooding spells each year (Do et al.,

2014). Understanding about topographical, hydrological as well as environmental

conditions which cause potential of flood risk is really important for flood management of

any area. In the coastal lowland of Danang City, The topography and geomorphology