Study on Characteristics of Future Cyclone and Associated Storm Surge Inundation Damage in Bangladesh

九州大学学術情報リポジトリ Kyushu University Institutional Repository

モハマド, アブドゥル, アル, モヒット

https://doi.org/10.15017/2534441

出版情報：九州大学, 2019, 博士（工学）, 課程博士バージョン：権利関係：

Study on Characteristics of Future Cyclone and Associated Storm Surge Inundation Damage in Bangladesh

Ph.D. Dissertation

This thesis is submitted to the department of Maritime Engineering, Kyushu University, in partial fulfillment of the requirements for the Degree of Doctor of Engineering.

By Md. Abdul Al Mohit

Department of Maritime Engineering Graduate school of Engineering Kyushu University Fukuoka, japan June 2019

Abstract

In this dissertation, the characteristics of cyclone which can strike in future due to climate change in the south-asian overpopulated country Bangladesh, has been investigated as well as the risk factors of storm surge and flood, caused by the severe cyclone have also determined. Bangladesh is one of the most affected countries as a consequence of climate change. Every year, due to different types of natural disasters thousands of people lost their lives. Among all these disasters, severe cyclone is one of them. As a result of this severe cyclone, the flood and the time afterward the flood hinder the nature. Moreover, recently, because of pogrom at least ten lakhs (one million) Rohingya refugees have taken shelter in the coastal region of Bangladesh. The government of Bangladesh is planning the migration of these Rohingyas to the newly emerged island “Bhashan Char” which is acute flood and severe cyclone prone region. Therefore, it is very much essential to know the risk factors of living congeniality in this island. In my research, for this region I have determined the alterations of the characteristics of the cyclones and also figured out the severe cyclones which can strike in future. To determine the storm surge, I have developed a non-linear numerical bay-river coupled model, which represents the accurate results. Overall, I have measured the risk factors of severe cyclones which can occurred in future in the south-east coastal part of Bangladesh like Chittagong and the new habitation of Rohingyas, also figured out the measures which can be taken by the government of Bangladesh to confront these disasters.

In Chapter 1, the aim, objectives, background, and framework of the research were explained. Basically, this chapter is a general introduction chapter.

In Chapter 2, reviewed the features of the storms in the Bay of Bengal such as landfall angle, Power Dissipation Index, Accumulated Cyclone Energy, cyclone genesis, seasonal activity, and so on. It is found from the analysis that the behavior of the cyclone is changeable due to climate change. The genesis of the present cyclone will move slightly upwards at lower latitudes in future. Besides, there is a seasonality of the present cyclone that will change in the future. Though the actual number of cyclone will be decreased but the accumulated cyclone energy will increase.

In Chapter 3, cyclone strike rate for the Bangladesh coast and other coast of Bay of Bengal were investigated, The characteristic features of a cyclone like landfall angle, wind speed, central pressure, and translation speeds were investigated for present and future climate condition. In this analysis, it is found that the east coast of India is a prone zone for future cyclone but most of the severe cyclone strike the Bangladesh

coast. This phenomenon indicates that the east coast of India and the west part of Bangladesh will face the huge precipitation due to atmospheric depression.

In Chapter 4, a non-linear bay-river coupled numerical storm surge model that incorporate the major river system of Bangladesh is developed, which can simulate the accurate results of storm surge. It is found from the analysis that the water level reduces near the junction area due to inland penetration. In the rainy season and the drought season, surge height is flactuated due to fresh water discharge through a river.

In Chapter 5, FVCOM is used to simulate the surge and its associated inundation along the coast of Bangladesh. The model result is compared with the developed bay- river coupled model to find the preformance of bay-river coupled model. Both the model simulation results make a good agreement with the observed result. To find the dangerous cyclone, the cost effective bay-river coupled model is used. Finally, an inundation area is measured by FVCOM.

In Chapter 6, storm surge inundation due to the future dangerous cyclone is investigated along the east zone of Bangladesh coast. The summerized characteristics of future severe cyclone were used to modify the present severe cyclone and investigate the flood risk. This study found that the future dangerous cyclone inundated 90% land of the Bhasan Char and 20% of the Chittagong airport area. This study also found that the present embankment of Bhasan Char is not suitable for flood disater prevention. This analysis of future disaster risk may helpful to the government of Bangladesh for the mitigation of future disaster risk and a good adaptation policy.

In Chapter 7, the overall findings of the study were briefly summerised in this chapter. In addition, some recommendation were presented based on the limitations of this study.

Acknowledgements

First of all, I am grateful to the Almightily Allah, who keep me healthy to submit this thesis paper.

I would like to express my deep gratitude and respect to my honarable supervisor Dr. Noriaki Hashimoto, Professor, Department of Maritime Engineering, Kyushu University, Japan, for his constant inspiration, invaluable advice, proper guidance, constructive criticism, stimulating discussion and valuable suggestions. My cordial thanks to him for his kindness and special support.

I wish to express my respect and cordial thanks to Dr. Masaru Yamashiro, Associate Professor, Department of Maritime Engineering, Kyushu University, Japan, for providing me the necessary suggesstions, guidelines and especially for helping me to understand daily live in Japan. I am very greatful to him for his patience and gentles and precious help. Grateful thanks to my committee members, Professor Shinichiro Yano and Associate Professor Yoshinari Hiroshiro for their good examination and review on my study with their prudent knowledge.

I want to mention here other two people who contributed to the work by inspiring feedback and technical support, much more thanks to Dr. Yoshihiko Ide and Mr. Mitsuyoshi Kodama. Without their great effort, it would have been quite impossible to complete this study.

I would like to expresses heartfelt thanks to all students of Coastal and ocean engineering laboratory who always showed graciousness and friendship during the period of my study. Specially, I thanks to Mr. Taiki Shibata and Mr. Kazuhiro Nakatani for their help regarding laboratory activities.

I acknowledge MEXT (Ministry of Education, Culture, Sports, Science and Technology) for financial support and the authority of Kyushu University to give me the facility for continuing my Ph.D. research. I also acknowledge the authority of Islamic University, Kushtia, Bangladesh to give me the permission for continuing my study in abroad.

Finally, I am deeply indebted to my parents and other members of my family for their moral support, blessings and encouragement through the study. I also would pay tribute to my wife (Zannat) and my son (Sadil), who had to deal a long absence of my take care and stay lonely due to my study in abroad.

Md. Abdul Al Mohit June, 2019

iii

CONTENTS

Page No Chapter 1: General introduction

1.1 Background 1 1.2 Study area 2 1.3 Previous study and problem identification 3 1.4 Present study with the aim 7

Chapter 2: Cyclone activity analysis under present and future climate condition

2.1 Chapter summary 8 2.2 The data 8 2.3 Activity analysis 11 2.3.1 Genesis analysis 15 2.3.2 Occurrence analysis 21 a) Yearly and monthly activity 21 b) Seasonal activity 25 c) Occurrence behavior at El Niño and La Niña 26 2.3.3 PDI and ACE Analysis 27 2.4 Conclusions 29

Chapter 3: Analysis of tropical cyclone characteristics in the Bay of Bengal

3.1 Chapter summary 30 3.2 Basic assumptions 30 3.3 Cyclone strike analysis in the Bay of Bengal 31 3.4 Cyclone strike analysis in Bangladesh coast 37 3.5 Characteristics of cyclone parameter 39 3.5.1 Central pressure 39 3.5.2 Landfall angle 41 3.5.3 Maximum sustained wind radius 43 3.5.4 Translation speed 45 3.6 Conclusions 47

Chapter 4: Development of numerical storm surge model 4.1 Chapter summary 48 4.2 Shallow water model equations 48 4.2.1 Average procedure 49 4.2.2 Vertically integrated equation 52 4.3 Boundary conditions 59 4.4 Determination of the forcing terms 61 4.5 Numerical procedure 62

4.5.1 Model setup 62 4.5.2 Numerical discretization of the model 64 4.5.3 Computation process 68 4.6 Model validation and outcomes 69 4.7 Conclusions 75

Chapter 5: Comparative analysis of two numerical storm surge models 5.1 Chapter summary 76 5.2 Concept of FVCOM 76 5.3 FVCOM model equations 77 5.4 FVCOM model setup 78 5.4.1 Unstructured triangular grid 78 5.4.2 Boundary conditions 79 5.4.3  Coordinate 80 5.5 Study domain 82 5.6 Forcing factors 83 5.6.1 Atmospheric pressure 83 5.6.2 Wind profile 85 5.7 Model data 87 5.7.1 Bathymetry data and tide 87 5.8 Simulation and result discussion 89 5.9 Conclusions 95

Chapter 6: Disaster risk of future cyclone inundation 6.1 Chapter summary 96 6.2 Objective 96 6.3 Dangerous cyclone analysis 98 6.3.1 Water level elevation analysis of severe storms in 98 different conditions 6.3.2 Inundation risk analysis for different conditions 103 6.4 Conclusions 111

Chapter 7: Conclusions 112

References Appendix-1

List of figure

Page No. Chapter 1

Fig.1.1 Bangladesh coastal area with tide station location 2 information Fig.1.2 Climate change impact in Bangladesh 5 Fig.1.3 The present situation of refugees (Rohingya) and their 6 accommodation

Chapter 2

Fig.2.1 Monthly mean inter annual variation of SST 10 Fig.2.2 Cyclone eye point for present climate condition 11 Fig.2.3 Cyclone activity for the future climate scenario of AGCM 12 Fig.2.4 Cyclone activity of 90-member ensemble simulation for six 15 scenarios Fig.2.5 Cyclone genesis location for present climate condition 16 Fig.2.6 Cyclone genesis location for future climate condition of 17 AGCM data Fig.2.7 Cyclone genesis location for future climate condition of 18 d4PDF data Fig.2.8 Genesis probable area of present climate condition 20 Fig.2.9 Genesis probable area of d4PDF future climate scenario 20 Fig.2.10 Genesis probability area of AGCM future climate scenario 21 Fig.2.11 Yearly average number of cyclone occurrence for present 22 and future climate scenario. The upper left and lower left figure (a)(c) shows the yearly occurrence number in present scenario. The upper right and lower right figure (b)(d) represents the future scenario. Fig.2.12 Average occurrence number of each scenario in each year 23 for present and future climate condition Fig.2.13 Monthly average of cyclone occurrence. (a) Present climate 24 scenario, (b) future climate scenario, (d) Average of Present climate scenario data, (d) Average future climate scenario data Fig.2.14 Monthly mean occurrence. (a) Present climate scenario, (b) 25 future climate scenario, Fig.2.15 Seasonal occurrence behavior. (a) Present climate condition, 26 (b) Future climate condition. Fig.2.16 Cyclone occurrence behavior at El Niño and La Niña 26 climate pattern Fig.2.17 PDI analysis of present and future climate condition 29

Chapter 3

Fig.3.1 Strike rate and the change of strike rate 32 Fig.3.2 Actual number and the change of actual number 32 Fig.3.3 Strike rate in different coast of different countries 34 Fig.3.4 Strike rate in each country from all of cyclones on BoB 35 Fig.3.5 Coastal length based strike rate 36 Fig.3.6 Different parts of coastal area of Bangladesh. The upper 37 figure shows the coastal protections of the coastal areas Fig.3.7 Cyclone strike characteristics in each part of Bangladesh 39 coast Fig.3.8 Central pressure behavior of different coastal zone of 40 Bangladesh. Fig.3.9 Schematic figure of before and after landfall point 41 Fig.3.10 Polar histogram of landfall angle for low-intense cyclone 42 Fig.3.11 Polar histogram of landfall angle for intense cyclone 43 Fig.3.12 Mean and standard deviation of maximum sustained wind 45 radius Fig.3.13 Mean and standard deviation of translation speed 46

Chapter 4

Fig.4.1 Fundamentals of wave conditions 53 Fig.4.2 Model domain area 62 Fig.4.3 Mesh scheme with stair step representation 64 Fig.4.4 Surge height at different tide station when incorporate the 69 river and without incorporate the river Fig.4.5 Surge height at different tide station when incorporate the 70 river and without incorporate the river Fig.4.6 Comparison result of two different cases simulations (With 71 and without river) Fig.4.7 Combined figure of the time series of storm surge 71 Fig.4.8 Distribution of maximum water level and their difference 73 Fig.4.9 Surge height of a cyclone MORA at the junction point and 74 some different points. Left figure show the surge height at a single station for discharge case. Right side figure shows the maximum surge height in different tide station.

Chapter 5

Fig.5.1 Basic difference of structured and unstructured grids 78 Fig.5.2 Basic difference of Cartesian coordinate system and σ 81 coordinate system Fig.5.3 Unstructured grid of the model domain and the east part of 83 Bangladesh coast Fig.5.4 Pressure distribution of cyclone 1991 at the time of 84 maximum surge height and the time of coastline crossing. Figure (a) represents the distribution at 43 hour and Figure

vii

(b) represents the distribution at 48 hour Fig.5.5 Schematic diagram of gradient wind, field wind, synthetic 86 wind Fig.5.6 Wind profile of cyclone 1991. Figure (a) represent the 87 distribution at 43 hour and Figure (b) represent the distribution at 48 hour Fig.5.7 Bay of Bengal bathymetry 88 Fig.5.8 Nao99b tide comparison at the Hiron point tide station 89 Fig.5.9 Storm surge height at the simulation time of 43 and 48 hour, 91 upper figure (a) shows the surge height when the maximum surge was found at 43h and figure (b) shows the surge height when the cyclone eye point crossing the coastline at 48h Fig.5.10 FVCOM and FDM Model simulation result of water level 92 elevation at some tide station Fig.5.11 Comparison of surge height at two different tide station 93 Fig.5.12 Comparison of observed and simulation result 95

Chapter 6

Fig.6.1 Land features of Bhasan Char 97 Fig.6.2 Embankment condition of Bhasan Char 98 Fig.6.3 Different angle track and parallel track associated with their 99 surge height Fig.6.4 Surge height for the case of different wind speed and 100 translation speed Fig.6.5 Severe cyclone track in the different climate scenario 101 Fig.6.6 (a) Time series of water level elevation for different severe 102 cyclone, (b) Maximum surge height for different severe cyclone Fig.6.7 (a) Time series of water level elevation at different 102 bathymetry condition, (b) Maximum surge height of different cases Fig.6.8 Time series result of storm surge height for the different 103 conditions storm surge simulation, where the left figure (a) without tide simulation and (b) with tide simulation Fig.6.9 Physical view of inundation risk area 104 Fig.6.10 Inundation risk at Bhashan Char 105 Fig.6.11 Inundation risk at Chittagong 106 Fig.6.12 The figure shows the maximum water level due to the 107 present climate severe cyclone and future dangerous cyclone, where left figure shows the water level without tide and the right side figure show the water level with tide Fig.6.13 Analysis of inundation risk due to present climate severe 108 cyclone and future dangerous cyclone. The left side figures show the inundation condition at the time of 38 hour, 43 hour and 47 hour for dangerous cyclone. The right side figures shows the inundation condition at the same time for present climate severe cyclone

viii

Fig.6.14 Estimation of flood risk area for the present severe cyclone 109 and future dangerous cyclone. Left side figures show the different time of inundation condition due to a dangerous cyclone. The right side figures show the flood risk area due to present climate severe cyclonee. Fig.6.15 Inundation risk assessment due to the present climate severe 110 cyclone for the different embankment height. Fig.6.16 Inundation risk assessment due to the future dangerous 111 cyclone for the different embankment height.

List of Tables

Page No. Table 2.1 Exercised data in this study 14 Table 2.2 Details of future six scenarios 14 Table 2.3 Cyclone category along the Bay of Bengal 27 Table 2.4 Detailed summary of accumulated cyclone energy from 28 MRI-AGCM data Table 2.5 Detailed summary of accumulated cyclone energy of 28 d4PDF data Table 3.1 Number of cyclone strike in different country area. The 33 strike numbers are divided into two criteria: intense and low intense.

Table 3.2 Number of cyclone strike in different part of Bangladesh 38 coast Table 4.1 Overall maximum wave peak statistic from the past study 72 and the present study.

Table 5.1 Simulation condition of FVCOM 90 Table 5.2 Simulation results of different study and the simulation of 94 FVCOM and FDM

Chapter 1

Chapter1

General introduction

1.1 Background

A disaster prone low-lying country of south Asia is Bangladesh. The extended area of this country is 20o 45/ N to 26o 40/ N latitude and from 88o 05/ E to 92o 40/ E longitude. This area has a 710km coastline with long continental shelf with shallow bathymetry. Many small and big flat islands make the coastline more complex and the geographical location of this country makes more vulnerable for the different kinds of coastal disasters. The well-known coastal disasters are tropical cyclone, tornedo, coastal flooding, salinity, and so on. The rapid climate changes make this area more disaster prone, which makes the ecological imbalance. Huge number of tropical cyclones and its associated surge always cause a great loss of many lives and properties along this region. The concern is that Bangladesh will face various catastrophic cyclones every year due to global warming (Mohit et al. 2018a). According to Mohit et al. (2018b), about 500000 people were died by the past cyclone in this region (History record from 1970-2018). A proper cyclone warning system can be reducing the human death and economic losses resulting from the surge within the coastal area of Bangladesh. Recently, the Cyclone Preparedness Programme (CPP) under the Bangladesh Red Crescent Society (BDRCS) forwards cyclone-warning bulletins in the remote coastal area through 42,675 coastal volunteers. The government of Bangladesh is not always fully successful in recovering the situation by using this warning system. So, proper forecasting and the surge estimation is important. In the near future, Bangladesh will be the number one country of south Asia that mostly affected by the world climate change impact said by Vice President for South Asia Region of World Bank. Therefore, for this reason, the cyclone disaster will be increasing this area. To make disaster plane and develop an adaptation technique for the coastal people of Bangladesh, a well investigation of future cyclone characteristics and its associated surge is needed. Bangladesh government should take some proper policy to mitigate the

Chapter 1

cyclone disaster, as we cannot prevent the natural phenomenon. Thus, an effective storm surge modeling and the analysis of cyclone characteristics is highly desirable for making a future-disaster prevention policy. Through such a type of modelling, a proper warning system can be made which can minimize the huge losses resulting from storm surge. 1.2 Study area

The present study area of this research is the coastal area of Bangladesh. The coastal zone of Bangladesh covers 19 districts out of 64, and 153 police stations area. This coastal area covers 32% of the total area and 28% of the population of Bangladesh (Alam et al. 2002). But, the 12 districts and 51 police stations area are vulnerable to surge disaster risk. The lowest landmass of Bangladesh is known as delta that makes the coastal zone of Bangladesh, which is an extended Himalayan drainage ecosystem. The river system of Bangladesh is highly complex and dominated by three major rivers, namely the Gauges, Brahmaputra & Meghna, which formed this world's largest delta. The interested study region and the nearest tidal station is shown in Fig. 1.1. On average, 15,000 people were died annually by storms and flood (Smith 1989). Bangladesh coast is the region, which faces about 5% of the global tropical cyclones form. In 1970, one of the most devastating cyclones of the century struck Meghna estuary of Bangladesh killing about 300,000 people.

