FLOOD RISK MAP FOR KUMASI

KWAME NKRUMAH UNIVERSITY OF SCIENCE AND TECHNOLOGY COLLEGE OF ENGINEERING DEPARTMENT OF GEOMATIC ENGINEERING FLOOD RISK MAP FOR KUMASI A Thesis submitted to the Department of Geomatic Engineering, Kwame Nkrumah University of Science and Technology in partial fulfillment of the requirements for the degree of BSc. Geomatic Engineering By: Gideon Nkrumah (4083315) Ibrahim Suleman (9737113) Supervisor: Dr. Akwasi Afrifa Acheampong May 2019 DECLARATION We hereby declare that this submission is our work towards the achievement of BSc. Geomatic Engineering and that to the best of our knowledge, it contains no materials previously published by another person nor material which has been accepted for the award of any other degree of the University, except where due acknowledgment has been made in the text. Nkrumah Gideon ………………………... ………………………….. (4083315) Student’s Names & ID Suleman Ibrahim Signature ………………………... Date .………………………….. (9737113) Student’s Names & ID Certified By: Signature ……………………….... Date …………………………. Dr. Akwasi A. Acheampong Supervisor Signature Date ABSTRACT Flooding is a recurring phenomenon in Kumasi which has caused damage to properties and financial loss over the years. The risk of flooding is projected to increase due to the annual rise in rainfall and conversion of floodplains to settlement. Many kinds of research have been made to assess the risk of flood and map out the vulnerable zones using diverse methodologies. This gives the idea of flood prediction, mitigation, and prevention. To achieve the objective of flood risk assessment in Kumasi, the GIS modeling approach was used to produce a flood risk map with the help of Quantum GIS software. In this research, a digital elevation model which is important data for hydrological modeling was obtained from the Shuttle Radar Topography Mission (SRTM). Four hydrological factors which include digital elevation, slope, proximity to the river, and topographic wetness index were derived from the depressionless DEM of the study area. The hydrological factors were further reclassified and overlaid to produce the flood risk map. The flood risk map showed five risk zones – very high, high, moderate, low, and very low zones in the study area. The results showed that 30% of the study area lies within the very high and high-risk zones. These areas are considered very highly susceptible to flooding. However, only 2% of the very high and high population density areas fall within the high susceptible flood zones. This is because most of the highly susceptible flood zones fall in the southern and north-eastern parts of the study area. These areas cover the very low and low population density regions. Moreover, floods will be likely to occur during the rainy season. To further demonstrate and visualize the risk of flood in the high susceptible zones, a flood simulation method was adopted. With the aid of Quantum GIS geoalgoritms and a digital elevation model, a resampled DEM was used to produce a flood simulation map showing the areas at risk in a given flood depth along the Susan river at Atonsu-Junction in Kumasi. This was further overlaid on a georeferenced google earth image of the area to aid visualize the affected areas. The flood simulation map showed that areas that have experienced extensive flooding fall within the predicted flood depths. The research has proved that the application of GIS for hydrological modeling and flood risk simulation is important for the assessment of flood risk. ACKNOWLEDGMENT We want to acknowledge God for the utmost grace and glory He gives us throughout our daily activities. Our second appreciation goes to our supervisor; Dr. Akwasi Afrifa Acheampong, for his wonderful guidance and supervision throughout this project. We really appreciate the time we had and the data provided to us by the various organizations we approached. They have been really helpful throughout this project. Our special gratitude also goes to our guardians who have supported us throughout our program of study, their encouragement and words of perseverance also saw us through the successful completion of this project. Thank you to all our Lecturers, Technicians, and Teaching Assistants in the Department of Geomatic Engineering for their immerse academic training throughout our years of stay in KNUST. Table of Contents DECLARATION 1 ABSTRACT 2 ACKNOWLEGEMENT 3 LIST OF ABBREVIATIONS 9 CHAPTER ONE 1 INTRODUCTION 1 1.0 Background of Study 1 1.1 Problem Statement 2 1.2 Aims and Objectives 3 1.2.1 Main Objective: 3 1.2.2 Specific Objectives: 3 1.3 Research Questions: 3 1.4 Thesis Outline 3 CHAPTER TWO 4 FLOOD, GIS & RISK MAPPING 4 2.0 Floods 4 2.0.1 Flood Types 5 2.0.2 Causes of Flood in Kumasi 5 2.0.2 Negative Effects of floods in Kumasi 7 2.0.3 Flood Mitigation in Kumasi 10 2.1 GIS in Flood Management 11 2.2 Flood Risk Mapping 11 2.3 Flood Simulation 13 CHAPTER 3 15 METHODOLOGY 15 3.0 Introduction 15 3.1 Study Area 15 3.2 Data materials and Sources 16 3.4 Flow Chart of Methodology 17 3.5 Software and Data Geoprocessing 18 3.5.1 Quantum GIS 18 3.5.2 GRASS GIS 18 3.5.3 Microsoft Word and Excel 2016 18 3.5.1 Digital Elevation Model 18 3.6 Modelling of hydrological factors 20 3.6.1 Digital