Fig.1.1 Bangladesh coastal area with tide station location information

Chapter 1

April 1991 Cyclone made landfall near Chittagong, which killed about 138,000 people. Due to the climate change impact, the coastal people migrated to the inland area to find the new livelihood. In addition, in the near future the migration rate will be increased. Some parts of coastal area are under severing threat of future disaster like future sever cyclone, future sea-level submergence, salinity, and so on. Most of the farmer already adapted their cultivation process in a dangerous situation of surge and inland flood. However, they are worried about the future disaster. 1.3 Previous study and problem identification

Tropical cyclone, Hurricane, Typhoon are the greatest natural hazards coastal area of different country. The characteristics of the cyclone are changes due to the global warming and climate change impact. To protect the human lives from the environmental disaster and provide a scientific judgment on climate change, an Intergovernmental Panel on Climate Change was establish at 1988. This intergovernmental organization gives the potential reports on Environmental situation that is useful to analyze the future disaster in each country. Different countries faces different types of environmental disaster. Bangladesh faces some cyclone disaster, on an average, 5-6 storms form in this region every year, which is 80% of the global casualties (Debsharma 2009). Many studies have been done all over the world to predict the surge height accurately. Most of the study region is North pacific, Atlantic, Pacific, Gulf of Mexico, Gulf of Carpentaria, and Bay of Bengal. The pioneering work of storm surge research was done by the scholars, Jelesniaski (1965, 1966, 1967), and Thacker (1977). At that time, the extensive work of surge model development was done in the North-west European continental shelf. The well-known works were done by the scholar Heaps (1973), Flather (1976), Davies (1976), Heaps and Jones (1981). In the Japan coast, Hasimoto et al. (2016) did the inundation estimation, and Yamasiro et al. (2014) studied risk of storm surge amplification. The largest coast of Northwest pacific is Vietnam coast and some researches on this coast are Takagi et al. (2014) and Esteban et al. (2014). The Gulf of Mexico is the prone area for storm surge flood risk, some well-known work was done by some scholar in this marshes (see, Cahoon et al. 2011; Walker 2001). The coast of Carpentaria in Australia faces

Chapter 1

largest storm surge due to the largest expanse of shallow water and the long lifetime of a cyclone (Needham et al., 2015) The countries, which covered by the Bay of Bengal and the Arabian Sea, are threatened by the strong cyclone. Most of the cyclone genesis was found in the Malaysian peninsula, which strike the coast of Bangladesh, India, Sri-lanka, Myanmar. Tropical storm with tide makes a great devastation along the coast of Bangladesh. A strong circulatory wind can play a vital role to increase surge height, but some other factors like extreme bending of the coastline, offshore islands, shallow bathymetry, and huge discharge through the Meghna and other rivers may affect this increasing surge height. Many studies have been done in this Bay of Bengal region. Most of the work was storm surge model development and tide surge interaction and so on. Some well-known works are mentioned here (see, e.g., Das 1972; Das et al. 1974; Johns and Ali 1980; Johns et al. 1985; Dube et al. 1985, 1986; Murty et al. 1986; Talukder et al. 1992; Flather 1994; Ali et al. 1999; Roy 1995, 1999a; Debsarma 2009; Rahman et al. 2011; Paul et al. 2012a, b, 2013, 2014). At first, Das (1972) developed a linear storm surge model, after that the development was done by Johns and Ali (1980) and Murty et al. (1986). Murty conclude that the interaction with tide and surge at the same time is difficult to consider. In 1980, Johns and Ali first incorporate the stair step representation of the Bay of Bengal coast to develop the storm surge model, which is consider the dynamic effect of the Ganges–Brahmaputra–Meghna river system and offshore islands of Bangladesh coast. However, in their stair step representation model, the accuracy of the coastal and island boundaries are depending upon the grid resolution. In their study, it is not ensured that the fine gird resolution can make better result. So, keeping the above limitation, Roy (1995) developed the model of Johns and Ali (1980), which is known as nested grid model. However, they incorporated only two big islands for the stair step representation. But, Paul and Ismail (2013) stated that Bangladesh is thickly populated with low lying small and big lands and offshore islands may influence surge intensity. At that time, the tide and surge was incorporate linearly, which is not realistic. Therefore, Paul and Ismail (2013) developed the model of Roy (1995) as a very fine nested grid model, which is also a non-linear interaction model of tide and surge. Nonetheless, in the Bangladesh coast, there is a big river inlet, which has a noticeable impact in storm surge simulation. Dube et al. (1986) did the

Chapter 1

parallel work at the same time. He assumes the river system of Bangladesh in their model simulation as an idealized river system. Furthermore, They (Dube et al., 2004) developed a one directional river model and two-dimensional bay model for couple model. In subsequent, Agnihotri et al., (2006) introduced a two-dimensional bay-river coupled model, which incorporates the east coast of India. But, his study also consider the idealized river system. Their study was limited by the lack of proper geometric representation of the river system. Recently, the commercial 3D models of storm surge simulation are useful to predict the water level accurately. The well-known simulation models are SLOSH (Sea, Lake and Overland Surges from Hurricane) Jelesnianski et al. (1992), CH3D (Sheng et al (1987, 2009) POM (Princeton Ocean Model) (Peng et al. 2004), FVCOM (Finite Volume Coastal Ocean Model) (Chen et al. 2003a; Rego et al. 2009) and ADCIRC (Luettich et al. 1992). Some researchers use the commercial 3D model to simulate the surge height along the Bay of Bengal. Hussain and Tajima (2017) find the tidal phase shift and surge characteristics by using the FVCOM. Tasnim et al. (2014) find the future surge height along the Myanmar coast by using FVCOM model. All of this works are storm surge related work along the Bay of Bengal coast. So, most of the work associated with storm surge has been done.

Fig.1.2 Climate change impact in Bangladesh

Chapter 1

But, this study find some problems related to storm surge and its associated disaster risk. The people of Bangladesh coast face various types of coastal disaster. Due to climate change impact, Bangladesh environmental condition is dramatically changed. In the near future, the characteristics of future cyclone may be changed. Therefore, it is important to investigate the future cyclone characteristics and its associated surge within this area. Now a day Bangladesh faces an unwanted complication due to the refugee of Myanmar. About 10 million refugees are now living in the coastal inland. Fig.1.3 shows the current situation of refugees (Rohingya). This figure illustrate the escape time of refugee from Myanmar, present residence of the refugee, and the proposed resident at Bhasan Char. The global political situation and the present condition of refugee indicate their return policy is uncertain. So, Bangladesh government want to settle them in a new born island known as Bhasan char, but the place is not investigated yet properly for residence. The authority of Bangladesh government has investigated the impact of tide and surge in a small range. But, it is important to investigate the surge induced inundation for present and future climate condition. Therefore, the present and future cyclone characteristics and its associated surge disaster investigations are important.

Fig.1.3 The present situation of refugees (Rohingya) and their accommodation

Chapter 1

Furthermore, a numerical storm surge model is needed to develop for storm surge disaster measure. Finally, an appropriate surge model should be investigated for disaster risk measure for the coast of Bangladesh. 1.4 Present study with the aim

To solve the current problem of Bangladesh coast and future disaster risk- reduction, this dissertation describes the future cyclone characteristics along the coast of Bangladesh, and a storm surge model was developed to forecast surge height along the Bangladesh coast. Furthermore, the result of the develop surge model was compare with the conventional FVCOM. Finally, the dangerous cyclone was investigated for the east zone of Bangladesh coast to measure the inundation risk along this area. The key goals of this study are:

1. To find the activity of the Bay of Bengal cyclone, the big data of Meteorological Research Institute (MRI) - Atmospheric Global Climate Model (AGCM) (Mizuta et al., 2006) and Database for Policy Decision-Making for Future Climate Change (d4PDF) were investigated. Finally, the observed cyclone activity was compared with the model data.

2. To find the behavior of present and future cyclone characteristics, the present and future cyclone track was investigated. The characteristics of a cyclone in the different climate scenario were investigated to find the dangerous cyclone track for the east zone of Bangladesh coast.

3. To find the surge height accurately, an operational sure forecasting bay-river coupled model was developed for the coast of Bangladesh.

4. To find the dangerous cyclone and justify the simulation result, the result of the bay-river coupled model was compared with the conventional storm surge model FVCOM.

5. Finally, to find the future disaster-risk analysis, an inundation was investigated through the dangerous cyclone track along the east zone of Bangladesh coast.

Chapter 2

Chapter2

Cyclone activity analysis under present and future climate condition

2.1 Chapter summary

This chapter discusses the cyclone activity analysis along the coast of Bangladesh. We have discussed the seasonal landfall impact, power dissipation index, accumulated cyclone energy, and the genesis location for the present and future cyclone of Bangladesh. This analysis was performed from the d4PDF, AGCM, JTWC, and Bangladesh Meteorological Department (BMD) data. The occurrence behavior at the pre-monsoon and monsoon season was also investigated. The bilateral relation was found for the seasonal activity of a tropical cyclone of Bangladesh. It is found that the future occurrence probability of tropical cyclone along the coast of Bangladesh will be increased. The seasonal activity shows the autumn (September, October and November) season is the prone season for cyclone formation along this region.

2.2 The data

In this dissertation, we have used the Meteorological Research Institute (MRI)- Atmospheric Global Climate Model (AGCM), Database for Policy Decision- Making for Future Climate Change (d4PDF), Joint Typhoon Warning Center (JTWC), and Bangladesh Meteorological Department (BMD). The MRI-AGCM and d4PDF data is the model-simulated data for present and future climate scenario. The BMD and JTWC data are the best track data or we called observed data. At first, Williamson et al. (1987) develop a NCAR Community Climate Model version 1 (CCM1) which is now known as an Atmospheric Global Circulation Model (AGCM). In this model, heat momentum, dynamic mass were used to the horizontal spectrum conversion method where 18 vertical levels were used to adjust the system. In each layers, the delta-Eddington calculation was performed which was developed from the solar radiation scheme of Thompson et 8

Chapter 2

al. (1987). In the present situation, the Meteorological Research Institute (MRI) and Japan Meteorological Agency (JMA) jointly develop a new operational numerical weather prediction model known as MRI-AGCM (Mizuta et al., 2006). For the further development, a semi Lagrangian three-dimensional advection scheme was used, which is accelerating the time integration with 20 Km horizontal grid and 60 level altitude 0.1 hPa vertical grid spacing (Yoshimura and Matisumura, 2005). The higher resolution (20 Km) AGCM experiment was performed by time-slice method, which has two layers. One is the global warming projection system that consists with an Atmospheric-Ocean General Circulation Model (AOGCM), and the higher part of the vertical level AOGCM generated by AGCM. For the further development of the climate experiment, the database for policy decision making for future climate change (d4PDF) data was revised for global experiment data. The d4PDF data was produce from the weather prediction model of Japan Meteorological Agency (http://www.jma.go.jp/jma/jma- eng/jmacenter/nwp/outline-nwp/index.htm) with the modified model of MRI- AGCM 3.2 (Mizuta et al. 2012). The data was 60 km grid for global simulation and 20km grid for Japan area. The developed model used triangular truncation with a linear Gaussian grid (TL319) and the 64 vertical levels with the top at 0.01 hPa (Imada et al. 2017). This clear that the model wave number is 319 and the grid resolution is 60 km in horizontal level and 64 for vertical level. The boundary conditions of the model was sea surface temperature (SST) and sea-ice concentration (SIC, and sea-ice thickness (SIT) for lower boundary. The external forces were considered as Global-mean concentrations of greenhouse gases, three- dimensional distributions of ozone and aerosols (Mizuta et al. 2012). The data were collected from the Data Integration and Analysis System (DIAS) under the Global environmental information integration program (Kitamoto et al., 2009). The data is divided in three features with past experiment (1951 to 2011), Non-Global warming experiment (1951 to 2010), and 4o C rise experiment (2051 to 2110). In addition, the experimental result of global atmospheric-ocean coupled model contributes to the Coupled Model Inter comparison Project Phase 5 (CMIP 5). There are 100-member simulation for present climate condition, and 90-member simulation for future climate condition. In the future climate simulation, there are six different ensemble simulation model of CC (CCSM4), GF (GFDL-CM3), HA

Chapter 2

(HadGEM2-AO), MI (MIROC5), MP (MPI-ESM-MR), and MR (MRI-CGCM3). The authorities are responsible for the future-climate model simulation. The authorities are National Centre for Atmospheric Research (USA), NOAA Geophysical Fluid Dynamics Laboratory (USA), Met Office Hadley Centre (UK), AORI, NIES, JAMSTEC (Japan), Max Plank Institute for Meteorology (Germany), and Meteorological Research Institute (Japan). Different SST condition is used for different climate simulation. Fig. 2.1 represents the SST pattern (Mizuta et al. 2017). Monthly mean inter-annual variation of SST named Centennial Observation-Based Estimates of SST version2 (COBE-SST2) and the sea-ice concentration with +4k Global Warming condition are the radiative forcing of the model (Hirahara et al. 2014). For the analysis of cyclone, Oouchi et al. (2006) developed a tropical cyclone (TC) detection model. After summarized the technique and adopted the new definition of cyclone, Murakami et al. (2012) developed the model. The assumption of their study was: the vorticity greater than 8.0x10-5 s-1 , the warm core at 300, 500 and 700 hPa exceeds 0.8oC, maximum wind speed greater than 13 ms-1 at 850 hPa, and the duration of life-time longer than 36 hours.

Fig.2.1 Monthly mean inter annual variation of SST (Ministry of Education, Culture, Sports, Science and Technology, et al., 2015)

Chapter 2

2.3 Activity analysis

Bangladesh has six seasons. The seasons depend upon the latitude change of earth axis relative to sun. However, the cyclone season depends upon the occurrence behavior in each month. According to the climate conditions, Bangladesh is mainly subtropical monsoon country. The seasons are the summer season (mid-April to mid-June), the rainy season (mid-June to mid-August), the autumn season (mid- August to mid-October), the late autumn season (mid-October to mid-December), the winter season (mid-December to mid-February), and the spring (mid-February to mid-April). To understand the cyclone seasonal behavior, we have investigated the cyclone formation analysis along the Bay of Bengal. The present scenario cyclone eye point of AGCM, d4PDF (member 1), and BMD is shown in the Fig.2.2. This figure represents the eye point position of a cyclone in each 6-hour interval.

Fig.2.2 Cyclone eye point for present climate condition

In addition, the pattern of future cyclone path is shown in the Fig 2.3 for the AGCM future climate condition.

Chapter 2

Fig.2.3 Cyclone activity for the future climate scenario of AGCM

Researchers are worried about the future climate and its changes due to the global warming and ice concentration. The MRI-AGCM present data can explain the present situation well because the MRI-AGCM is forced with observed SST, but the model input SST from CIMP3 has some biases. For the 21st century (2075– 2099, 25 years), the global simulation under the consideration of future scenario MRI-AGCM3.2H model forecast the future climate. At the first step, the Couple Model Intercomparison Project 3 (CMIP3) multi-modeled data set was used to project the Multi-Model Ensenble (MME) of SST. The second stage represents the linear trend in MME of SST that projected MIP3 data set. At the third step, a difference between 20th century experiment of Intergovernmental Panel on Climate Change (IPCC) fourth assessment report (AR4; IPCC 2007) and future simulation under the Special Report on Emission Scenario (SRES) A1B emission Scenario (IPCC 2000) in MME of SST. In addition, to simulate the future

Chapter 2

scenario, four different multi-model projected SST distributions were considered. The internal variability of the climate system was calculated by the two distinct initial conditions, which were compared with two different simulation. So, the future simulation output was four SST distributions with two initial conditions, 24 ensemble experiments with three convection schemes, and 25-year integration period of simulation. The four SST distributions were obtained from the projected change of SST which was collected from the cluster analysis of multi-model projected changes SST of CIMP3. The future prejection M, C1, C2, and C3 represents a mean of future SST projections. In this projection system, the SST range consider from 1.95°C to 2.14°C (Reinhardt E, 2017). The projected pattern of SST is similar as the Pacific Ocean and the Caribbean Sea. The SST pattern in the Caribbean Sea was large to small and the Pacific Ocean is different, from small to large. The Bay of Bengal cyclone information for the different scenario of C0, C1, C2, and C3 are represented in the above figure (see, Fig.2.3). From the above figure, it is evident that the probable cyclone eye point was found near the east coast of India. The cyclone activity information of the Bay of Bengal for the different scenario of C0, C1, C2, and C3 can represent the cyclone formation probability along the Bay of Bengal for the 21 century. From the above figure (see, Fig.2.3), it is evident that the cyclone probability was found near the east coast of India. However, some cyclone was also found near the east part of Bangladesh coast for the C0 and C3 scenario, and the west part of Bangladesh for C1 and C2 scenario. For the detail analysis of future cyclone activity, we have analyzed the d4PDF data. Fig. 2.4 represents the cyclone activity of 90-member ensemble simulation for six scenarios of future climate condition. Most of the cyclone was found near the east coast of India in each scenario except MI scenario. In the MI experiment, some cyclone was found near Bangladesh, India and Sri-Lanka. Most of the cyclone of Bay of Bengal forms in the CC scenario. It is observed that Bangladesh coast faces some cyclone in different scenario. Most of the cyclones that strike in the Bangladesh coast are found in the CC, GF, MI, MP, and MR scenario. This figure only shows the cyclone path, which can explain the general behavior of cyclone and the probability of cyclone in each scenario. To understand the cyclone behavior properly, we have investigated the cyclone genesis point in each scenario.

Chapter 2

The data that I have used in my study are represented in the table. The table 2.1 represents the data properties and the table 2.3 shows the scenario details. In this study a huge number of dataset is analyzed to investigate the present and future features of cyclone that changed by the climate change.

Table 2.1 exercised data in this study d4PDF MRI-AGCM JWTC BMD Present Future Present Future Present Present 100 90 member’s Present C0, C1, Observed (Best Observed member observed C2, and C3 track data) (Best track ensemble Six scenario & data) (1.95°C to simulation CC, GF, HA, 1951-2011 Present 2.14°C MI, MP, and simulated (6000 MR, 2051- simulation) 2111 (5400 simulation)

Table 2.2 details of future six scenarios

CC=CCSM4 National Centre for Atmospheric Research (USA) GF=GFDL-CM3 NOAA Geophysical Fluid Dynamics Laboratory (USA) HA=HadGEM2-AO Met Office Hadley Centre (UK) MI=MIROC5 AORI, NIES, JAMSTEC (Japan) MP=MPI-ESM-MR Max Plank Institute for Meteorology (Germany) MR=MRI-CGCM3 Meteorological Research Institute (Japan)

These six scenarios are based on six SST warming patterns that clustered in each ocean area from RCP 8.5. Every SST warming pattern was scaled so that the simulated global-mean surface air warming from the pre-industrial level is equal to 4 K.

Chapter 2

Fig.2.4 Cyclone activity of 90-member ensemble simulation for six scenarios

2.3.1 Genesis analysis

I have considered the initial detection point of tropical disturbance as a genesis point. Genesis is an important characteristic for cyclone movement behavior. Due

Chapter 2

to climate change, the genesis location may change. There are some factors depend on the cyclone genesis occurrence. The accountable to the factors are the small magnitude vertical wind shear of horizontal wind, low-level relative vorticity, and a large magnitude of thermal energy. However, Leipper and Volgenau (1972), suggested that the SST and high magnitude ocean thermal energy is the key factor for the existence of cyclone. In the earlier 19th century, Dunn (1940) observed that the earth rotation near the equator is an important factor for developing a genesis in the upper part of equatorial area. The future projection of geneses point along the Bay of Bengalis presented in the Fig.2.5. This figure shows the present climate genesis location along the Bay of Bengal. It is observed from the figure that the cyclone genesis location was found near the middle and east coast of India for present climate situation.

Fig.2.5 Cyclone genesis location for present climate condition

Chapter 2

However, in the future climate condition the genesis location may change. Because, in the 4o-5o latitude band the wind are very weak. Therefore, there is no possibility to create the cyclone genesis area near the 5o equatorial belts. In the present research, we have found the key factors of cyclone genesis, which are Coriolis parameter, low-level relative vorticity, sea surface factor like SST>26o C in between ocean surface to 60 meter higher position, change vertical gradient with respect to pressure is 500 hPa, and the humidity at the middle troposphere. The role of low-level relative vorticity is important for cyclone genesis. The role of low-level relative vorticity is important for cyclone genesis. A continuous import of mass, water vapor, and momentum are needed to develop a tropical cyclone.

Fig.2.6 Cyclone genesis location for future climate condition of AGCM data

Chapter 2

Fig.2.7 Cyclone genesis location for future climate condition of d4PDF data

From the report of Sartor (1968) and Wachtmann (1968), the tropical cyclone forms in the region where the large position of low level vorticity makes. So, all the factors are depending upon the local climate condition. To understand the cyclone genesis behavior due to climate change condition in different scenario, we have extracted the cyclone genesis point from the original cyclone track data. Fig.

Chapter 2

2.6 shows the genesis location for the AGCM future climate condition. Similarly, Fig.2.7 shows the cyclone genesis for six different scenarios of d4PDF future scenario data. To understand the probability of the genesis location of each (2o x2o) latitude longitude area, we have investigated the genesis probable area. Fig.2.8 shows the probability area of cyclone genesis in each two-degree latitude longitude area. From this analysis, it is found that the area of cyclone genesis probability will be shifted in the near future. Fig. 2.8 also shows the cyclone genesis probability of present climate condition. Color bar indicates the genesis probability in each individual area. In the present climate condition, the genesis area is found in the middle of the Bay of Bengal. But, in the near future, it will be shifted to the upper latitude and the maximum probability of genesis location would be found near the east coast of India. We have found some genesis area outside the Bay of Bengal area due to select the first point of depression as a genesis point. Some genesis area also found outside of interested area due to select a big cyclone track separate region. However, the study gives the positive summary about the genesis probability. Fig. 2.9 shows the cyclone genesis probability of future climate condition of d4PDF data. Most of the scenario shows the same genesis probability area except MI scenario. So, it is evident that the climate change has an impact on cyclone genesis. Due to the global warming, future climate will change and its associated cyclone genesis will also change. Fig. 2.10 shows the future genesis prone area for the different scenario of AGCM. From the above analysis, I have seen that the east coast of Indian region and its adjacent area is the most genesis prone area. Due to the climate change, the genesis area will be changed along this region.

Chapter 2

Fig.2.8 Genesis probable area of present climate condition

Fig.2.9 Genesis probable area of d4PDF future climate scenario

Chapter 2

Fig.2.10 Genesis probability area of AGCM future climate scenario

2.3.2 Occurrence analysis a) Yearly and monthly activity

I have investigated the occurrence of Tropical cyclone (TC's) within the Bay of Bengal. The overall occurrence behaviors of the Bay of Bengal are investigated for yearly occurrence, monthly occurrence. In both cases, I have investigated for present climate condition and future climate condition. The yearly occurrence of cyclone is represented the formation number of cyclone in a year. In this analysis, I have investigated the present climate for 1979 to 2003 cyclone information. We have collected the cyclone information for each scenario of JTWC, BMD, AGCM and d4PDF. After analyzing the cyclone information, I have found that the yearly average number of cyclone occurrence is 4-7 along this region. Fig. 2.11 show the yearly average number of cyclone occurrence in present climate condition and future climate condition. From this figure, it is found that the occurrence fluctuation is lower for d4PDF scenario than the AGCM future scenario and the

Chapter 2

future cyclone occurrence will be decrease along this region. To understand the yearly occurrence more clearly, Fig.2.12 shows the each scenario average occurrence number in each year of present and future climate condition.

Fig. 2.11. Yearly average number of cyclone occurrence for present and future climate scenario. The upper left and lower left figure (a)(c) shows the yearly occurrence number in present scenario. The upper right and lower right figure (b)(d) represents the future scenario

Chapter 2

Fig.2.12 Average occurrence number of each scenario in each year for present and

future climate condition

So, it is evident that the future cyclone occurrence will be decrease along this region. The box plot or whisker diagram is a standardized way of displaying the distribution of data based on some characteristics. The lower and upper part of the box spans represents the first quartile to the third quartile. The segment line between the two parts of the box shows the median number of cyclone events and whiskers above and below in the box shows the maximum and the minimum number of cyclone events. From the above analysis, it is found that the yearly occurrence number of cyclone is 4 to 7 within the Bay of Bengal. After analyzing the yearly occurrence of cyclone event, I have analyzed the monthly and seasonal activity. The above study is not the brief assumption of occurrence behavior. Therefore, for this region, I have investigated the cyclone occurrence behavior in each month and checked the seasonality.

Chapter 2

Fig.2.13 Monthly average of cyclone occurrence. (a) Present climate scenario, (b) future climate scenario, (d) Average of Present climate scenario data, (d) Average future climate scenario data

From the Fig.2.13, I have found that the monthly occurrence of present climate cyclone has a periodic nature. However, the future occurrence has no such like activity. The present cyclone occurrence has seasonal impact, it show the higher occurrence probability at the post-monsoon season (October-December). But, in the future climate condition, it will be changed. To find the present and future occurrence with the inter variability of each scenario; I have analyzed the data to find the general behavior of monthly occurrence. Fig. 2.14. can explain the general behavior of the monthly occurrence of tropical cyclone along the Bay of Bengal. The monthly average occurrence number is 0.6 to 0.8. The occurrence behavior is high for the post monsoon months but the instability is very high. For the future scenario, the monthly occurrence behavior will be changed.

Chapter 2

Fig.2.14 Monthly Mean occurrence. (a) Present climate scenario, (b) future climate scenario b) Seasonal activity

The seasonal activity analysis explain the cyclone events of the pre-monsoon, monsoon and post monsoon which can make an important role in cyclone activity analysis along this region. Due to seasonal diversity, the effects of the cyclone in the Bay of Bengal region are also different. For the Bangladesh and the west part of Indian region, there are six seasons usually observed, for example, Summer (mid-April to mid-June), rainy season (mid-May to late-October), Autumn (mid- August to mid-October), Late autumn (mid-October to mid-December), Winter (mid-December to mid-February) and Spring (mid-February to mid-April). In this study, we have divided the three seasons of cyclone, which are pre-monsoon, monsoon and post monsoon respectively. The pre-monsoon season consider for January to April, monsoon season consider for May to August and the post monsoon season consider for September to December. It is found that, the cyclone occurrence behavior has an impact on seasonal activity. The post monsoon season is the prone season for cyclone occurrence in this area. The seasonal occurrence behavior under present and future climate condition shows in Fig. 2.15. Different scenario shows the same seasonal activity of cyclone occurrence for this region. But, in the future the present occurrence behavior of cyclone will be changed due to climate change. The changes were the occurrence number and seasonal occurrence probability.

Chapter 2

Fig. 2.15 Seasonal occurrence behavior. (a) Present climate condition, (b) Future climate condition c) Occurrence behavior at El Niño and La Niña

To understand the tropical circulation pattern effect of cyclone occurrence behavior, El Niño and La Niña impact has been investigated. It is found that the occurrence behavior of ensemble data that generated by the model simulation of MRI-AGCM and d4PDF, and the observed cyclone occurrence behavior shows the different characteristics. Due to El Niño, the cyclone occurrences is higher than La Niña, but for the moderate case of La Niña shows higher occurrence than El Niño. The occurrence behavior of cyclone at El Niño and La Niña climate pattern are explained in the Fig. 2.16. The El Niño and La Niña are the opposite face of climate pattern. In this study, it is found that the El Niño and La Niña has a negligible impact on cyclone occurrence behavior.

Fig. 2.16 Cyclone occurrence behavior at El Niño and La Niña climate pattern

Chapter 2

2.3.3 PDI and ACE analysis

Bangladesh Meteorological Department classified the tropical cyclone based on the maximum sustained wind speed and lowest central pressure for warning and evacuation purpose. Basically, the intensity of the storm depends on the velocity of the wind during the storm. The bulletins of cyclone warning are issued by the BMD based on the specified cyclone classification. The detail classification is shown in table 2.3.

Table 2.3 Cyclone category along the Bay of Bengal

Serial Category Maximum wind speed (in knots) 1. Low (L) <10kt 2. Well Marked Low (WML) 10-16kt 3. Depression (D) 17-27kt 4. Deep Depression (DD) 28-33kt 5. Cyclonic Storm (CS) 34-47kt 6. Severe Cyclonic Storm (SCS) 48-119kt 7. Severe Cyclonic Storm with a core of >120kt Hurricane Winds (SCS (H))

To understand the cyclone activity, it is important to analyze the Power Dissipation index (PDI) and Accumulated Cyclone Energy (ACE). Tropical cyclone energy of different ocean basin is different, because of the lifetime of a cyclone. The PDI and ACE of Bay of Bengal storm for present and future climate conditions are calculated from the relation of Emanuel (2005). For this analysis, track archives from MRI-AGCM and d4PDF are used.

 PDI 2 v3 dt (2.1)  max 0

The ACE can also be obtained from the relation of Emanuel (2005) as

 ACE v2 dt (2.2)  max t0

where vmax is the maximum sustained wind speed and  is the total duration of the cyclone. The square velocity of maximum sustained wind is the kinetic energy of a cyclone. The total summation of kinetic energy of a cyclone represents the

Chapter 2

accumulated cyclone energy of a cyclone. This cyclone energy (CE) approximates the wind energy, which can calculate every six hours by a tropical system over its lifetime. This analysis can explain the destructiveness of a cyclone for the specified ocean basin. Stronger cyclone may more destructive in this region. The overall analysis of ACE for AGCM and d4PDF is presented in the table 2.5 and table 2.5

respectively.

Table 2.4 Detailed summary of accumulated cyclone energy from MRI-AGCM data AGCM present (1979- AGCM future (2075-2099) 2003) Year ACE Year ACE (kn2) 2 (kn ) C0 C1 C2 C3 Mean 1979-1983 80.20 2075-2079 0 0 6.38 129.55 33.98 1984-1988 39.57 2080-2084 24.41 35.04 3.67 1.06 16.04 1989-1993 36.22 2085-2089 50.42 27.65 10.21 19.1 26.84 1994-1998 55.25 2090-2094 15.91 0 12.58 0 7.12 1999-2003 67.23 2095-2099 7.98 11.87 62.03 0 20.47

Table 2.2 represents the ACE of present and future cyclone, which is calculated from the MRI-AGCM data. Consecutive five years interval cyclone is used to calculate the ACE, which is presented in the Table 2.3. Both the present and future climate scenario of AGCM, it is found that the ACE of future cyclone is lower than present cyclone ACE. But, the d4PDF data shows different results.

Table 2.5 Detailed summary of accumulated cyclone energy of d4PDF data

D4PDFpresent (1979- d4PDF future (2075-2099) 2003) Year ACE Year ACE (kn2) 2 (kn ) CC GF HA MI MP MR 1979-1983 9.73 2075-79 29.15 27.39 9.07 2.83 9.26 14.7 1984-1988 14.03 2080-84 18.55 14.56 8.02 6.38 6.01 25.0 1989-1993 10.67 2085-89 23.49 17.38 25.50 6.66 11.7 24.3 9 1994-1998 7.44 2090-94 10.43 11.60 13.60 6.28 5.84 18.4 1999-2003 13.95 2095-99 15.3 10.64 18.32 4.66 4.11 16.1 Table2.5 shows the ACE analysis result of d4PDF data with six different climate scenarios.

Chapter 2

For the future climate condition, the data were reviewed for the different cases as like CC, GF, HA, MI, MP and MR.

Fig.2.17 PDI analysis of present and future climate condition

It is found that the accumulated cyclone energy of a cyclone will be increased in future. The PDI and ACE of Bay of Bengal cyclones at present and future clearly identify that the Bay of Bengal cyclone has sufficient oceanic thermal energy, which conducive for the intensification of cyclone in this region. PDI metrics of present and future cyclone in the Bay of Bengal during 25 years (see, Fig.2.17). Finally, it is found that the future cyclone occurrence will decrease in this region but the accumulated cyclone energy will be increased, which represents the strongest cyclone formation probability. Thus, it is important to investigate the cyclone behavior properly within this region to understand the future probability of cyclone characteristics.

2.4 Conclusions

This study investigates the cyclone activity along the Bay of Bengal region. From this study, it is observed that the constructed storm in the Bay of Bengal has various types of behavior. Cyclone genesis will be shifted to the upper latitude and the number of occurrence will decrease in the future climate condition. Bay of Bengal cyclone has seasonal effects. Present climate shows that the post-monsoon season is the prone season for cyclone occurrence. But, due to the climate change, the seasonal behaviour of cyclone occurrence will be changed in future. The cyclone occurrence number will decrease in future but the accumulated cyclone energy will be increase. This represents the future intense cyclone formation in this region.

Chapter 3

Chapter3

Analysis of future tropical cyclone characteristics in the Bay of Bengal

3.1 Chapter summary

Tropical cyclone characteristics depend upon some basic parameters of a cyclone. The well-known parameters of a tropical cyclone are landfall angle, maximum sustained wind radius, translation speed, central pressure, wind velocity and so on. In this chapter, a cyclone strikes behavior associated with the landfall angle, maximum sustained wind radius, translation speed, central pressure has been investigated at the present and future climate condition. To find the future disaster risk due to the changes of the tropical cyclone characteristics is important. The changes of a cyclone characteristics due to the climate change may have influence on surge height. Therefore, I have investigated the precise summary of the characteristics change of a cyclone due to climate change.

3.2 Basic assumptions

To investigate the characteristics of a cyclone parameter, I have considered some assumptions. In the chapter two, subsection 2.3.3, BMD categorized the cyclone for warning and evacuation purpose. In this chapter, for this study, I have considered the cyclone in two different categories; (i) low-intense cyclone and (ii) intense cyclone. The low intense cyclone has the central pressure limit between 1013 hPa to 970 hPa, and the intense cyclone central pressure is below 970 hPa. In this study, I have separated the cyclone in such a way that the central pressure between 1013 hPa to 970 hPa is considered as low-intense cyclone. On the other hand, if a cyclone has the central pressure below 970 hPa in its life time (Genesis to landfall) then I consider the cyclone as an intense cyclone. I have also separated the cyclone track for each country area like Bangladesh, Sri-Lanka, Myanmar, and India to investigate the track features properly. According to the studies of Rao et

Chapter 3

al. 2010 and Jain et al. 2010, cyclone track information is separated in each country. If the landfall occurred in the specified region then I consider the track for that region. Therefore, the sample size of cyclone parameter is different in each region. After that, the landfall angle was calculated from the study of the calculation process of azimuth distance. Those storms hit in each country of the study area are used to analyze the landfall properties of that country. So, I considered the strike cyclone whose consecutive two eye point is sea and land. If this situation occurs in a specific country region then I have considered the cyclone strike in that country. The strike rates are calculated by the ratio of the number of strike cyclone and the number of total formation. The actual number of cyclone is also calculated from the ratio of the number of strike cyclone and the number of simulation year.

n(Strikecyclone) Strike Rate(%) x100 (3.1) n(Totalformation)

n(Strikecyclone) Actualnumber  (3.2) n(Simulation year) Here the "Strike cyclone" means that the cyclone has landfall and the "Total formation" means the total number of cyclone formation without landfall and with landfall. In the equation 3.2, the term "Simulation year" represents the total number of year for data generated. In this study, I have used d4PDF data, which has 60 years simulation data.

3.3 Cyclone strike analysis in the Bay of Bengal

A high attention is needed for the government of Bangladesh to concentrate the coastal area to make a useful plan to preventing the coast from future cyclone disaster risk. This study may help to take the overall concept of future-cyclone strike information along the Bay of Bengal coast. Bay of Bengal coast covers the countries Bangladesh, India, Sri-Lanka, and Myanmar. Fig. 3.1 shows the strike rate of all cyclones in the Bay of Bengal. Figure (a) shows the cyclone strike rate in different climate scenarios and the Figure (b) represents the changes of the strike rate. The change rate was calculated from the difference of future and present strike rates. In the similar way, actual number of cyclone and its changes in

Chapter 3

Fig. 3.1 Strike rate and the change of strike rate different scenarios are presented in the Fig. 3.2. From both the figures, it is evident that the strike rate will be increase in the Bay of Bengal but the actual number will be decrease.To understand clearly the strike rates in different country, I have investigated the cyclone strike analysis from those cyclones which landfall occurred in different coast of different country. Finally, I have found that the total number of formation 21,862 in the Bay of Bengal but the number of strike cyclone only 11,110 in the Bay of Bengal coast. This figure shows that the actual number is decreasing in future scenario. From the all scenarios, the MI scenario shows the lowest actual number of strike cyclone. Thus, I think that the actual number of strike cyclone will be decrease in future climate.

Fig. 3.2 Actual number and the change of actual number

Chapter 3

From the formation number and strike rate analysis, it is found that the 50% cyclone does not strike the coast. This is the overall fact of all cyclones in the Bay of Bengal coast. To investigate the strike rate in each country, it is important to separate the strike cyclone for each country. To make clear investigation, the intense and low-intense cyclone behavior is investigated. The statistical information is given in Table 3.1

Table 3.1 Number of cyclone strike in different country area. The strike numbers are divided into two criteria: intense and low intense Present CC GF HA MI MP MR

Total 21862 2622 2563 2382 825 1934 2332 Track

Sri-Lanka Strike 412 17 28 15 13 13 4 Track

Low- 412 17 28 15 13 13 4 intense

Intense 0 0 0 0 0 0 0

India Strike 7927 1098 1105 1029 277 588 1011 Track

Low- 7417 980 1033 939 245 541 920 intense

Intense 510 118 72 90 32 47 91

Bangladesh Strike 952 154 107 93 51 89 110 Track

Low- 803 114 86 76 37 67 82 intense

Intense 149 40 21 17 14 22 28

Myanmar Strike 1819 210 152 186 43 84 167 Track

Low- 1762 189 149 179 40 81 163 intense

Intense 57 21 3 7 3 3 4

Chapter 3

From the strike cyclone, I have studied the strike rate for intense and low-intense cyclone in each scenario for different country. Fig. 3.3 explains the intense and low intense cyclone strike behavior in each country region.

Fig.3.3 Strike rate in different coast of different countries

From this analysis, I have found that the strike rate of intense cyclone will be rise in future. From this analysis, it is found that strike rate of intense cyclone will be higher in this region and found to be higher than other countries. Actually, this analysis shows the strike rate in each country from their strike cyclone. Both the intense and low-intense cyclone strike information in different scenario is also shown in the figure. But, if we investigate the strike number from all formation of cyclone in the Bay of Bengal then we found that the cyclone infected country is India. The reason may be the coastal

Chapter 3

Fig 3.4 Strike rate in each country from all of cyclones on BoB

Chapter 3

length and the geographical location. That's why I have investigated the strike rate in each cases in each country coast. Fig 3.4 shows the strike rate from all cyclone formation in each scenario of each country. From the information of BMD and Indian Meteorological Department (IMD), the coastal length of different countries in the Bay of Bengal are as follows; Bangladesh: 710 Km, India (East coast): 3,480Km, Sri-Lanka: 1,340km, and Myanmar: 1,930 Km. The coastal length based strike rate analysis can make clear idea of strike rate along this region.

Fig 3.5 Coastal length based strike rate

Chapter 3

Fig 3.5 shows the coastal length based strike rate analysis. Thus, though the low- intense and intense cyclone found the east coast of India but the intense cyclone strike rate will be higher in the Bangladesh coast.

3.4 Cyclone strike analysis in Bangladesh coast

From the above discussion, I have found that Bangladesh is the prone country for the risk of intense cyclone. So, I have investigated the cyclone risk in different part of Bangladesh coast. To understand the cyclone activity in the coast of Bangladesh, I have separated the coast into three different parts. Each of the coasts has its own characteristics. The east part is economic zone, middle part is low land zone, and the west part is mangrove forest zone. The physical difference of the three different part of Bangladesh coast is known as the eastern part, middle part and western part (Khan and Kawasaki, 2016). Fig. 3.6 shows the different part of the coast of Bangladesh. In a similar way, I have separated the cyclone track information in each part of Bangladesh coast. The longitudinal difference of each part of the coast is same for east, middle, and west zone. The eastern zone has some coastal protection but there is no coastal protection in middle and western part.

Fig.3.6 Different parts of coastal area of Bangladesh. The upper figure shows the coastal protections of the coastal areas

Chapter 3

The middle part is low-level land and the west part has a salt marsh tidal halophytic mangrove forest. The mangrove is known as Sundarbans. According to the report of the daily newspaper (the daily star), about 60% trees are in Sundarbans whereas Bangladesh has 17.3% forest area. So, it is important to protect the mangrove due to future cyclone risk. Though the west and middle parts of Bangladesh coasts are also prone for the intense cyclone, this research concentrates the east part of Bangladesh coast. Because, the east coast will faces most intense cyclone in future. Further, there are 10 million Rohingya Settlement Island is in this eastern zone. Bangladesh government recently developed the island (Bhasan Char) to settle the refuge of Myanmar (Rohingya). So, it is important to concentrate the development of the coastal structure and disaster prevention policy along this area. This part is also important for the future-disaster prevention plan due to the development of the coastal structure and future safety of Rohingya. I have separated the cyclone tracks for each coastal zones of Bangladesh. The cyclone tracks are separated for intense, low-intense category in six different climate scenarios. Total track represents the sum of intense and low- intense cyclone strike number. The detail strike information is given in Table 3.2. Finally, I have calculated the strike rate in each individual coastal area. The strike rates are shown in Fig. 3.7. From this figure, it is evident that the west part of Bangladesh coast will face most of the cyclone in future.

Table 3.2 Number of cyclone strike in different part of Bangladesh coast

Present CC GF HA MI MP MR Total All 952 154 107 93 51 89 110 Track Low In. 803 114 86 76 37 67 82 Intense 149 40 21 17 14 22 28 East Strike Track 296 42 25 23 15 18 24 Part Low-intense 253 31 20 17 11 15 17 Intense 43 11 5 6 4 3 7 Middle Strike Track 252 34 34 21 17 20 36 Part Low-intense 203 26 26 20 14 12 26 Intense 49 8 8 1 3 8 10 West Strike Track 404 78 48 49 19 51 50 Part Low-intense 347 57 40 39 12 40 39 Intense 57 21 8 10 7 11 11

Chapter 3

But, it is important to find the parameter behavior of a cyclone in each part of Bangladesh coast to find the general behavior of a cyclone.

Fig. 3.7 Cyclone strike characteristics in each part of Bangladesh coast

3.5 Characteristics of cyclone parameter

3.5.1 Central pressure

One of the most important characteristics of a cyclone is central pressure. To find the intensification, wind speed, and the category of a cyclone central pressure makes an important role. Thus, I have investigated the mean, standard deviation, maximum and minimum central pressure of a cyclone in each region of Bangladesh coast. The central pressure of a cyclone in different climate scenarios is also investigated. The strike cyclone of each coastal zone in Bangladesh has different characteristics of central pressure. The detailed central pressure behavior

Chapter 3

of low-intense and intense cyclone in different scenarios for every part of the coastal zone is given in Fig. 3.8.

Fig. 3.8 Central pressure behavior of different coastal zone of Bangladesh

From this figure, it is evident that the central pressure for low-intense cyclone is stable but for the intense cyclone, it is unstable. West part cyclone is more unstable than east part. The mean central pressure of each scenario for east is lower than the middle and west part. I have found that the western parts of the Bangladesh coast will more venerable due to the intense cyclone in future. But, most of the dangerous and catastrophic cyclone will strike on the eastern coast of Bangladesh. Because, the cyclone of lowest central pressures are found in this region more than the western part. So, it is important to concentrate the development of the coastal

Chapter 3

structure and disaster prevention policy along this area. Though this part is very important for future disaster-prevention plan due to the development of the Bangladesh economy and economic zone, the present development did not meet the expectation. Moreover, there is no proper investigation of future disaster risk along this region.

The central pressure of the strike cyclone in each part of the coastal area of Bangladesh shows different characteristics under present and future climate condition. In the scenario of CC, GF, HA, MI, MP and MR, the central pressure of a cyclone is unstable for intense cyclone but the low-intense cyclone represents the stable changes due to climate change. The mean and standard deviation uncertainty of cyclone central pressure for present and future scenario is shown in the figure (see Fig.3.8). From this study, it is evident that the eastern part of Bangladesh coast faces dangerous cyclone but most of the cyclone strike western part of Bangladesh.

3.5.2 Landfall angle

Landfall angle is an important factor for disaster risk analysis. The landfall angle may have an influence in the surge height along this region.

Fig.3.9 Schematic figure of before and after landfall point

Due to the climate change impact, the cyclone landfall behavior also changed. For the different climate scenario, I have observed the characteristics of cyclone landfall angle. The overall characteristics of landfall angle depend on the coastal structure. Landfall angle was calculated from the study of azimuth distance. To find the landfall angle, the consecutive two points of a cyclone eye are needed,

Chapter 3

which are in sea and land. Then the following relation is used to calculate the landfall angle.

Landfallangle( )  a tan(x2  x 1 , y 2  y 1 ) (3.3)

where, landfall angle, (x11 , y )  latitude and longitude before landfall,

(x22 , y )  latitude and longitude after landfall.

The polar histograms of landfall angle are calculated for both the low-intense and intense cyclone. For the different climate scenario, I have also observed the characteristics of cyclone landfall angle.

Fig.3.10 Polar histogram of landfall angle for low-intense cyclone

Chapter 3

Fig.3.11 Polar histogram of landfall angle for intense cyclone

Fig.3.10 and Fig.3.11 show the polar histograms of low-intense and intense cyclone landfall angle. In this study, it is found that the landfall angle would be shifted northeast to north-west direction in the future climate scenario along the coast of Bangladesh. The landfall angle of low-intense cyclone has some uncertainty in the west part of Bangladesh than intense cyclone. It is found that the landfall angle is same or stable for the intense cyclone. The polar histogram gives the detail information about the landfall angle in different parts of Bangladesh coast in different climate scenario for present climate and the future CC scenario. 3.5.3 Maximum sustained wind radius

Maximum sustained wind radius is an important parameter of a tropical cyclone. This parameter depends upon the central pressure and wind velocity. To explain the tropical cyclone, the tropical cyclone is considered a function of some parameters like wind speed, central pressure, maximum sustained wind radius etc. The maximum sustained wind radius has great impact on surge height. Due to the global warming, the maximum sustained wind radius of a cyclone may change. To understand its nature in the Bangladesh coast, I have investigated the cyclone information for the six different climate scenarios. From the study of Hsu and Yan,

Chapter 3

(1998) the mean Rmax was 47 km for the central pressures of 909–993 hPa in 1893– 1979. Fujii (1998) found that the typhoon with central pressure ≤ 980 hPa that hit the Japan main islands has the Rmax of 84–98 km. The maximum sustained wind radius Rmax is depending on the cyclone track and the characteristics of each cyclone. Therefore, some scientist proposed several estimation models for Rmax (see, Kossin et al. (2007); Quiring et al. (2011)). However, the maximum wind velocities are differently defined depending on different ocean basin where the tropical cyclone transits. Different countries classified the sustained wind in different consideration. For instance, 10 min maximum sustained wind speed was considered for Japan Meteorological Agency (JMA) and 1 min for United States National Weather Service (NWS). Some well-defined empirical relationship developed by some scholars (see, e.g., Kawai et al., 2005, Takagi et al., 2012; Nakajo et al., 2014;). In this study, I have calibrated the cyclone track data under present and future climate condition to find the appropriate empirical relation. The modified empirical relation from Kawai et al., 2005 is

1  0.644 Vmax Pc  1013  (3.4) 13

Pc  967  61.5 Rmax  94.89*exp (3.5)

where, Pc =Central pressure (hPa); Vmax = maximum velocity (m/s); Rmax = maximum sustained wind radius (km).

Fig.3.12 shows the mean and standard deviation of maximum sustained wind radius for the different part of Bangladesh coast. In the figure (a) and (b) represent the mean value for intense and low-intense cyclone and the figure (c) and (d) represent standard deviation. From this analysis, it found that the mean radius range is 62-64 km for the cyclone of Bangladesh.

Chapter 3

Fig.3.12 Mean and standard deviation of maximum sustained wind radius

This study found that the maximum sustained wind radius decreases in the east coast and increases in the middle and west coast in future. From the wind-pressure relation, it is evident that the intense cyclone activity will increase in the east coast. It also found that the fluctuation of maximum sustained wind radius is very high along the east coast of Bangladesh. I have used this information to find the dangerous cyclone for surge simulation in the chapter 6.

3.5.4 Translation speed

Translation speed of a tropical cyclone represents the movement speed of a cyclone. This study investigated the movement speed of a cyclone, which strike in the coast of Bangladesh. Due to the climate change, there is an impact on cyclone translation speed. Cyclone translation speed has a great impact on storm surge. Therefore, for the future dangerous cyclone risk analysis, it is important to find the behavior of the cyclone movement speed.

Chapter 3

Fig.3.13 Mean and standard deviation of translation speed

To find the cyclone translation speed, I have used the general definition of speed. At first, find the distance of consecutive two points (before landfall and after landfall). After that, the distance is dividing by the passing time. Basically, this movement speed is the speed of crossing the coast. For each part of Bangladesh coast, I have found the movement speed of individual cyclone. I have found that the climate change and the coastal location have some impact on movement speed. Different climate scenario shows different translation speed. Fig. 3.13 shows the mean and standard deviation of translation speed. Figure (a) and (b) show the mean translation speed for the low-intense and intense cyclone. Figure (c) and (d) show the standard deviation of translation speed for the low-intense and intense cyclone. From this analysis, it was found that the fluctuation (standard deviation) of translation speed is zero for the HA scenario at the middle part. The mean translation speed is 15-20 km/h for the low-intense cyclone but 20-25km/h for intense cyclone.

Chapter 3

3.6 Conclusions

In this chapter, I have investigated the future changes of the cyclone parameter due to climate change. The general characteristics that I have found from the previous analysis will be used to find the dangerous cyclone and its associated storm surge along the coast of Bangladesh. In this chapter, I have found that the Indian coast faces most of the cyclone but Bangladesh is the prone region for intense cyclone. West part of Bangladesh will be faces huge number of tropical cyclone, but the severe cyclone strike the east part of Bangladesh. Landfall angle will be shifted slightly from Southeast to Southwest side. Intense cyclone moving speed will be decrease in future. Maximum sustained wind radius will fluctuated for intense cyclone but stable for low intense cyclone.

Chapter 4

Chapter4

Development of numerical storm surge model

4.1 Chapter summary

In this chapter, a numerical storm surge model known as bay-river couple model is developed to simulate the surge height along the coast of Bangladesh. This model is a shallow water Cartesian coordinate model, which is discretized by the finite difference method with forward in time and central in space. To compute the numerical simulation, the boundaries for the islands and coast are approximated through proper stair steps representation and solved it by a conditionally stable semi-implicit manner on a staggered Arakawa C-grid. In the southern open boundary, a most energetic tidal constituent, M2, is used in the bay model. A one- way nested scheme technique is used in the bay model to include coastal complexities as well as to save computational costs. Finally, I have used this developed model to foresee the sea-surface elevation associated with the catastrophic cyclone of 1991 and cyclone MORA. After that, I tried to found the water level elevation through model simulations with and without the inclusion of the river system. From this study, I have found that the surge height in the bay- river junction area decreased by 20% and the surge height reduced by about 3–8% outside the junction area. The obtained result makes a good agreement with some of the observed data.

4.2 Shallow water model equations

From the concepts of atmospheric or oceanic phenomenon, it is found that the z- component of the momentum equation may be approximated by the hydrostatic equation when the horizontal length scale is much larger than the vertical scale. However, the continuity equation may reduce to the non-divergence of velocity, when the density variation in all directions is negligible. Finally, neglecting the molecular viscosity of the oceanic fluid, the basic shallow water equations are as follows (e.g., see Paul and Ismail, 2013)

Chapter 4

u  u  u  u 1  p u  v  w  fv   , (4.1) t  x  y  z   x

v  v  v  v 1  p u  v  w  fu   , (4.2) t  x  y  z   y

p  g , (4.3) z

u  v  w    0 , (4.4) x  y  z where u , v , w are the instantaneous components of velocity (m/s) in the directions of x, y, and z, respectively; t is the time (s); p is the pressure (hPa); ρ 3 the density of the sea water supposed homogenous and incompressible (kg/m ); f 2  sin  the Coriolis parameter, where  is the angular speed (m/s) of the

Earth rotation and  is the latitude of the place of interest; g the acceleration due to gravity (m/s2).

4.2.1 Average procedure

Using Eq. (4.4), Eqs (4.1) and (4.2) can be expressed in the following form

u  u  u  u1  p u  v  w  fv   , t  x  y  z  x

u     u  v  w1  p or, ()()()uu  uv  uw  u  u  u  fv   , t  x  y  z  x  y  z  x

u     u  v  w1  p or, ()()()uu  uv  uw  u    fv   , t  x  y  z  x  y  z  x

up   1  or, ()()()uu  uv  uw  fv   . (4.5) t  x  y  z  x

Similarly,

v (uv) (vv) (vw) 1 p     fu  . (4.6) t x y z  y

Chapter 4

To make velocity average, I have chosen the instantaneous velocity components. u u u', v  v v ', w  w w ', with the replace ofu , v and w in Eqs.(4.4)-(4.6). where, u is the average time mean velocity of a fluid particle. Also u' designates deviation from the mean velocity at any time. To memorize that the average in deviation from mean velocity must be zero. So, u'   v '   w '  0, i . e ., To make more simplification, if we neglect the small fluctuations in density associated with the turbulence, then the resulting equations can be averaged in time to get a simple partition of the flow between the mean flow and turbulent fields. Then the average of the term uw can be written as  uw    ( u  u)( w  w)    u  w    u w  , where the terms like  u  w  and  u  w  vanish because u and  w are constants over the averaging interval.

Carrying out this averaging process for all terms in Eq. (4.4), we obtain

 u    v    w     0 . (4.7) x  y  z

By using this averaging process in Eq. (4.5), and using Eq. (4.7), it is found that  up    1     ()()()uu   uv   uw  f v , t  x  y  z  x  u    or, (  (u u )(  u u  )  ) (  ( u u  )(  v v ')  ) t  x  y

 1  p (  (u u )(  w w )  ) f v  , zx

 u    or, ( u  u  u  u   ) (  u  v u   v   ) t  x  y

 1  p (), u w u  w  f v  zx

Chapter 4

u  u  v  u  w  u or, 2 u  u  v  u  w t  x  y  y  z  z

   1  p ()()(), u u    u  v    u  w   f v x  y  z  x u  u  u  u1  p or, u  v  w  f v t  x  y  z  x

     u v w   u u   u  v  u  w    u    , x  y  z    x  y  z 

u  u  u  u1  p or, u  v  w  f v t  x  y  z  x

    u u     u  v     u  w   . x  y  z

Similarly, the averaging process in Eq. (4.6) with the aid of Eq. (4.7), it is found that v  v  v  v1  p or, u  v  w  f v t  x  y  z  y

    u v     v  v     v  w   . x  y  z Hence, the combined form of continuity and momentum equations, respectively, as  u    v    w     0 , (4.8) x  y  z

u  u  u  u1  p u  v  w  f v t  x  y  z  x (4.9)  u'''''' u    u v    u w     , x  y  z

v  v  v  v1  p  u    v    w   f  u   t  x  y  z  y (4.10)  u'''''' v    v v    v w     . x  y  z

In the right hand side of the equation (4.9) and (4.10), the terms on the square brackets called the eddy stress terms, which depend on the turbulent fluctuations.

Chapter 4

According to the Prandtl's mixing length theory, these eddy stress terms are parameterized in terms of the mean field variables by assuming that the eddy stress is proportional to the gradient of the mean velocity. Since the vertical gradients of the eddy stress are much larger than the horizontal gradients, so, only the vertical gradients are considered here. The eddy stress terms  wu   and  wv   can be expressed as (see, Holton, 2004)  u    u w   Az   z x    v  , (4.11)  v w   Az    z y 

where Az is the eddy exchange coefficient. By using the Eq.(4.11), Eqs. (4.8) - (4.10) can be written as (drop the angle brackets) u  v  w    0 , (4.12) x  y  z u  u  u  u11  p  u  v  w  fv    x , (4.13) t  x  y  z  x  z

v  v  v  v 1  p 1  u  v  w  fu    y . (4.14) t  x  y  z   y   z The Eqs. (4.12)-(4.14) are the basic shallow water equations with averaged velocity components. In the above equations uv, and w represent Reynolds averaged components of velocity in the directions of x, y and z, respectively, x ,

 y are the x and y components of friction stress. 4.2.2 Vertically integrated equation

At first, considering the curvature of the earth surface is assumed to be zero to derive the vertically integrated shallow water equations. To derive the vertically integrated shallow water equations, a system of rectangular Cartesian coordinates is used in which the origin, O, is in the undisturbed level of the sea surface as the xy -plane and OZ is directed vertically upwards. Let us consider the displaced position of the free surface as z (x, y, t) and position of the sea floor as z h(x, y) , so that the total depth of the fluid layer is h . To measure the movement, considering the

Chapter 4

Fig. 4.1 Fundamentals of wave conditions direction of the free surface point as OX points towards the south, OY points towards the east and OZ is directed vertically upwards (see, Fig 4.1). When the storm propagated from the free surface, the upper surface stress is generated due to circulatory wind and the bottom stress acts as the wastage term. This term is known as bottom friction. The surface stress and the bottom frictions of the x and y- components are Tx andT y and Fx and Fy . Then the bottom surface conditions are given by

(xx , yy )  (FF , ) anduvw    0 atz   hxy ( , ) (4.15)      ( ,  )  (TT , ), PPandw    u  v atz   (,,) xyt (4.16) x y x y a t  x  y

The condition is known as kinematic surface boundary condition (KSBC). Now integrate each term of the Eqs. (4.12)-(4.14) in the vertical direction from z  h x y),( to z   tyx ),,( , where the bottom and surface conditions are given by Eqs.(4.15) and (4.16). By using Leibnitz rule of differentiation under the integral sign, the modified relations are

 u   u dz dz  u  u()  h . (4.17) xhh  x  xh  x

At first, integrated each term of the Eq. (4.12) over depth, Then,

uh    ()    dz udz [][][] u  u  udz  u , (4.12a)  h  hx  x  h  x  x  xh  x

Chapter 4

vh    ()    dz vdz [][][] v  v  vdz  v  h  , (4.12b) hy  y  h  y  y  yh  y

 w      dz[ w ] [ w ]   [ u ]  [ v ]   h   . (4.12c) h z  t  x  y

Adding all the terms, the vertically integrated form of the continuity equation become

   udz  vdz  0, (4.18) t  xhh  y

1  1  Defining u U udz and v V vdz , simplified form of the   h h   h h above equation is

    h  U     h  V   0. (4.19) t  x   y   From Equation (4.13)

u  u  u  u  u  v  w11  p x u  v  w  fv  u       , t  x  y  z  x  y  z  x  z

u  u  u  v  u  w11  p x or, 2u  v  u  w  u  fv    , t  x  y  y  z  z  x  z

up   11   or, u2   uv   uw   fv    x , t  x  y  z  x  z

up   11   or, u2   uv   uw   fv    x . (4.20) t  x  y  z  x  z p Vertically integration of the hydrostatic equation g , assuming  to be z

pa p  independent of z , dz  g dz  p  p   g (   z ) , a pzz

p p  or, a g . (4.21) x  x  x

Chapter 4

Equation (4.20) can be written as

u   11p   u2   uv   uw   fv  ax  g  . (4.22) t  x  y  z  x  x  z

Now, integration of each term of the Eq. (4.13), which is equivalent to the integration of each term of (4.22). Integrating each term of Eq. (4.22), then

u       dz udz  u  u()  h  udz  u . (4.22a) ht  t   h z  t z  h  t  t h z  t

    u2  dz u 2 dz  u 2  u 2 ()  h , hhx  xz  x z  h  x

   h or, u2  dz u 2 dz  u 2  u 2 . (4.22b) hhx  xz  x z  h  x

    uv  dz uvdz ()()() uv  uv  h , hhy  yz  y z  h  y

   h or, uv  dz uvdz ()() uv  uv . (4.22c) hhy  yz  y z  h  y

  uw  dz ()() uw  uw . (4.22d) h z z z  h

 1xx 1  1 1 dz dz  x   T x  F x . (4.22e) hhzz  h 

1p 1  p 1  p adz a dz  a   h . (4.22f) hhx   x   x

     g dz g dz  g  h . (4.22g) hhx  x  x

Therefore, integration of each term of Eq. (4.22) and simplified then

Chapter 4

     h   udz u  u2 dz  u 2  u 2  uvdz th  t  x   h z  x z  h  x  y h

 h  1 p uv  uv  uw  uw  f vdz  a   h  (4.23) zy z  h  y z  z  h h   x  1 g  h   Txx  F , x 

 1  1  1 If we defineU udz , U22 u dz , UV u v dz and so on. h h    h   h   h h Then,

 2    Uh    U   h    UVhuuvw        t  x  y  x  y

hh  1 pa  u u  v w u  fV h     h  g   h  x  y  t  x  x 1 TFxx , 

    h or, U h    U2    h    UV   h    u  u t   x  y   z  t z  h  t

11pa u  fV  h      h   g   h   Txx  F , z t  x  x

   1 p or, U h    U2    h    UV   h    fV   h   a    h  t   x  y     x (4.24)  1 g  h   Txx  F . x 

Similarly integrating vertically Eq. (4.14) over depth and defining

 1  1  1 U udz , U22 u dz and UV u v dz and so on. h h    h   h   h h Finally, It is found

   2 1 pa V h     UV   h    V   h   fU   h      h  t  x  y  y (4.25)  1 g  h   Tyy  F y  

Chapter 4

The first term on the right hand side of each of the Eqs. (4.24) and (4.25) depends upon horizontal gradient of atmospheric pressure that is generally very small. However, during a stormy period, the horizontal gradient may not be negligible, but at that time, the stress due to storm wind is more prominent and dominating. Therefore, we may approximate Eqs. (4.24) and (4.25) dropping the mentioned negligible terms, one can see them to be

 2   U h    U   h    UV   h    fV   h    g   h  t  x  y  x (4.26) 1 TFxx , 

  2  V h     UV   h    V   h   fU   h    g   h  t  x  y  y (4.27) 1 TF.  yy

Eqs. (4.19), (4.26) and (4.27) can be set to the forms:

    h  u     h  v   0 , (4.28) t  x   y  

 2   u h    u   h    uv   h    fv   h    g   h  t  x  y  x (4.29) 1 TFxx , 

  2  v h     uv   h    v   h   fu   h    g   h  t  x  y  y (4.30) 1 TF.  yy

According to the conventional quadratic law a parameterization of the bottom stress may be made in terms of the depth averaged currents, so the bottom friction may express

2 2 2 2 Fxf  C u u  v and Fyf  C v u  v , (4.31)

Chapter 4

where Cf is known as bottom friction coefficient, which be kept as a constant in our model. Using Eq. (4.31) in Eqs. (4.29) and (4.30), we get     T u h   u2    h   uv   h   fv  h  g   h  x t   x  y    x  (4.32)

2 2 Cf u u v ,

    T v h    uv   h    v2    h  fu  h  g   h  y t   x    y  y  (4.33)

2 2 Cf v u v .

For the numerical treatment, it is convenient to express Eqs. (4.28), (4.32) and (4.33) in the form as,

 uv     0, (4.34) t  x  y

u ()() uu  vu  T C u() u22 v   fv   g()  h x  f , (4.35) t  x  y  x  h

v ()() uv  vv  T C v() u22 v   fu   g()  h yf  , (4.36) t  x  y  y  h where (u,v) (   h)(u,v). (4.37)

Here, u and v in the bottom stress terms of Eqs. (4.32) and (4.33) have been replaced u~ and v~ in Eqs. (4.35) and (4.36) in order to solve the equations numerically in a semi-implicit manner. To incorporate the river domain with realistic geometry the model was develop for the river system. For the developed river model, the origin O′ was taken at the equilibrium level of the free surface and was located at the left corner of the head of the river latitude (23.25° N, 90.40° E). The x -axis was taken as positive in the stream direction while the y -axis was taken parallel to the coast. The elevation of the free surface above its mean level was denoted by z  (,) x y z h(,) x y riv and the bottom of the river by riv . The depth

Chapter 4

averaged velocity components in the developed river system, uriv and vriv , satisfied the following equations

riv  Huriv   Hv riv  0, (4.38) t  x  y

    1/2 Hu   Hu2  Huv   Hv   griv  Cuu2  v 2 , (4.39) triv  x riv  y riv riv riv  x f riv riv riv 

1/2    2  2 2 Hv  Huv   Hv  Hu   griv  Cvu  v , (4.40) triv  x riv riv  y riv riv  y f riv riv riv 

where H[riv h riv ]( x , y ) represents the total water column depth where uvriv,, riv are the velocity component. The bottom stress, following the study (Agnihotri, 2006),

22 was parameterized as Cf u riv v riv .

4.3 Boundary conditions

In this model boundary condition a radiation type of boundary condition is used, (Heaps,1973).

g ucos v sin    ( )  , for all t  0. (4.41) h

Where,  denotes the direction of the outward normal to the coast measured clockwise from north direction. The radiation type boundary condition allows the propagation of energy only outwards from the interior in the form of simple progressive wave. The modified form of the equation (4.41) is used for the coastal boundaries. At the coastline, east boundary, southern boundary, and west boundary the direction angle  0 ,  270,  180, and   90 respectively (see, Fig 4.1). The following radiation types of boundary conditions are taken for open sea boundaries:

g 1 At the west boundary : v ( )2  0 , (4.42) h

Chapter 4

g 1 At the east boundary : v ( )2  0, (4.43) h

1 1 2 g2  g  2 t  At the south boundary: u( )   2  a Sin    (4.44) h h   T  The normal components of the mean current are taken as zero at the closed boundaries of the mainland and the islands. In Eq. (4.44), a , and T denote the amplitude and phase of the tidal forcing and period of the tidal constituent under consideration. The boundary condition of river model was considered as the eastern and western river boundaries, whenever it is a vertical sidewall, the boundary condition can be set as

riv  0 (4.45) Q uriv  , and uriv  0 (4.46) Bhriv riv

when fresh water discharge and no fresh water discharge is considered, where Briv is the river breath and Q represents the river discharge. Both the bay and river model simulation, a matching is important. In the model simulation, the bay influences the southern boundary of the river model. The continuity of volume flux should ensure at the mouth of the river. In this study, the two-dimensional matching condition was used in our model by the following equation.

riv bay , uubay riv 0 and vvbay riv 0 (4.47) Equation (4.47) can ensure that the sea surface elevation that influences the river and the continuity of volume flux of the flow. This matching equation can exchange the information of the bay and river model in such a way that the model simulation’s updated value from the river model is again used as the bay model input. The following equation prepares the input for the bay model after the next iteration. Q uubay riv (4.48) Bhriv() riv riv This is an appropriate process through the matching to find the appropriate result from the bay and river couple model. In our model simulation, the bay model inner nested grid size and the river model grid size were the same, so that the matching

Chapter 4

could exchange the flow velocity and volume easily with less time and cost. A contraction mapping is needed when the grid size does not match.

4.4 Determination of the forcing terms

The major forcing terms of the surge forecasting model equation are horizontal gradient of the atmospheric pressure, seabed friction, wind stress, and Coriolis force. Basically, the long wave shallow water equation are used for computing tide, surge, and their interaction. In the mesoscale sea model, the horizontal gradient of atmospheric pressure is negligible. On the other hand, during a storm period, the horizontal gradient of atmospheric pressure is not negligible, but at that time, the main generating mechanism of surge amplitude is the tangential stress due to strong wind associated with the storm. So, the atmospheric pressure gradient terms may be neglected to compute tide, surge and their interaction. The Coriolis force can be determined by knowing the latitudinal position of the area and the bottom friction may be parameterized in terms of depth-averaged currents by a quadratic law. Generally, the wind stress parameterized in terms of the wind field associated with the storm, which done by the conventional quadratic law. The conventional quadratic law can be written as

1 1 222 222 Tx C d  a u a (u a  v a ) and Ty C d  a v a (u a  v a ) , (4.49)

where Cd is known as the drag coefficient. It is difficult to get the wind field data from the meteorological department of Bangladesh. Rather the information is available in terms of the maximum sustained wind velocity and the corresponding radial distance from the eye of the storm and the difference of the atmospheric pressure between the eye and periphery. For the Bay of Bengal region most frequently used formula is due to Jelesnianski (1965), which is given by

 3 V0  raa R for all r R Va   (4.50) V R/ r for all r R  0  aa

Chapter 4

where V0 is the maximum sustained wind at the radial distance R and ra is the radial distance at which the wind field is desired. The x and y components (ua, va) of the wind field are derived from Va given by the above empirical formula.

4.5 Numerical procedure 4.5.1 Model setup In order to estimate the appropriate water level due to a surge in the designated study area, it is important to investigate the activity of the cyclone over the area at least 3 days before crossing the coast.

Fig. 4.2 Model domain area

Fig.4.2 shows the model domain area. It is observed that the surge response along the coast becomes significant well before crossing the coast. Therefore, the mesh

Chapter 4

size (two consecutive grid points distance) should be smaller to incorporate the major islands of the estuary. To make smaller the mesh size, I have faced two practical problems over the whole study area: (i) Number of grid points where the computation parameter ζ , u, v need to be computed there takes more space of computer memory and much CPU time in each time step of the solution process. (ii) To make ensure the stability of an explicit finite difference scheme, smaller time step requires smaller mesh size. So that, the time step, t , should be related to the mesh size of x , y , and the ocean depth, h, by the following relation of CFL (Courant- Friedrichs-Lewy) condition x t 2gh where, g  gravitational acceleration. Considering this fact, a nested mesh scheme has been used to predict the water level due to a storm. The physical location of the scheme is shown in the figure Fig.4.2. The prescribed nested schemes are: fine mesh scheme (FMS), coarse mesh scheme (CMS) and a very fine mesh scheme (VFMS). In this study, the nesting is made sequentially from the CMS to FMS and FMS to VFMS. The CMS cover the domain 150 N to 230 N latitude and 850E to 950E longitude. In this mesh size, the gird space is x 15.08 km and y 17.52 km associated with the 60x61 grid points in the computation domain. In a similar fashion, FMS cover the area between the latitude 21o-15 to 23o N and 89o E to 92o E longitudes with 92x95-grid point associated with the grid distance x 2.15 km and y 3.29 km. To incorporate the landfall target area properly, the VFMS cover the area between 21.770 N to 23o N latitude and 90.40o E to 92o E longitudes. In this scheme, the grid size of the x-axis (north-south direction) is x 720.73m and the y-axis (east-west direction) y 1142.39 m with 190x145 grid points. In the river model, the study domain was extended between 23° N to 23.25° N and 90.40° E to 90.68° E. For the extended region, we considered 40 × 27 computational grid points with a grid spacing of 0.72 km along the north–south direction and 1.14 km in the east–west direction, which was referred to as a very fine grid model for river (VFGMR). The grids of the VFGMR were taken in such a way that along the west–east direction for the y-axis direction, the north–south

Chapter 4

direction was an x-direction river flow and the VFGM parent grid coincided with the grids of the VFGMR (matching points at the river and bay). The VFGMR depends on VFGM and the parameters  , u and v of the VFGM can pass through the boundary 23° N to 23.25° N and 90.40° E to 90.68° E. The process of representing the approximation of island boundaries through a rectangular grid is known as stair-step representation (see, Fig.4.3).

Fig.4.3 Mesh scheme with stair step representation

The boundary condition of the CMS scheme is open scheme for nonlinear tide interaction (see, Mohit et al.(2018a)). The CMS scheme is fully independent and the parameter  prescribed from those obtained in CMS in each time step of the solution process used as boundaries of the FMS. In a similar way for VFMS, parameter  prescribed from those obtained in FMS in each time step of the solution process used as boundaries of the VFMS. But, for the bay-river couple model, the matching condition is used to simulate the surge considering the river dynamics. The river discharge was consider Q 5100.00m3 from the study of Jain et al. (2007).

4.5.2 Numerical discretization of the model

The governing Eqs. (4.34), (4.35), and (4.36) and the boundary conditions (4.42), (4.43), and (4.44) are discretized by finite-difference (forward in time and central in

Chapter 4

space) and are solved by conditionally stable semi-implicit method using a staggered grid system as mentioned earlier. For the variable  at a grid point (i , j) at time tk is represented by

k xi,, y j t k   i, j

According to the definition of differentiation

 kk1   kk  kk  i,, j i j ;  i1, j i 1, j , and  i, j 1 i , j 1 tt xx2 yy2

Any variable U for averaging operation

kk kk y x UU y UU xy x U k  i1, j i 1, j ; U k  i, j 1 i , j 1 , and UUkk ij, 2 ij, 2 i,, j i j

The last term of the model equation (4.35) and (4.36) also discretized in a semi implicit manner. For example the term u~ u 2  v 2  is discretized as

22 uk1 u k v k where the superscript k+1 indicates that u~ is to be evaluated in advanced time level. Thus, the model equation Eq. (4.34) is as follows:

kk1 i,, j i j t TL12  TL  (4.51) where

kk kk uui1, j i 1, j vvi, j i i , j 1 TL1  ; and TL2   ; 2x 2y  

k1  ij, in Eq. (4.51) is computed at i = 2, 4, 6,....., M – 2. and j = 3, 5, 7,...... , N –2.

Similarly, Eq. (4.35) becomes

Chapter 4

uk   t( TL 1  TL 2  TL 3)  t TR 1  TR 2  (4.52) uk1  ij, ij, 1t . FR 3  where

uk U k u k U k TL1  i2, j i  2, j i  2, j i  2, j ;  4x

y x y x uk V k u k V k TL2  i, j 1 i , j  1 i , j  1 i , j  1 2 y 

xy k TL3  fi vij, ;

kk11 k 1 i1, j i 1, j TR1   g i,, j  h i j  ; 2x

 TR2  x ; 

xy 2 kk2  CUV  f i,, j i j and, FR3  k1 ;  i,, j h i j ~l1 u , ji in Eq. (4.52) is computed at i = 3, 5, 7,....., M –1 and j = 3, 5, 7,...... , N –2

Also, Eq. (4.36) becomes

k k1 vij,   t. TL 1  TL 2  TL 3    t .( TR 1  TR 2) vij,  1t . FR 3  (4.53) where

y x y x Uk v k U k v k TL1  i1, j i  1, j i  1, j i  1, j ; 2 x

Chapter 4

k k k k V vi, j 2 V v i , j 2 TL2  i, j 2 i , j 2 ; 4 y

xy k1 TL3  fi uij, ;

kk11 k1 i, j 1 i , j 1 TR1  g i,, j  h i j  2y

 TR2  y ; 

xy 2 kk2 CUV f i,, j i j and, FR3  k 1  i,, j h i j

k 1 vij, in Eq. (4.53) is computed at i = 2, 4, 6,....., M –2 and j = 2, 4, 6,...... , N –1

For the boundary conditions (4.42) - (4.43), the elevations at j = 1, j = N and i = M are computed respectively in the following manner.

k11  k  2 (h / g ) V k i,1 i ,3 i ,2 i ,2 (4.54)

k11  k  2 (h / g ) V k i, n i , n 2 i , n  1 i , n  1 (4.55)

kk11   2 (h / g ) U k m, j m 2, j m 1, j mj1, (4.56) where i = 2, 4, 6, . . ., M –2. and j = 1, 3, 5, 7, . . ., N. At the initial, consider the initial values of  , u and v are zero. The time step of the model simulation is taken as 60 seconds that ensures stability of the numerical scheme. In the solution process, the value of the friction coefficient Cf and the drag coefficient CD are 0.0026 and 0.0028 respectively. Discretization of river model is given in appendix 1.

Chapter 4

4.5.3 Computation process

The governing equations were discretized by forward in time and central in space finite difference scheme by considering discrete points in the xy plane for the bay and river model. The model equations are (4.34), (4.35), and (4.36) for bay model and (4.38), (4.39), and (4.40) for river model. In the bay model, (,)ij represent the position of a grid point, where xi ( i  1)  x , i  1,2,3,... m (even), yj ( j  1)  y , j  1,2,3,... n (odd) and the sequence of time instant is defined bytk  k  t, k  1, 2, 3,.... By using the conditionally stable semi-implicit manner staggered C-grid to solve the model equation there are some distinct types of computational points. The distinct types of computational points are , u and v for the parent (bay) model and riv , uriv and vriv for the river model.

Both the bay and river model the distinct points are considered with i even and j odd, the point is a  point at which  is computed. If i and j are both odd, the point is a u point at which u is computed and if i and j are both even, the point is v , at which v is computed. The M number of computation grid was chosen to be even so that the eastern open sea boundary consisted of both  -points and v - points. The N number of the computation grid was chosen to be odd to ensure that there were only  -points and u -points for all of the grid points of the river mouth, where the continuity of volume flux was ensured. To consider for cold start time of simulation, the initial value of  , u and v were taken as zero. A 60 second-time step process was used for both the model simulation in order to meet the Courant- Friedrichs-Levy (CFL) stability criteria. In the southern open boundary of CGM, a stable tidal regime was then generated with the M 2 constituent. The computed boundary data was exchanged from each nesting model. The boundary data was exchanged by the linear interpolation method. For the entire nested model, the computation time step was checked by the CFL stability criteria. To find the effect of river, a matching is important for the river model to work properly. The obtained data of VFGM were assigned in the river model through the matching condition. The General Bathymetric Chart of the Oceans (GEBCO) bathymetry data was interpolate to the gird point of river model domain. In the river model, for

Chapter 4

the solution process, the extrapolated value of the elevation from the bay model at the bay-river junction point provided a forcing for the motion in the river. The use of Equations (4.38-4.40) demonstrated the updated elevations and currents at the  and u -points in the river, except for the current at the head of the river where the u -velocity was given by one of the boundary conditions specified by Equations (4.45) and (4.46). The updated elevation at the junction point from the river model was chosen to update the elevations in the bay model by using Equations (4.47) and (4.48) for the next iteration. The first point of  and u in the river model, the elevation and current were computed from the bay model by using the matching phenomenon. This matching procedure guided the updated elevations in the river and, additionally, yielded boundary values of the mouth of the river to be used in the next update of the bay model variables. The similar procedure was adopted for the v velocity.

4.6 Model validation and outcomes

To validate my model simulation, I have used the cyclone information of 1991. The model simulation was computed for 72 h of the storm from 0800 local time (LT) of 27 April to 0800 LT of 30 April for the storm April 1991 and the results are presented for the last 48 h at some of the stations. The numerical computation considered in such a way that the model run 4290 time step, the model storing data after every 10 min, 289 was the number of data the model stores and the model started to store data after it was run 2040 times.

Fig.4.4 Surge height at different tide station when incorporate the river and without incorporate the river

Chapter 4

To find the impact of a river on the surge height, the model was run for the two cases: one is incorporating the river system and another is without incorporate the river system. Fig. 4.4 shows the surge height at some tide station near the coast of Bangladesh. The model results fairly agree with the observed data of Bangladesh Inland water Transport Authority (BIWTA). In this study, it is found that the average surge height around the coast of Bangladesh is 3-6 m. The recession was found at the west part of Bangladesh when the storm strikes the east part of Bangladesh coast. That's why, the lower surge height was found near the Hiron point area (Sundarbans) for the storm 1991. Fig. 4.5 shows the surge height for both the case with non-linear interaction of tide. The water level elevation due to storm was very high due to the height peak time of tide. In this analysis, I have found that the river system has a noticeable impact on surge height and the river influence the surge height near the junction area. I have found the changes of surge height along the coast of Bangladesh.

Fig. 4.5 Surge height at different tide station when incorporate the river and without incorporate the river

The overall surge height changes is 10-30cm due to the river incorporate in the model simulation. But, near the junction point of bay and river is found 1-1.2m.

Chapter 4

Fig.4.6 Comparison result of two different cases simulations (With and without river)

Fig.4.7 Combined figure of the time series of storm surge

Fig. 4.6 shows the impact of river in surge simulation. The combined figure may easy to understand the overall situation of surge height due to consider the river in the model simulation. Fig.4.7 show the time series of surge height in different condition in a single tide station with observed data and simulated data. The observed data scarcity is high for comparison. The reason behind this may be due to the rough weather and the manual process of data collection. The red circle shows the observed data collected from the Bangladesh Inland Water Transport Authority (BIWTA), which were very scarce. The figure legend ‘WRDCH’ represents the surge height considering the river impact at the Cox’s Bazar tide

Chapter 4

station and ‘WORDCH' stands for without river discharge at the Chittagong tide station.

Table 4.1 Overall maximum wave peak statistic from the past study and the present study. Simulated Overall Simulated Overall Simulated Simulated Observed Max. Overall Max. Max. Overall Max. Overall Overall Coastal Water Water Level Water Water Level Max. Water Max. Location Level (m) (m) by FDM Level (m) (m) (without Level (m) Water by [Paul [Mohit et al. by [Roy River) (with River) Level (m) et.al 2014] 2017] 1999a] Cox’s Bazar -- -- 6.14 5.98 5.89 6.00 Moheshkhali -- -- 4.12 4.59 4.57 -- Banshkhali -- -- 3.58 ------Chittagong 6.25 5.45 4.50 4.60 4.61 5.4 Sitakunda 5.78 -- 4.48 5.15 5.11 -- Sandwip 5.63 5.33 4.38 5.21 5.20 -- Mirsharai -- -- 5.66 5.05 5.03 -- Companiganj 7.28 -- 6.15 5.90 5.89 6.1 Chital Khali -- -- 4.50 ------Char Jabbar 6.35 5.18 5.69 5.51 5.47 -- Char Changa 5.81 4.31 4.12 4.60 4.58 -- Char Madras 5.81 -- 4.32 4.99 4.95 -- Rangabali 4.50 -- 4.07 3.56 3.56 -- Kuakata 3.86 -- 3.96 3.60 3.58 -- Patharghata -- -- 4.36 3.55 3.57 -- Tiger Point 4.57 -- 4.21 4.30 4.30 -- Hiron Point 4.01 0.70 3.80 3.48 3.45 3.5 Monpura ------5.20 4.88 -- Island Sona Char ------4.51 3.61 --

Chapter 4

Fig.4.8 Distribution of maximum water level and their difference

Chapter 4

Similarly, ‘OBCH’ represents the observed data at the Cox’s Bazar tide station. The results ‘WRDJ’, ‘WORDJ’, ‘WRDWOTJ’ and ‘WORDWOTJ’ represent the simulated surge height at the junction point with the condition of the inclusion of river (WRDJ), without river (WORDJ), inclusion of the river and tide (WRDWOTJ) and without inclusion of the river and tide (WORDWOTJ). Table 4.1 shows the overall peak of the water level at some location near the coast. This table also shows the similarity to our study with the results of other research. The overall distribution of maximum surge height around the coast of Bangladesh is shown in Fig. 4.8. The upper figure shows the distribution of maximum surge height when there is a river in the model simulation, the middle figure shows the simulation result when there is no river in the model, and the lower figure shows the difference of surge height. The decreasing of the surge height found near the junction area due to the inland penetration of water flow through the river channel. Due to the cyclone track features the surge height of cyclone MORA shows the lower fluctuation. In this study, I have found that the fluctuation of surge height is negligible for the storm of MORA due to the incorporate of river and without river condition in the model simulation. However, I have checked the surge height for the mean annual river discharge effect. Fig. 4.9 shows the surge height for river discharge condition.

Fig. 4.9 Surge height of a cyclone 1991 at the junction point (left side) and some different points (right side). Left figure show the surge height at a single station for discharge case. Right side figure shows the maximum surge height in different tide station.

Chapter 4

4.7 Conclusions In this chapter, a numerical storm surge model is developed. The developed numerical storm surge model is known as bay-river couple model. This study investigates the effect of the Ganga–Bhramutra–Meghna river system on the water levels associated with a storm along the coast of Bangladesh. It is found from this study that a river has an impact on storm surge along this region. From this study, I have found that the surge height reduces 20% near the bay-river junction point, when a river system is incorporated in the model. The reason behind this fact is inland penetration of water due to a storm. Thus, the response of a river on the surge development is appreciable and cannot be ignored for storm surge prediction purposes. However, the model has some limitations (ignore the wet dry technique), but the model can be used for operational predictions of storm surges along the coast of Bangladesh.

Chapter 5

Chapter5

Comparative analysis of two numerical storm surge models

5.1 Chapter summary

This chapter deals with the comparative analysis of storm surge simulation along the coast of Bangladesh through the finite difference numerical nested model (bay- river) and the Finite Volume Coastal Ocean Model (FVCOM) to find the bay-river model performance. The model equations of the numerical nested model are discretized through Finite difference Method FDM (bay-river) and solved it by a conditionally stable semi implicit manner on an Arakawa C-grid system. For the FVCOM, σ-coordinate is used for the irrational bottom slope representation and the mesh grid of the study domain is generated by the unstructured triangular cells. In both the cases, the models are applied to extrapolate sea-surface elevation associated with the catastrophic cyclone 1991(BOB 01) along the seashore of Bangladesh. The simulation result of the two models makes a good agreement with the observed result.

5.2 Concept of FVCOM

In the recent world the civilization developed near the coastal area due to business, economy, and for other reasons. But, to protect this area from the ocean disaster is necessary. In the present situation, a serious challenge for oceanographers to developed a stable model even though the governing equations of ocean circulation are well defined and numerically solvable in terms of discrete mathematics. There are two well-defined method for calculating the ocean circulation and storm surge model. The methods are finite difference method (Blumberg and Mellor, 1987; Blumberg, 1994; Haidvogel et al., 2000) and finite element method (Lynch and Naimie, 1993; Naimie, 1996). Both the methods have some good features and some limitation also. Considering both the methods, a 3-D unstructured-grid, free-

Chapter 5

surface, primitive equation, Finite Volume Coastal Ocean circulation Model (called FVCOM) (Chen et al. 2003a; Chen et al. 2004b, Chen et al., 2006a,b) was developed.

5.3 FVCOM model equations

Considering the advantages of FDM and FEM, a primitive and prognostic, unstructured-grid, Finite-Volume, free surface, three-dimensional (3-D) primitive equations Community Ocean Model was developed (Chen et al. 2003a). The unstructured triangular grid mesh can easily change its size depending upon the location and the domain or applied region. The simplex discrete scheme of FDM was considered to the integral form of the governing equation. In the FVCOM unstructured grid model, momentum flux calculations are used to deal with the integral equations like FDM and they are performed over an arbitrarily size of triangular mesh like FEM, hence one can obtain suitable results on mass conservation in terms of both the individual control element and the entire computational domain. The combination of the process of FDM and FEM can express the factors, mass, moment, heat conservation, and so on, in a complex geometry (Chen et al., 2006a). The governing equations of FVCOM contains momentum, continuity, temperature, salinity and density without snow and ice, which are solved by finite volume discrete method when calculating fluxes between arbitrarily sized triangle mesh adjoin each other. The governing equations of the momentum of x, y and z direction, continuity equation, temperature- advection diffusion equation, salt advection diffusion equation, density equations are (Chen et al. 2003a):

u  u  u  u 1(pHa p )   u u  v  w  fv    (Kmu )  F (5.1) t  x  y  z 0  x  z  z

v  v  v  v 1(pHa p )   v u  v  w  fu    (Kmv )  F (5.2) t  x  y  z 0  y  z  z

w  w  w  w   w u  v  w  (K )  F (5.3) t  x  y  z  zm  z

Chapter 5

u  v  w    0 (5.4) x  y  z

TTTTT      u  v  w  (K )  F (5.5) t  x  y  z  zhT  z

SSSSS      u  v  w  (K )  F (5.6) t  x  y  z  zhs  z

  (T,S) (5.7) where, x, y,z : east, north, and vertical axes in the Cartesian coordinate system; u,v,w : x, y,z velocity components; T ：temperature; S : salinity;  :density; pa

;air pressure at sea surface; pH :hydrostatic pressure; f ：Coriolis parameter; g： gravitational acceleration; Km：vertical eddy viscosity coefficient Kh：thermal vertical eddy diffusion coefficient; Fu,Fv,Fw,FT,FS：horizontal momentum, thermal, and salt diffusion terms; D=H+ζ：total water column depth; H：bottom depth，ζ：height of the free surface (relative to z = 0). 5.4 FVCOM model setup

5.4.1 Unstructured triangular grid

It is important to make accurate coastal representation for storm surge simulation. It is also a big challenge for storm surge modeler when configuring the computational domain.

Fig.5.1 Basic difference of structured and unstructured grids (Chen et al. 2013)

Chapter 5

Some model of storm surge simulation like, ROMs, POM, IIT-Delhi has a Cartesian coordinate rectangular gird to represent the coast, which can’t represent the coastline properly. Whereas, FVCOM applies unstructured grids that have non- overlapping triangular cells to resized the flexibly, hence, one can realize the accurate fitting even with the complex geometry compared to the structured grids. A smaller mesh size in the complex costal structure can make the good representation, but it is not time consuming. So, large mesh can be adapted to the open boundary while small one can be used around the region of the study area with irregular geometry. Fig.5.1 shows the visible difference of structured and unstructured grid.

5.4.2 Boundary conditions

To accrue the model simulation results, boundary conditions are needed to solve it numerically. In the model simulation, some boundary conditions are considered. The surface and bottom boundary conditions for u,v , and w are represented as:

u  v 1    E  P K(,)m (,);w  sx  sy   u  v  ;z   (x,y,t) (5.8) z  z 0  t  x  y 

u  v 1  H  H Qb Km ( , ) (  bx ,  by ); w   u  v  ; z   H(x, y). (5.9) z  z 0  x  y 

The x and y components of surface wind and bottom stresses can be

22 parameterized by the relation (sx ,  sy ),(  bx ,  by )  C d u  v (u,v) . Where the notations Qb is the groundwater volume flux at the bottom and Ω is the area of the groundwater source, E is the Energy flux and P represent the pressure. The drag coefficient Cd is determined by matching a logarithmic bottom layer to the model at a height zab above the bottom. The drag force can estimated from the relation

22zab Cd  max k / ln( ) ,0.0025 , where k = 0.4 is the von Karman constant and z0 z0 is the bottom roughness parameter.

The other boundary conditions of surface and bottom boundary conditions for temperature is

Chapter 5

T1 [Qn (x,y,t)  SW(x,y,  ,t)],z   (x,y,t) (5.10) z cph K

TTA tan  H ,z   H(x, y) (5.11) z Kh n

where Qn  x, y,t  is the surface net heat flux, which consists of net downward shortwave radiation, net downward longwave radiation, sensible, and latent fluxes. The shortwave flux incident SW x, y, ,t  is at the sea surface. The specific heat of seawater is cp,, AH is the horizontal thermal diffusion coefficient, α is the slope of the bottom bathymetry, and n is the horizontal coordinate (Pedlosky, 1974; Chen et al., 2004b). For the long wave, shortwave flux can be represented as the following relation when sensible and latent heat fluxes are assumed zero. The shortwave flux SW(x,y,0,t) is approximated by

z z SW tzyx ),,,(  SW yx ,0,,( t)[Re a  1(  R)e b ] (5.12)

According to the study of Kraus (1972), Simpson and Dickey (1981a), the absorption profile of upper ocean diurnal heating can estimated from the relation (5.13), and the surface and bottom boundary conditions for salinity can represented by (5.14, 5.15).

SW(x,y,z,t) SW(x,y,0,t) Rzz 1 R H(x,ˆ y,z,t)  [ eab  e ] (5.13) z cp a b

S S(Pˆˆ E)  cos  ,z   (x,y,t) (5.14) zKh

SSA tan  H ,z   H(x, y) (5.15) z Kh n

5.4.3  coordinate

A vertical coordinate system is important to represents the actual ocean bottom topography. In the FVCOM, σ coordinate system is used to represent the vertical coordinates of ocean bottom topography. The representation of bottom topography

Chapter 5

in both the representation of Cartesian coordinate system and σ coordinate system is shown in Fig. 5.2.

Fig.5. 2 Basic difference of Cartesian coordinate system and σ coordinate system

where, σ varies from the value of 0 to 1 and the notations were describe previously. After simplified the model equation through σ coordinate system the governing equations are as follows:

 Du  Dv      0 (5.17) t  x  y 

uD  u2 D  uvD  u      fvD t  x  y  (5.18) gD 0  D 1   u  gD  [(Dd)  '   ]  (K )DF   mx x 0  x  x D  

vD  uvD  v2 D  v      fuD t  x  y  (5.19) gD 0  D 1   v  gD  [(Dd)  '   ]  (K )DF   my y 0  y  y D  

TD  TuD  TvD  T  1   T    (K )  DHˆ  DF (5.20) t  x  y  D hT 

SD  SuD  SvD  S  1   S    (K )  DF (5.21) t  x  y  D hS 

Chapter 5

In the all direction the horizontal diffusion term in the σ coordinate system can represented as

 u   u  v DF [2A H ]  [A H(  )] (5.22) xx m  x  y m  y  x

 u  v   v DF [AH(  )]  [2AH ] (5.23) yx m  y  x  y m  y

    D(F ,F ,F ,F ) [ (A H ) (A H )](T,S,q22 ,q l) (5.24) T Sq22 q l x h  x  y h  y

where, Am and Ah are horizontal eddy diffusion coefficient and horizontal thermal diffusion coefficient, respectively. The boundary condition can be represented at water surface (σ=0)

u  v D Eˆˆ  P ( , ) (  ,  ),     K sx sy  0m (5.25) T D S S(Pˆˆ  E)D [Qn (x,y,t)  SW(x,y,0,t)],    cp K h Kh  at water bottom (σ=1)

u v D Q ( , ) (  ,  ),   b   K bx by  0m (5.26) TTSSAHH D tan   A D tan  22, Kh  A H tan   n  K h  A H tan   n

5.5 Study domain

Storm surges associated with intense tropical cyclones originating in the Bay of Bengal are responsible for major devastation along the coastal region of Bangladesh. Eighty percent of the global casualties occur in Bangladesh due to its complex geographical location at the northern tip of the Bay of Bengal. In this study, the study domain was considering the Bay of Bengal region (from 17o N latitude to 24o N latitude) to cover the total coastal area of Bangladesh. Fig. 5.3 shows the mesh grid with the Bay of Bengal coast.

Chapter 5

Fig.5.3 Unstructured grid of the model domain and the east part of Bangladesh coast

The southern open boundary unstructured mesh grid size is considered as 10 to 15 km and the innermost coastal area considered 500 to 150 m. For the inundation simulation, the mesh size of the Bhashan char and the Chittagong area is consider as 120 to 150 m. For this rough mesh size, the time step was consider 30s for accepting the stability condition.

5.6 Forcing factors

5.6.1 Atmospheric pressure

Some well-known model for atmospheric pressure distribution is used to compute the pressure profile numerically. Myers' model (Myers and Malkin, 1961) and Fujita's model (Fujita, 1952) are the well know model for Atmospheric pressure distribution. These models used the two parameters to distribute the isobaric line of the pressure distribution. The parameters are central pressure and the maximum wind speed radius. In this study, I have used Myers' model (Myers and Malkin, 1961) to estimate central pressure behavior of a cyclone. The air pressure distribution is expressed as a function on the distance from the center of cyclone as follows.

r p p   pexp(  0 ) (5.27) c r

Chapter 5

where, p : atmospheric pressure at the any point from the distance r of the center of a cyclone, pc : central pressure of the cyclone, p : pressure depth (=p∞-pc), p : the air pressure at the point which is located far from the cyclone, i.e. normal air pressure, 1013hPa , r0 : radius of the maximum winds; the distance from the center of the cyclone to the point with maximum wind velocity. Fig.5.4 shows the pressure profile of a cyclone 1991.

Fig. 5.4 Pressure distribution of cyclone 1991 at the time of maximum surge height and the time of coastline crossing. Figure (a) represents the distribution at 43 hour and Figure (b) represents the distribution at 48 hour

Chapter 5

These figures represent the pressure profile around the Bangladesh coast at the time of maximum surge height and the coastal crossing time. In my simulation result, I have found the maximum surge height at 43 hour where the simulation was performed for 48 hours. The simulation time was 28 April at 8.00 AM to 30 April 8.00:AM.

5.6.2 Wind profile

To estimate the wind profile, considering the wind in the free atmosphere because the wind velocity enough up above and there is no effect by the friction of sea surface. The wind profile estimation is calculated and multiplied with an empirical reduction coefficient so that the changes of the wind direction from tangential of an isobaric line to the direction of the cyclone center. The wind of the free atmosphere is obtained from the gradient wind components from the equilibrium of atmospheric pressure gradient force, Coriolis force and centrifugal force. The gradient wind components for cyclone movement are calculated from the study of (Fujii Mitsuda, 1986).

The relation of gradient wind component (Fujii and Mitsuda, 1986) is as

2 1p Ugr fU1 gr (5.28) a rr

After solving the above equation, the simplified form of gradient wind is

2 rf rf p rr00 Ugr     exp  (5.29) 2 2a r r

3 where, Ugr is gradient wind. a : Density of the atmosphere (= 1.22 kg/m ) f : Coriolis coefficient (= 2 sin , : Angular velocity of Earth rotation =7.29×10-5 rad/s，  : latitude)

UCU1 1 gr (5.30)

where the empirical coefficient value of C1 is used in between 0.6 to 0.7.

Chapter 5

Fig. 5.5 Schematic diagram of gradient wind, field wind, synthetic wind

The wind direction at sea is supposed to be deflected by the tangential direction of isobaric line, which is 300 towards the center side of a cyclone compared with the wind direction. For the another component, wind speed U2 of the wind component of the field at sea is

Ur1   UCV2 2 T (5.31) Ur10 

Here, C2 is empirical reduction coefficient, and generally the same value as C1 is used, VT is the traveling speed of the typhoon. Fig. 5.6 shows the velocity profile of a cyclone 1991. The figure shows the wind profile at the simulation time of 43 hour and 48 hour. Maximum wind speed is found near the east coast of Bangladesh at the simulation time of 43 hour, which is shown in Fig 5.6 (a). In the same figure, figure (b) shows the wind profile condition when the cyclone crossing the coastline.

Chapter 5

Fig. 5.6 Wind profile of cyclone 1991. Figure (a) represent the distribution at 43 hour and Figure (b) represent the distribution at 48 hour

5.7 Model data

5.7.1 Bathymetry data and tide

An important factor of storm surge simulation is the depth of ocean; in general, we called it Bathymetry. In my simulation, the bathymetry data was taken from the GEBCO’s gridded bathymetric data sets, which are the global terrain models for ocean and land. GEBCO_2019 Grid is a global 15 arc-second interval grid has been developed through the Nippon Foundation and GEBCO, which is known as

Chapter 5

Seabed 2030 Project. The Nippon Foundation of Japan is a non-profit philanthropic organization active around the world, whereas the international group of mapping experts of GEBCO develops the data set. The bathymetric data sets and data products are operating under the joint auspices of the International Hydrographic Organization (IHO) and UNESCO's Intergovernmental Oceanographic Commission (IOC). The GEBCO water depth data was linearly interpolated to each mesh grid point of the study area. Fig. 5.7 shows the bathymetry condition around the domain area. In my model simulation, the ocean tide estimated by NAO.99b (Matsumoto et al., 2000), which is a global tide model. For the nonlinear interaction of tide and surge, tide is estimated at the northern open boundary as a boundary condition. It is found that the simulated tide and the observed tide have a good agreement at some tidal station near the coast of

Bangladesh.

Fig. 5.7 Bay of Bengal bathymetry of GEBCO

Fig. 5.8 shows the comparison of simulated tide and observed tide at Hiron point (see, Fig 1.1, chapter 1) tide station.

Chapter 5

Fig.5.8 Nao99b tide comparison at the Hiron point tide station

5.8 Simulation and result discussion

In this chapter, I have investigated the storm surge due to the cyclone 1991 for comparing the result with the FDM (bay-river) simulation. The detail process of FDM (bay-river) simulation is explained in the previous chapter (Chapter-4). To perform the simulation, I have completed to calculate the input data. The input data were, air pressure and wind, which are calculated from the equation (5.27) and (5.29). The simulation was performed from the April 28 at 8.00 AM to April 30 at 8.00 AM in the year 1991. The 1991 cyclone of Bangladesh, known as BOB 01, was the most catastrophic and deadliest tropical cyclone that struck near Chittagong on the southeast side of Bangladesh on April 29, 1991. Due to this cyclone a big economic losses and human death was occurred. About 138,000 people were died and 1.72 billion economic losses occurred. The first circulation was formed on 22 April 1991 near the equator in the Indian Ocean and most of the mass cloud surrounded the Bay of Bengal region. The first circulation of cyclone 1991 found from the satellite picture of NOAA-I1 and GMS-4. After that, the storm then moved northeastward from its genesis position. From the latitude 11.80°N to 13° N, the storm was a cyclonic storm, having a central pressure of 996 hPa. By this time, on April 27, the system reserved its intensity until April 28

Chapter 5

(UTC 6.00) and changed its nature as a severe cyclonic storm. The system was then the forwarded slightly to the north and northeast side with increasing the central pressure 938 hPa. Finally, the storm crossed the coastal area of Bangladesh with a 6-m surge height on 30 April at 2:00~3.00 a.m. local time. To find the simulation result of the cyclone 1991, I have set the simulation condition (see Table 5.1)

Table 5.1 Simulation condition of FVCOM

Initial tide level 0m Horizontal mesh size 0.05~16km Sea water density 1025.99 km/m3 (200 C, 35psu) Vertical layers 3 Meteorological condition Typhoon model (input time interval is 10min) Calculating time interval 0.3s Calculating time range 28 Apri 8.00 AM to 30 April Open boundary condition Nao.99b tide

FVCOM simulates the surge height by using the external forces data. FVCOM is not the model intended for simulating surge height thus the simulation can be performed only with constant atmospheric pressure. In the first step of simulation, the topographical input data on computational domains were created by SMS (Surface-water Modelling System). After that, the external forces estimated form the pressure wind model and then run the FVCOM. Finally, SMS was used to visualize and analyze the outputs from FVCOM with various visualization tools such as contours and vectors. In the model simulation, I have found the maximum surge height 6m near the Chittagong and Cox-Bazar at the time of 43 hours. Fig. 5.9 shows the detail distribution result of surge height at the hour of 43 and 48.

Chapter 5

Fig.5.9 Storm surge height at the simulation time of 43 and 48 hour, upper figure (a) shows the surge height when the maximum surge was found at 43h and figure (b) shows the surge height when the cyclone eye point crossing the coastline at 48h

To make clear the distribution result in the individual location, the time series result of different tide station are represented in the Fig. 5.10. This figure shows the time series of water level elevation that simulated by the FVCOM model and the FDM (bay-river) model. The station location of this figure is presented in the Fig.1.1 in chapter-1. Fig.5.10 depicts the water level at seven locations from the west part to east part of Bangladesh coast. Early recession of surge levels at the western costal stations are evident from the figures.

Chapter 5

Fig.5.10 FVCOM and FDM (bay-river) Model simulation result of water level elevation at some tide station

The track of the cyclone path is very favorable for high surge along east part of Bangladesh because of the anticlockwise direction of wind movement. Thus, we see that surge intensity is sensitive to the path of the storm. The path of the April 1991 shows that the cyclone crossed the coast just north of Chittagong keeping Sandwip to its left. That’s why, the maximum surge height was found near the Chittagong and its adjacent areas. The comparison of FVCOM and FDM (bay-river) is important for finding the dangerous cyclone. Because, the huge number of surge simulation

Chapter 5

performed by FDM (bay-river) can reduce the time and cost. To understand the comparison clearly, the location-based comparison is shown in Fig. 5.11.

Fig.5.11 Comparison of surge height at two different tide station

From the Fig.5.11, I have found that the water level elevation at some stations makes an acceptable agreement. Though the models are different in nature, they make an acceptable agreement. FVCOM model simulation cover all the factor properly, whereas the FDM (bay-river) model has some limitation on geometric representation. The computation process of FDM (bay-river) is much easier than the computation of FVCOM. So, for the next analysis, I have used the develop bay-river coupled model to find the dangerous track for the future climate scenario. The observed data scarcity is high in the respective department. Due to the weather condition and the analog system of data collection process make this problem higher. So, the model simulation result of this analysis compare with the other study and some limited observed data. Table 5.2 explains the simulation results of different study to compare with the result of FVCOM and FDM (bay-river). Fig. 5.11 shows the comparison of observed and simulation result. Table 5.2 shows

Chapter 5

only the maximum surge height at 17 tide stations among all the coast of Bangladesh.

Table 5.2 Simulation results of different study and the simulation of FVCOM and FDM Coastal Overall Simulated Overall Simulated Simulated Simulated location peak overall peak overall overall overall Surge max. Surge max. peak by peak by level by Surge level by Surge FDM FVCOM Paul et level by Paul et level by (bay- (this al. 2014 Paul et al. al. Roy et al. river) study 2012a 2012b 1991 (this study) Cox’s Bazar ------6.14 5.80 Moheshkhali ------4.12 4.61 Banshkhali ------3.58 6.81 Chittagong 6.25 4.81 5.95 5.45 4.50 6.18 Sitakunda 5.78 ------4.48 5.28 Sandwip 5.63 5.83 5.74 5.33 4.38 5.35 Mirsharai ------5.66 5.05 Companiganj 7.28 7.02 6.90 -- 6.15 5.52 Chital Khali ------4.50 3.82 Char Jabbar 6.35 6.81 6.29 5.18 5.69 4.80 Char changa 5.81 5.49 4.95 4.31 4.12 4.50 Char Madras 5.81 -- 4.97 -- 4.32 4.00 Rangabali 4.50 -- 3.93 -- 4.07 3.92 Kuakata 3.86 3.65 3.30 -- 3.96 4.80 Patharghata ------4.36 4.95 Tiger point 4.57 ------4.21 4.47 Hiron point 4.01 4.52 2.69 0.70 3.80 4.30

So, it is evident that the maximum surge height around the coast of Bangladesh was 3-6 meter due to the cyclone 1991. The observed data also shows the good agreement with model simulation result. The statistical analysis of model simulation result and the observed result shows the good correlation. Due to the lacking of observed data, the correlation coefficient of observed data (one day) at Hiron point is positively correlated with the correlation value 0.78, Root Mean Square Error (RMSE) and Mean Absolute Deviation (MAD) value 0.43 and 0.81 of the FDM (bay-river) simulation. On the other hand, a positive correlation with the correlated value 0.89, RMSE and MAD is 0.24 and 0.58 respectively of the FVCOM simulation.

Chapter 5

Fig. 5.12 Comparison of observed and simulation result

So, from this investigation it is evident that the two models are useful for storm surge simulation. But, FVCOM is useful for the inundation assessment. For the next chapter, I have used both the models for future disaster risk due to the dangerous cyclone.

5.9 Conclusions

Form the above study, it is concluded that the FDM (bay-river) and FVCOM model has a good efficient to simulate the surge accurately. In our previous study, it was found that the bay-river model and the observed data make good agreement. Thus, for the disaster analysis due to the storm surge of future cyclone, the develop model is useful for storm surge simulation. FDM (bay-river) is less complex than FVCOM, but FVCOM can measure the inundation properly. Due to the capacity of computation, FVCOM will be useful for accurate measure of inundation in any complex coast.

Chapter 6

Chapter6

Disaster risk of future cyclone inundation

6.1 Chapter summary

In this chapter, disaster risk of storm surge inundation due to the future cyclone is investigated along the east zone of Bangladesh coast. Most of the severe cyclone of present and future climate is investigated to find the dangerous situation of surge level. Some different cases of cyclone characteristics are also investigated for the future disaster risk analysis. The brief concept of cyclone characteristics that collected from the previous chapters is used to simulate the surge for finding the dangerous condition. Finally, an inundation risk is investigated for the newly establish settlement area of Rohingya (people come from Myanmar) people at Bhashan Char and the economic zone of Chittagong, Bangladesh. This analysis of future disaster risk may helpful to the government of Bangladesh for the mitigation of future disaster risk and a good adaptation policy. According to this study, the newly settlement area at Bhashan Char is more vulnerable for the future cyclone disaster risk.

6.2 Objective

Bangladesh, one of the most disaster prone and most crowded nations. Government plan to develop an isolated, flood-prone island in the inner most side of the Bay of Bengal to make some temporarily house for the thousands of Rohingya Muslims fleeing violence in neighboring Myanmar. Government wants to settle them in the newly born island Bhashan Char, though some people say that the island is not be a suitable place for them. This island is the most cyclone lashed island and prone for flood inundation. Bhashan Char is a floating island emerged from the slit around 15-20 years ago and the regular flood occurred during monsoon season (June- September). Chinses and British engineers build 13 km embankment to protect the specified area from the regular flood. However, this embankment may not be protecting the flood of dangerous cyclone in future. Though the island is 13 square

Chapter 6

km, only the 6.7 square km used for housing. Fig. 6.1 shows the changes pattern of the new island Bhashan Char.

Fig. 6.1 Land features of Bhasan Char

To protect the inside house, 32 meter wide 3.47 meter high embankment was made by the Bangladesh government through a Chinese construction company Sinohydro. Though, the embankment has an anti-erosive layer, a massive cyclone destroy the erosive protection due to a heavy torrential rain. In the future climate scenario, the characteristics of cyclone will be changed. Therefore, an appropriate analysis is crucial for addressing flood disaster. It is the time decision of how much this flood control embankment will play a significant role. Fig.6.2 provides complete information about the embankment that will help to understand the matter. From this figure it is found that the nearest foreshore distance 300 to 400 meter, the length of the embankment is 13 km and breadth of the embankment 32m, an average height of the embankment is 3.47m (from the study of Sinohydro company). A sustainable plan and future risk review for the different climate scenario is very important for the safety and economic development of the coastal region.

Chapter 6

Fig. 6.2 Embankment condition of Bhasan Char

The purpose of this research is to clarify the risk of flood caused by the surge of future storms and to help the Government of Bangladesh for a sustainable contemporary planning. 6.3 Dangerous cyclone analysis

6.3.1 Water level elevation analysis of severe storms in different conditions

Based on the different cyclone characteristics, such as landfall angle, parallel and angular changes of landfall, different wind speed and different translation speed of storms, the height of the storm surge in the eastern region of Bangladesh has been assessed, which helps to find a dangerous cyclone along this area.

Chapter 6

Fig.6.3 Different angle track and parallel track associated with their surge height

Fig. 6.3 shows that the storm tracks of 5-degree angle of each track and 20 km parallel tracks. The maximum storm surges were found for the 85-95 degrees angle track. On the other hand, the parallel track of original track, which is 60 km far from the original one at the southwest mostly in the west side, is more dangerous cyclone along this region. Likewise, in this study, from the analysis of different wind speed and translation speed, it is found that the cyclone of 60-70 m/s wind speed is the most dangerous cyclone for the east coast of Bangladesh and the rapidly moved cyclone is also dangerous. Although, the surge height depends on the landfall location and continuous change of the cyclone position, yet this research adds a different dimension on it. In this study, there has been an analysis of the storms that have crossed the coast faster. Fig 6.4 shows the height of storm surge due to the changes wind speed and changes translation speed.

Chapter 6

Fig.6.4 Surge height for the case of different wind speed and translation speed

Finally, from this discussion it is seen that the storm strike the coast at 85-95 degree angle, the storm that has 60-70 m/s wind speed, the storm coming from southwest mostly in west side and the storms crossed the coast faster are risky cyclone for the east zone of Bangladesh coast. To find the disaster risk due to the severe cyclone of different climate scenario, I have investigated the cyclone, which has lowest central pressure and strike at the east zone of Bangladesh coast. In this study, I have considered the minimum central pressure in its course as a severe cyclone. I have investigated the severe cyclone for the present climate scenario and future climate scenario. For the present climate scenario, I have simulated the storm surge for the present climate simulation of AGCM, present climate simulation of AGCM m0 case, present climate simulation of d4PDF, observed best track data of Bangladesh Meteorological Department (BMD), and JWTC. For the future climate scenario, I have simulated the future climate simulation of AGCM- c2 case, future climate simulation of AGCM-c3 case, future climate simulation of d4PDF-CC, future climate simulation of d4PDF-GF ,future climate simulation of d4PDF-HA, future climate simulation of d4PDF-MI, future climate simulation of d4PDF-MP, future climate simulation of d4PDF-MR. Fig. 6.5 shows the severe cyclone track in different climate scenario.

100

Chapter 6

Fig.6.5 Severe cyclone track in the different climate scenario

Cyclone disaster risk due to the severe cyclone of different climate scenario was investigated from the simulated surge height. To understand the future risk of dangerous cyclone I have separated the track from each scenario based on the lowest central pressure. The separated track are shoen in the Fig. 6.5. By using this extracted track, I have simulated the surge height along the coast of Bangladesh and found the dangerous situation of storm surge. For the MR and CC scenario, the maximum surge height was found in the east zone of Bangladesh coast. For the present climate scenario, cyclone 1991 was the severe cyclone. But, in the future scenario, the surge height will be increase due to the lowest central pressure increase. So that, the east coast of Bangladesh will be faced dangerous cyclone disaster. The detail investigation of dangerous cyclone and its associated inundation analysis is given below section 6.3.2. After analyzing the surge height, I have taken the characteristics of MR scenario to find the future dangerous cyclone. From the all scenario, I have found the maximum surge height 10.3m for the MR scenario.

101

Chapter 6

Fig.6.6. (a) Time series of water level elevation for different severe cyclone, (b) Maximum surge height for different severe cyclone

The time series result of water level elevation due to the severe cyclone of different climate scenario is given in Fig. 6.6.

To find the disaster risk due to the surge height, it is important to investigate the influence of water depth on surge height. Therefore, I have investigated the surge height due to the changes of water depth in different cases. I have found that the water depth has some influence on water level elevation due to a storm. In this study, I have consider five cases and the conditions are: 1m increases of bathymetry, 2m increases of bathymetry, 1m decreasing of bathymetry, uniform bathymetry 3m at the target area, and uniform bathymetry 5m at the target area.

Fig.6.7. (a) Time series of water level elevation at different bathymetry condition, (b) Maximum surge height of different cases

102

Chapter 6

Fig. 6.7 shows the surge height in different cases. From this analysis, it is found that the bathymetry has a noticeable influence on water level elevation due to storm. In this study, I have found that the surge height increasing when the bathymetry (water depth) increasing. Similarly, when the water depth decreases then the surge height also decreases. However, for the uniform depth, surge height is dramatically changed. For the uniform 3 meter depth, I have found the lowest surge height. So, for the counter measure of storm surge, a proper planning of bathymetry and bottom topography is important. 6.3.2 Inundation risk analysis for different conditions

In this section, I have investigated the inundation risk due to the different case of cyclone characteristics changes. Due to the climate change impact, different types of future cyclone characteristics change are noticed. In this section, I have considered seven extreme cases of cyclone characteristics that changed by the climate change impact. The considered characteristics are collected from the different climate scenario of CC, GF, HA, MI, MP, and MR and modified the 1991 cyclone track. Such cases are; original cyclone of 1991, parallel track of 1991 cyclone that 60 km far at the east side, parallel track of 1991 cyclone that 60 km far at the west side, 15 degree landfall angle fluctuation from the original track, case4 for south-west track that mostly in west,

Fig.6.8 Time series result of storm surge height for the different conditions storm surge simulation, where the left figure (a) without tide simulation and (b) with tide simulation

103

Chapter 6

and case5 for south-west track that mostly in south. Another two cases are considered for central pressure added with15hPa, and 15hPa eliminate from original central pressure. From those considerations, an inundation risk was investigated along the east zone of Bangladesh coast and the important island (Bhashan Char). The maximum surge height at Cox Bazar tide station was investigated for the different conditions (see, Fig 6.8). The physical view of inundation risk area and the maximum inundation for different case simulation is shown in Fig.6.9, Fig.6.10 and Fig.6.11. To find the flood risk area from the presented area, I have used the 120m unstructured mesh scheme. Both the target area is important to their own situation. The Bhashan Char is important for the development of newly settlement area for Rohingya people, and the economic city of Bangladesh called Chittagong is also important for the development of economic activity of Bangladesh. To find the future disaster risk along this region due to the future dangerous cyclone should be investigated. FVCOM model was used to simulate the surge height for different condition of cyclone characteristics. A well-known cyclone 1991 was modified by the changes characteristics of future cyclone.

Fig. 6.9 Physical view of inundation risk area

104

Chapter 6

I have found the maximum surge height at 43h simulation. Therefore, the inundation risk was investigated at the same time for different cases (see Fig. 6.10). In this study, I have found that the parallel track, 60km far from the original track that mostly in the west side is more dangerous. In the similar fashion, I have found the dangerous situation for another two cases. Thus, the dangerous conditions are found as: the parallel track that 60km far from the 1991 track mostly in the west side ,

Fig. 6.10 Inundation risk at Bhashan Char

105

Chapter 6

Fig. 6.11 Inundation risk at Chittagong

15 degree above landfall angle that mostly strike from the west side, and the cyclone central pressure that more than 15 hPa central pressure from original track. Finally, for the future disaster-risk analysis, I have taken all the dangerous characteristics to change the severe cyclone 1991 and consider this cyclone as a dangerous cyclone to future disaster risk analysis. The inundation risk was investigated from the surge simulation of modified dangerous cyclone. Fig.6.12 shows the time series result of present climate sever cyclone 1991 simulation and

106

Chapter 6

the modified cyclone 1991. An inundation risk was investigated due to the present climate severe cyclone and future dangerous cyclone along the east zone of Bangladesh coast. It is also investigated the future risk of newly established Rohingya settlement area of Bhasan Char. I have found some problems in the present project in Bhasan Char. The project has a great fault on embankment planning. For the future disaster risk, the project should be needed further development. In this study, I have investigated the inundation risk at Chittagong and Bhashan Char area. Where, I have found the inundation conditions in the different time of cyclone position. Fig. 6.13 shows the inundation situation at Bhashan Char due to the present climate severe cyclone and the future dangerous cyclone.

Fig.6.12 The figure shows the maximum water level due to the present climate severe cyclone and future dangerous cyclone, where left figure shows the water level without tide and the right side figure show the water level with tide

107

Chapter 6

Fig.6.13 Analysis of inundation risk due to present climate severe cyclone and future dangerous cyclone. The left side figures show the inundation condition at the time of 38 hour, 43 hour and 47 hour for dangerous cyclone. The right side figures shows the inundation condition at the same time for present climate severe cyclone

So, it is found that the future dangerous cyclone will destroy the island completely. Fig.6.13 shows the inundation condition at the time of 38h, 43h, and 47h. I have found that the embankment plan makes the flood secure area more dangerous. When a cyclone passes away the island then the inside, the area of the embankment seems like a pond. So, for the flood risk reduction, the embankment planning should be improved, or construct some floodwater drainage system.

108

Chapter 6

Fig.6.14 Estimation of flood risk area for the present severe cyclone and future dangerous cyclone. Left side figures show the different time of inundation condition due to a dangerous cyclone. The right side figures show the flood risk area due to present climate severe cyclone

Fig. 6.14 shows the same analysis for the Chittagong area, which is an economic zone of Bangladesh. It is found from the above analysis that the maximum flood occurred at the time of 43hour-simulation. Where, most of the area near the coast was inundated due to dangerous cyclone. Though the flood decreases from the inundated area, but I have found that the new area of inner most island was flooded at the time of 47 hour-simulation. However, this newly flooded area is not so risky. On the other hand, the present climate severe cyclone has some flood impact on the east side of the karnafully river. So, for the future cyclone flood risk will be higher along this area. The most riskable area will be east part of the Karnafully River and the Chittagong international airport area. In this study, I have also investigated the embankment impact of Rohingya settlement area in Bhasan Char due to the present climate severe cyclone and future dangerous cyclone. I have found the flood risk 109

Chapter 6

due to present climate severe cyclone in the Bhasan Char area. However, this risk may reduce if the embankment height will increase. In this study, I have found that the southern part of the island is more risky area due to a storm. Fig. 6.15 shows the flood risk at the different embankment height. This analysis was done for the present climate severe cyclone. To understand the future dangerous cyclone impact on the Rohingya settlement area, I have simulated the dangerous cyclone and measure the flooded area at two different embankment heights. Fig. 6.16 shows the inundation risk at two different embankment heights due to a dangerous cyclone. So, the island is more hazardous for future dangerous cyclone and its associated flood. Thus, Bangladesh government should be more observant to settle the Rohingya people. A proper planning may improve the safety of the settlement area, such as, improvement of the embankment height, installation of the proper drainage system, to overcome the waterlogging, some water-pump should be establish. Finally, this settlement area is good for Rohingya people but there is some risk of future dangerous cyclone.

Fig.6.15 Inundation risk assessment due to the present climate severe cyclone for the different embankment height

110

Chapter 6

Fig.6.16 Inundation risk assessment due to the future dangerous cyclone for the different embankment height

6.4 Conclusions

Proper planning and appropriate measures to reduce the disaster caused by storms can greatly reduce the risk of flooding. It is important to properly investigate the right plans. My research is to measure the flood prone areas caused by the horrific storms in the future. I have discussed the flood disaster and its remedies caused by the present and future dangerous cyclone at the east zone of Bangladesh coast and Rohingya settlement area. I have found that the future dangerous cyclone can be a terrible catastrophe in this region. However, horrible disasters can be reduced by increasing the height of flood control dam at Bhasan Char and developing some water draining pump to reduce water logging, and make some multipurpose cyclone shelter center to evacuate the victim people quickly.

111

Chapter 7

Chapter7

Conclusions

This study discusses the disaster and its impact due to the current and future cyclone in Bangladesh. This study has reviewed the trend of cyclone strike in the future along the coast of Bangladesh and Bay of Bengal. Apart from this, a bay- river couple numerical model has been developed to analyze the effect of river on the water level elevation due to a storm. Finally, an inundation survey and future disaster risk has been done for the most violent storms of the future. In view of the above work, it is found that the number of storms in the Bay of Bengal will increase in the future, but the number of violent storms will be slightly reduced. However, those cyclones, which are created in the Bay of Bengal, in between them, the possibility of severe storms to hit the coast of Bangladesh is too much. So it is assumed that in the future, less severe storms will turn into depression and cause heavy rainfall. Moreover, this heavy rainfall will be seen across the eastern suburbs of India and the western part of Bangladesh. For the coast of Bangladesh, the western side will be much more prone to storm than comparatively middle and eastern coasts. However, sometimes some dangerous cyclones can hit the eastern coast of Bangladesh. From this study, it is also found that the strike angle characteristics of a cyclone will be slightly changed in future, and the change will occur from southeast to southwest side. The path deviation of a cyclone will be higher for the ordinary cyclone but the path deviation will be less for intense cyclone. The center pressure of a cyclone will be lower in the future climate but in the current situation, the deviation of central pressure is negligible. A bay-river couple model has been developed to investigate the storm surge of present and future cyclone, which incorporates the river dynamics. From this developed model, it is observed that the surge height reduces around the bay-river junction area, because the inland penetration of huge water through the river channel. However, due to excessive amount of fresh water discharge, height may increase. Therefore, due to storms in the rainy season of Bangladesh, the impact of flooding in this region has increased. However, due to the limitation of this model to measure the

112

Chapter 7

flooded area, the FVCOM model has been used to measure the floodplain area. Before measuring the flooded area, the results of the FDM model and the FVCOM model have been compared because the FDM model can reduce time and cost to investigate the dangerous cyclone from the huge number of surge simulation. The observed result and the simulated result of FDM and FVCOM make a good agreement. For the future dangerous cyclone, I have used the summarized characteristics of the future severe cyclone in a known cyclone. Finally, I have found the inundation risk along the east part of Bangladesh coast. I have found that the Rohingya settlement area of Bhashan char and the Chittagong international airport is risky area for the future dangerous cyclone. To settle the Rohingya people in the Bhashan Char, Bangladesh government should take some policy to prevent the future flood risk along this area. A proper planning may improve the security of the Rohingya settlement area, such as, improving embankment height, set-up a proper drainage system, To overcome the waterlogging, some water-pump should be establish, make a modern warning system and multipurpose cyclone shelter center. Finally, this settlement area is good for Rohingya people but there is some risk of future dangerous cyclone. The major summary of this thesis are as followings:

1. The storms of the Bay of Bengal are more likely to cross the eastern coast of India, but the possibility of severe storms falling into the Bangladesh coast is much higher. 2. The west coast of Bangladesh will face much more storms in the future, but there will be dangerous cyclones, which strike in the east coast of Bangladesh. 3. The future cyclone of Bangladesh will be more intense than the present cyclone. However, due to less severe storms as like a depression, the east coast of India and the western part of Bangladesh will precipitated due to heavy rainfall. 4. Seasonal changes of future cyclone of Bangladesh can be noticed due to climate change. 5. Due to the inland penetration of water through the big river, the surge will decrease near the bay-river junction area.

113

Chapter 7

6. For the future dangerous cyclone, the inundation risk in the east part of the Karnaphuli River and Chittagong airport areas will be higher. 7. Bangladesh government should need to take some plan and policy for the new settlement areas of Rohingya, to combat the risk of future climate change and violent cyclone.

114

Reference

Alam, S.M.N., Demaine, H. & M.J. Phillip. (2002): Landuse diversity in south western coastal areas of Bangladesh. The Land, 6.3: 173–184.

Ali A. (1999): Climate change impacts and adaptation assessment in Bangladesh, Clim. Res., 12, 109–116.

Agnihotri, N., Chittibabu, P., Jain, I., Sinha, P.C., Rao, A.D., Dube, S.K. (2006): A Bay River coupled model for storm surge prediction along the Andhra coast of India. Nat. Hazards, 39, 83–101.

Blumberg, A.F., Mellor, G.L., (1987): A description of a three-dimensional coastal ocean circulation model. In Three-dimensional Coastal Ocean Model, N.S. Heaps, Ed., Coastal. Estuar. Sci., 4, 1-6.

Blumberg, A. F., (1994): A primer for Ecom-si. Technical Report of HydroQual, Inc., pp.66.

Cahoon, D.R., White, D., Lynch, J.C.(2011): Sediment infilling and wetland formation dynamics in an active crevasse splay of the Mississippi River delta, Geomorphology, 131, 57–68, doi:10.1016/j.geomorph.2010.12.002.

Chen, C., Liu, H.(2003): An unstructured grid, finite-volume, three-dimensional, primitive equation ocean model: Application to coastal ocean and estuaries, J. Atmos. Ocean. Technol., 20, 159–186.

Chen, C, Beardsley, R. C., and Franks, P. J. S., Keuren,. J. V., (2003b): Influences of the diurnally varying heat flux on stratification and residual circulation on Georges Bank, J. Geophys. Res., 108(C11), 8008, DOI 10.1029/2001JC001245.

Chen, Cowles, C, G., and Beardsley, R. C., (2004b): An unstructured grid, finite- volume coastal ocean model: FVCOM User Manual. SMAST/UMASSD Technical Report-04-0601, pp183.

Chen, C., Huang, H., Beardsley, R. C. , Liu, H., Xu, Q., and Cowles, G., (2006a): A finite-volume numerical approach for coastal ocean circulation studies: comparisons with finite difference models, Journal of Geophysical Research, in press.

Chen, C., Beardsley, R. C., Huang., H., Qi, J., Cowles, G., Gao, G., and Lin, H., (2006b) : A spherical coordinate version of FVCOM: validation and application, Journal of Atmospheric and Ocean Technology, to be submitted.

Chen, (2013): An Unstructured Grid, Finite-Volume Community Ocean Model FVCOM User Manual, Fourth Edition, Marine Ecosystem Dynamics Modelling Laboratory.

Debsarma S.K. (2009) : Simulations of storm surges in the Bay of Bengal, Mar. Geod, 32, 178–198. Davies, A. M. (1976): A numerical model of the North Sea and its use in choosing locations for the deployment of offshore tide gauges in the JONSDAP ’76 oceanographic experiment, Dtsch. Hydrogr. Z., 29, 11-24.

Das P.K. (1972): Prediction model for storm surges in the Bay of Bengal, Nature, 239, 211–213.

Das P.K., Sinha M.C. and Balasubrahmanyam V. (1974): Storm surges in the Bay of Bengal, Quart. J. Roy. Met. Soc., 100, 437–449.

Dube S.K., Sinha P.C. and Roy G.D. (1985): The numerical simulation of storm surge along the Bangladesh coast, Dynam. Atmos. Oceans, 9, 121–133.

Dube S.K. Singh P.C. and Roy G.D. (1986): The numerical simulation of storm surges in Bangladesh using a bay-river coupled model, J. Coastal Engineering, 10, 85-101.

Dube, S.K., Chittibabu, P., Sinha, P.C., Rao, A.D. (2004): Numerical modeling of storm surge in the head Bay of Bengal using location specific model, Nat. Hazards, 31, 437–453.

Dunn, G. E., (1940) : Cyclogenesis in the tropical Atlantic. Bull. Am. Meteor. Soc., 21, 215-229.

Esteban, M., Thao, N.D., Takagi, H., Valenzuela, P., Tam, T.T, Trang, D.D.T, Anh, L.T. (2014): Storm Surge and Tsunami Awareness and Preparedness in Central Vietnam, Coastal Disasters and Climate Change in Vietnam: Engineering and Planning Perspectives, Elsevier, pp. 321-336.

Emanuel, K., (2005): Increasing destructiveness of tropical cyclones over the past 30 years, Nature, 436, 686–688.

Flather, R.A. (1976): A tidal model of the northwest european continental shelf, Mem.Soc.Roy. Sci. Liege,10,141-164.

Flather R.A. (1994): A storm surge prediction model for the northern Bay of Bengal with application to the cyclone disaster in April 1991, J. Phys. Oceanogr., 24, 172–190.

Fujii, T.,(1998): Statistical analysis of the characteristics of severe typhoon hitting the Japanese main islands, Mon. Weather Rev., 126, 1901–1907.

Fujita, T., (1952): Rep. Met. Lab., Kyushu Inst. Tech., Vol.2, No.1-2.

Fujii, T., & Mitsuta, Y. (1986): Simulation of winds in typhoons by a stochastic model, J. Wind Eng. No.28, pp.1-12

Heaps N.S. (1973): A three dimensional numerical model of the Irish Sea, Geophys, J. Roy. Astr. S, 35, 99-120.

Heaps N.S. and Jones J.E. (1981): Three-dimensional model for tides and surges with eddy viscosity prescribed in two layers: II. Irish Sea with bed friction layer, Geophys, J. Roy. Astrom. Soc. 64:303-320.

Hashimoto, N., Yokota, M., Yamashiro, M., Kinashi, Y., Ide, Y., & Kodama, M. (2016): Numerical simulations of storm-surge inundation along innermost coast of ariake sea based on past violent typhoons. Journal of Disaster Research, 11(6), 1221-1227. https://doi.org/10.20965/jdr.2016.p1221.

Hussain, M.A., Tajima, Y., Hossain, M.A., Das, P.(2017): Impact of Cyclone Track Features and Tidal Phase Shift upon Surge Characteristics in the Bay of Bengal along the Bangladesh Coast, J. Mar. Sci. Eng., Vol. 5, pt. 52.

Hirahara, S., Ishii, M., Fukuda, Y.(2014): Centennial-scale sea surface temperature analysis and its uncertainty. J. Clim., 27, 57–75.

Hsu, S. A. and Yan, Z. (1998): A Note on the Radius of Maximum Wind for Hurricanes, J. Coast. Res., 14, 667–668, doi:10.2307/4298820.

Haidvogel, Arango, D. B, H. G., Hedstrom, K., Beckmann, A., Rizzoli, P. M., Schepetkin, A. F., (2000): Model evaluation experiments in the North Atlantic Basin, Simulation in nonlinear terrain-following coordinates, Dyn. Atmo.Oceans., 32, 239-281.

Holton, J.R. (2004): An introduction to dynamic meteorology, Elsevier Academic Press.

Imada, Y., Maeda, S., Watanabe, M., Shiogama, H., Mizuta, R., Ishii, M., Kimoto, M.(2017): Recent Enhanced Seasonal Temperature Contrast in Japan from Large Ensemble High Resolution Climate Simulations. Atmosphere , 8, 57.

Jain, S.K., Agarwal, P.K., Singh, V.P., (2007): Hydrology and water resources of India, Springer, Dordrecht, p 308

Jelesnianski C.P., (1965): A numerical calculation on of storm tides induced by a tropical storm impinging on a continental shelf, Mon.Wea. Rev., 93, 343-358.

Jelesnianski C.P. (1966): Numerical computation of storm surges without bottom Stress, Mon. Weather Rev., 94, 379-394.

Jelesnianski C.P. (1967): Numerical computation of storm surges with bottom Stress, Mon. Weather Rev., 95: 740-756.

Johns B. and Ali A. (1980): The numerical modeling of storm surges in the Bay of Bengal, Quart. J. Roy. Met. Soc., 106, 1-18.

Johns B., Rao A.D., Dube, S.K. and Sinha P.C. (1985): Numerical modeling of tide surges interaction in the Bay of Bengal, J. Phil. Trans. R. Soc. of London A, 313, 507-535.

Jelesnianski, C.P., Chen, J., Shaffer, W.A. (1992): SLOSH: Sea, Lake and Overland Surges from Hurricanes, Technical Report for NOAA Technical Report, U.S. Department of Commerce, National Oceanic and Atmospheric Administration, National Weather Service (NWS), Silver Spring, MD, USA.

Jain, I., Rao, A.D., Jitendra, V., Dube, SK. (2010a): Computation of expected total waterlevels along the east coast of India, Journal of Coastal Research, vol. 26(4), pp681-687.

Kitamoto, A., Nakahara, M., Washitani, I., Kadoya, T., Yasukawa, M., Kitsuregawa, M. (2009): Information Visualization and Organization for Participatory Monitoring of Invasive Alien Species, Proceedings of the 20th International Workshop on Database and Expert Systems Applications (DEXA) / 1st International Workshop on Database Challenges in Biodiversity Informatics, IEEE, 345-349.

Khan, MNH., Kawasaki, S. (2016): Formation of artificial beachrock towards inhibit of coastal erosion in Bangladesh: a review. The 15th Asian reg. conference on soil mechanics and geotechnical engineering, Japan Geotechnical Society Special Publication. https://doi.org/10.3208/jgssp.TC303-01.

Kawai, H., Honda, K., Tomita, T., and Kakinuma, T. (2005): Characteristic of Typhoons in 2004 and Forecasting and Hindcasting of Their Storm Surges, Technical Note of the Port and Airport Research Institute, No. 1103, pp.34.

Kossin, J. P., Knaff, J. A., Berger, H. I., Herndon,D. C., Cram, T. A., Velden, C. S., Murnane, R. J., and Hawkins, J. D., (2007): Estimating hurricane wind structure in the absence of aircraft reconnaissance, Weather Forecast., 22, 89–101, doi:10.1175/WAF985.1.

Kraus, E. B., (1972): Atmosphere-Ocean Interaction. Clarendon Press, 275pp.

Leipper,D. Volgenau, D.(1972): Hurricane heat potential of the Gulf of Mexico, J. Phys. Oceanogr, 2 , pp. 218-224

Luettich, R.A., Westerink, J.J., Scheffner, N.W.(1992): ADCIRC: An Advanced Three Dimensional Circulation Model for Shelves, Coasts and Estuaries. Report 1: Theory and Methodology of ADCIRC-2DDI and ADCIRC-3DL, Technical Report for Dredging Research Program DRP-92-6, US Army Engineers Waterways Experiment Station: Vicksburg, MS, USA, 1992.

Lynch, D. R., and Naimie, C. E., (1993): The M2 tide and its residual on the outer banks of the Gulf of Maine, J. Phys. Oceanogr., 23, 2222-2253.

Matsumoto, K., Takanezawa, T., and Ooe, M., (2000): Ocean Tide Models Developed by Assimilating TOPEX/POSEIDON Altimeter Data into Hydrodynamical Model: A Global Model and a Regional Model Around Japan, Journal of Oceanography, 56, 567-581.

Mohit, A.A.; Yamashiro, M.; Ide, Y.; Kodama, M.; Hashimoto, N. (2018): Tropical cyclone activity analysis using MRI-AGCM and d4PDF data, In Proceedings of the Twenty-Eighth International Ocean and Polar Engineering Conference, Sapporo, Japan, 10–15 June 2018.

Mohit, M.A.A., Yamashiro, M., Hashimoto, N., Mia, M.B., Ide, Y., Kodama, M. (2018): Impact Assessment of a Major River Basin in Bangladesh on Storm Surge Simulation, J. Mar. Sci. Eng. Vol. 6(3).

Murty T.S., Flather R.A. and Henry R.F. (1986): The storm surge problem in the Bay of Bengal, J. Prog. Oceanogr, 16, 195–233.

Mizuta, R., Oouchi, K., Yoshimura, H., Noda, A., Katayama K., Yukimoto, S., Hosaka, M., Kusunoki , S., Kawai, H., Nakagawa, M. (2006): 20-km-mesh global climate simulations using JMA-GSM model—mean climate states, J Meteorol Soc Jpn, 84, 165-185.

Mizuta, R., Yoshimura, H., Murakami, H., Matsueda, M., Endo, H., Ose, T., Kamiguchi, K., Hosaka, M., Sugi, M., Yukimoto, S.,(2012): Climate simulations using MRI-AGCM3.2 with 20-km grid, J. Meteorol. Soc. Jpn., 90A, 233–258.

Mizuta R., (2017): Over 5000 years of ensemble future climate simulations by 60 km global and 20 km regional atmospheric models, Bull. Amer. Meteor. Soc., DOI: http:// dx.doi.org/10.1175/BAMS-D-16-0099.1.

Murakami, H., and Coauthors,(2012a): Future changes in tropical cyclone activity projected by the new high-resolution MRI-AGCM. J. Climate, 25, 3237– 3260, doi:10.1175/JCLI-D-11-00415.1.

Mohit, A.A., Ide., Y., Yamashiro, M., Hashimoto, N.,(2017): A comparative study of two different numerical methods on storm surge. In Asian and Pacific Coasts 2017, Proceedings of the 9th International Conference on APAC 2017, Pasay, Philippines, 19–21 October, 2017; World Scientific Publishing Co. Pte Ltd.: Singapore, pp. 163–174.

Myers, V. A., and Malkin, W., (1961): Some properties of hurricane wind fields as deduced from trajectories, U. S. Weather Bureau, National Hurricane Research Project, Report 49.

Nakajo, S., Mori, N., Kim, S. Y., Yasauda, T., Yamada, F., and Mase, H. (2014): Basic study on estimation method of return period and variation range of severe storm surge event, Proceedings of 34th Conference on Coastal Engineering, pp.11.

Naimie, C. E., (1996). Georges Bank residual circulation during weak and strong stratification periods: prognostic numerical model results, J. Geophys. Res. 101(C3), 6469-6486.

Needham, H.F, Keim, B.D, Sathiaraj,D. (2015): A review of tropical cyclone generated storm surges: Global data sources, observations, and impacts, Reviews of Geophysics, https://doi.org/10.1002/2014RG000477 ‐

Outline NWP March. (2007): Available online: http://www.jma.go.jp/jma/jma eng/jma center/nwp/outline-nwp/index.htm (accessed on 30 December 2016).

Oouchi, Yoshimura, K. J. , Yoshimura, H. , Mizuta, R. , Kusunoki, S. , Noda, A.(2006) : Tropical cyclone climatology in a global warming climate as simulated in a 20 km-mesh global atmospheric model: Frequency and wind intensity analysis, J. Meteor. Soc. Japan, 84, 259–276, https://doi.org/10.2151/jmsj.84.259.

Paul G.C. and Ismail A.I.M. (2012a): Numerical modeling of storm surges with air bubble effects along the coast of Bangladesh, Ocean Eng., 42, 188–194.

Paul G.C. and Ismail A.I.M. (2012b): Tide–surge interaction model including air bubble effects for the coast of Bangladesh, J. Franklin Inst. , 349, 2530– 2546.

Paul G.C. and Ismail A.I.M. (2013): Contribution of offshore islands in the prediction of water levels due to tide–surge interaction for the coastal region of Bangladesh, Nat. Hazards, 65, 13–25.

Paul G.C., Ismail A.I.M. and Karim M.F. (2014): Implementation of method of lines to predict water levels due to a storm along the coastal region of Bangladesh, J. Oceanoger., 70, 137-148.

Peng, M., Xie, L., Pietrafesa, L.J. (2004): A numerical study of storm surge and inundation in the Croatan–Albemarle–Pamlico Estuary System, Estuar. Coast. Shelf Sci., 59, 121–137.

Pedlosky, J., (1974): Longshore currents, upwelling and bottom topography, J. Phys. Oceanogr., 4, 214-226.

Quiring, S., Schumacher, A., Labosier, C., and Zhu, L., (2011): Variations in mean annual tropical cyclone size in the Atlantic, J. Geophys. Res., 116, D09114, doi:10.1029/2010JD015011.

Rahman M.M., Hoque A., Paul G.C. and Alam M. J. (2011a): Nested numerical schemes to incorporate bending coastline and islands boundaries of Bangladesh and prediction of water levels due to surge, Asian J. of Math. and Statistics, 4, 21-32.

Roy G.D. (1995): Estimation of expected maximum possible water level along the Meghna estuary using a tide and surge interaction model, J. Environ Int., 21, 671 67.

Roy G.D., Kabir A.B.M.H., Mandal M.M. and Haque M.Z. (1999c): Polar coordinate shallow water storm surge model for the coast of Bangladesh, Dyn. Atoms. Ocean, 29, 397–413.

Roy, G.D., (1999): Inclusion of off-shore islands in transformed co-ordinates shallow water model along the coast of Bangladesh, J. Environ. Int., 25, 67–74.

Roy G.D. (1999a): Inclusion of off-shore islands in transformed co-ordinates shallow water model along the coast of Bangladesh, J. Environ Int., 25, 67–74.

Rego, J.L., Li, C.(2009): On the receding of storm surge along Louisiana’s low- lying coast, J. Coast. Res., 56, 1042–1049.

Reinhardt E. P., Kenshi H., Izuru T., and Tosiyuki N.,(2017): Iirtually experiencing future climate changes in Central America with MRI-AGCM: climate analogues study, Hydrological Research Letters 11(2), 106–113 (2017) Published online in J-STAGE (www.jstage.jst.go.jp/browse/hrl). DOI: 10.3178/hrl.11.106.

Rao, A.D., Jain, I., Venkatesan, R.(2010): Estimation of Extreme Water Levels Due to Cyclonic Storms: A Case Study for Kalpakkam Coast, International Journal of Ocean and Climate Systems, vol. 1 (1). 14pp.

Sartor, J. (1968): Monthly climatological wind fields associated with tropical storm genesis in the West Indies. Colorado State Univ., Ft. Collins, Atmospheric Science Research paper, 34 pp.

Smith D.K. (1989): Natural disaster reduction: How metrological and hydrological services can help. World Metrological Organization, Geneva Publication no. 722.

Sheng, Y.P.(1987): On modeling three-dimensional estuarine and marine hydrodynamics. Elsevier Oceanogr. Ser.1987, 45, 35–54.

Sheng, Y.P., Paramygin, V.A. (2009): Forecasting storm surge, inundation and 3D circulation along the Florida coast, In Proceedings of the 11th International Conference on Estuarine and Coastal Modeling, Seattle, DC, USA, 4–6 November 200, pp. 744–761.

Simpson, J. J. and Dickey, T. D., (1981a): The relationship between downward irradiance and upper ocean structure, J. Phys. Oceanogr., 11, 309-323.

Thacker W.C. (1977): Irregular grid finite difference, Simulation of oscillation in shallow circular basin, J. Ohys. Oceanogr, 7, 284-292.

Takagi, H., Thao, N. D. and Esteban, M. (2014) : Tropical Cyclones and Storm Surges in Southern Vietnam, Coastal disasters and climate change in Vietnam, Engineering and planning perspectives, Elsevier, pp.3-16.

Talukder , J., Roy G.D. and Ahmed M. (1992): Living with cyclone. Community Development Library, Dhaka.

Thompson, SL., Ramaswamy,V., Covey,C. (1987): Atmospheric effects of nuclear war aerosols in general circulation model simulations: influence of smoke optical properties, J Geophys Res, 92, 10942–10960.

Takagi, H., Nguyen, D. T., Esteban, M., Tam, T. T., Knaepen, H. L., and Mikami, T., (2012): Vulnerability of coastal areas in Southern Vietnam against tropical cyclones and storm surges , The 4th International Conference on Estuaries and Coasts (ICEC2012), pp.8.

Tasnim K. M., Shibayama ,T., Esteban, M, Takagi, H, Ohira, K., and Nakamura, R. (2014): Field Observation and Numerical Simulation of Past and Future Storm Surges in the Bay of Bengal: Case Study of Cyclone Nargis, Nat Hazards, DOI 10.1007/s11069-014-1387-x.

Walker, N. D. ( 2001): Tropical storm and hurricane wind effects on water level, salinity, and sediment transport in the river-influenced Atchafalaya- Vermilion Bay System, Louisiana, USA. Estuaries, 24, 498-508.

Wachtmann, R. F., (1968): Role of angular momentum transport in tropical storm dissipation over tropical oceans. Colorado State University, Fort Collins, Abmospheric Science Research Paper, 46 pp.

Williamson, DL., Kiehl, JT., Ramanathan, V., Dickinson, RE., and Hack, JJ. (1987): Description of NCAR Community Climate Model (CCM1), NCAR Technical Note NCAR/TN-285+STR National Center for Atmospheric Research, Boulder, Colorado.

Yamashiro, M., Tanabe, T., Shimada, G., Yokota, M., and hasimoto, N. (2014) :A study on possible critical storm surge under present climate condition in the Ariake sea japan, proc. Of the int. conf. on coastal engineering (ICCE 2014) vol. 34.

Yoshimura, H, and Matsumura, T. (2005): A two-time-level vertically- conservative semi Lagrangian semi-implicit double Fourier series AGCM, CAS/JSC WGNE Research Activities in Atmospheric and Ocean Modeling, 35, 3.25–3.26.

Appendix-1

The river model equation are discretized by finite-difference technique (forward in time and central in space) and are solved by conditionally stable semi-implicit method using a staggered grid system Any variable riv at a grid point (i, j) at time tk is represented by: x,, y t  k riv i j k  R i, j Thus, the finite-difference approximated form of the equation 8 is as follows:   riv Hu    Hv   0 t  xriv   y  riv  H(,)[(,)] u v  u v  h riv riv riv riv

kk1 i,, j i j t RTL12  RTL  (a)

uukk vvkk RTL1  R i1, j R i 1, j and RTL2  R i, j i R i , j 1  2x 2y k 1 where, R i, j in equation (a) is computed at i = 2, 4, 6, ..., M–2 and j = 3, 5, 7, ..., N–2. Similarly, the finite-difference form of the Equation (9) as is follows     1/2 Hu Hu2  Huv  Hv   griv  Cuu2  v 2 riv  riv  riv riv  riv f riv riv riv  t  x  y  x

k k 1 Ru i, j   t( RTL 1  RTL 2  RTL 3)   t TR 1  Ru i, j  (b) 1t . RFR 3 

uk U k u k U k RTL1  Ri2, jRi  2, j Ri  2, jRi  2, j  4x

y x y x uk V k u k V k RTL2  Rij, 1 Rij ,  1 Rij ,  1 Rij ,  1 2y 

xy k RTL3 R vij, kk11 RTR1  g R i1, j R i 1, j 2x

xy 2 kk2  CUVf R i j  R i j ,, RFR3  k 1 R i,, j Rh i j

k 1 where, Ru i, j in equation (b) is computed at i = 3, 5, 7, ..., M–1 and j = 3, 5, 7, ..., N–2. In similar way the boundary and other equation are discretized.