elevation 20 3.6.2 Slope 20 3.6.3 Topographic Wetness Index (TWI) 20 3.6.4 Proximity (raster distance) of rivers 21 3.6.5 Overlaying of maps using the raster calculator 22 3.7 GPS Field Survey 24 3.8 Population Density map 24 3.9 Rainfall Pattern and Trend 24 3.9.1 Regression Analysis 24 3.10 Field Survey 25 3.11 Flood Simulation 25 CHAPTER 4 26 RESULTS 26 4.0 Flood Risk Mapping 26 4.0.1 Filling of Sink (Depressions) 26 4.0.2 Catchment Area 26 4.0.3 Stream Network 27 4.0.4 Digital Elevation Map 28 4.0.5 Slope Map 29 4.0.6 Topographic Wetness Index Map 31 4.0.7 Proximity/Raster distance of river 32 4.0.8 Results on Weighted Overlay - Flood Risk Map 33 4.1 Population Density Map 38 4.2 Rainfall Pattern and Trend 39 4.3 Field Survey 40 4.4 Flood Simulation 41 DISCUSSIONS 45 4.5 Discussions on Results of Flood Risk Map 45 4.6 Discussion on the Influence of Flood Risk on Population Density 46 4.7 Discussion on Rainfall pattern in Kumasi 46 4.8 Discussion on Flood Simulation 46 CHAPTER 5 48 CONCLUSIONS AND RECOMMENDATIONS 48 5.0 Conclusions 48 5.1 Recommendations 49 References 50 APPENDIX 53 Appendix A: Coordinates of Very High-Risk Zones from GPS Survey (Garmin 62c handheld GPS) 53 List of Tables Table 2. 1 Flood Risk Data sheet for Kumasi 8 Table 3. 1 Data materials and sources Table 3. 2 Formula table in raster calculator Table 3. 3 Weighting of reclassified maps 16 23 24 Table 4. 1 Area of Digital Elevation map classes Table 4. 2 Area of Slope map classes Table 4. 3 Area of Topographic wetness index classes Table 4. 4 Area of Flood risk map classes Table 4. 5 Area of Flood risk and Population Density classes Table 4. 6 Rainfall data of Kumasi from 2005 to 2015 28 29 31 33 37 40 List of Figures Figure 2. 1 Anloga junc on flooded a er con nuous rainfall (authors; 20th May, 2019) Figure 2. 2 Solid and liquid waste in a waterway Figure 2. 3 Solid waste carried by Flood water Figure 2. 4 A bridge being constructed at Daban river to control flood water 4 7 10 10 Figure 3. 1 A map of the study area showing its geographic loca on in Ghana Figure 3. 2 Flowchart of Methodology Figure 3. 3 Flowchart of flood simula on methodology 15 17 25 Figure 4. 1 Map of Depressionless DEM 26 Figure 4. 2 Map of Catchment area 27 Figure 4. 3 Map of Stream Networks in Kumasi 27 Figure 4. 4 Digital Eleva on Model of Kumasi 28 Figure 4. 5 Classified and inversed Digital Eleva on Model of Kumasi 29 Figure 4. 6 Slope map of Kumasi 30 Figure 4. 7 Classified and inversed Slope map of Kumasi 30 Figure 4. 8 Topographic Wetness Index map of Kumasi 31 Figure 4. 9 Classified Topographic Wetness Index map of Kumasi 32 Figure 4. 10 Proximity/Raster distance map of Kumasi 33 Figure 4. 11 Quan ta ve chart of Flood risk zones in Kumasi 34 Figure 4. 12 Flood risk map of Kumasi 34 Figure 4. 13 Vectorised Flood risk map of Kumasi 35 Figure 4. 14 Flood risk map with towns and flood points in Kumasi 36 Figure 4. 15 Chart showing percentage area of flood risk zones for Kumasi submetros 36 Figure 4. 16 Chart showing the percentage popula on density area for each Flood risk zone 37 Figure 4. 17 Popula on density map of Kumasi 38 Figure 4. 18 Mean monthly rainfall graph of Kumasi 39 Figure 4. 19 Annual rainfall pa ern of Kumasi 39 Figure 4. 20 Embankments built to direct runoff water Figure 4. 21 Buildings within river banks 41 Figure 4. 22 Extracted DEM of Susan River, Atonsu-Junc on 42 Figure 4. 23 Resampled DEM of Susan river, Atonsu-Junc on 42 Figure 4. 24 DEM of Susan river, Atonsu-Junc on 43 Figure 4. 25 Flood simula on map of Susan river, Atonsu-Junc on 43 Figure 4. 26 3d visualiza on of flood depth simula on on Susan river, Atonsu-Junc on with a ver cal exaggera on of 4.5 using Qgis2threejs plugin in QGIS 2.18 so ware. 44 LIST OF ABBREVIATIONS GIS - Geographic Information System DEM - Digital Elevation Model QGIS - Quantum Geographic Information System GRASS - Geographic Resources Analysis Support System KMA - Kumasi Metropolitan Assembly NADMO - National Disaster Management and Organization FEMA - Federal Emergency Management Agency GPS - Global Positioning Systems TCPD - Town and Country Planning Department GSS - Ghana Statistical Service SRTM - Shuttle Radar Topography Mission USGS - United States Geological Survey UTM - Universal Transverse Mercator WGS - World Geodetic System NASA - National Aeronautics and Space Administration CHAPTER ONE INTRODUCTION 1.0 Background of Study A flood is a situation in which water temporarily covers land to a certain depth. This water comes from mostly overflown rivers and torrential rainfall and inundates flood-prone areas. Flood risk is the damage that may be expected to occur in a given area arising from flooding. It is a combination of likelihood or probability of flood occurrence, the degree of flooding, and the impacts or damage that the flooding would cause (Kwang et. al, 2017). Kwang et. al, 2017 described urban flooding as one of the world’s problems in recent times because of its frequent occurrence which results in loss of lives and property. Globally, flooding has become one of the common occurrences of natural disasters. As they are naturally occurring, they cannot be prevented and have the potential to lead to fatal causes such as displacement of people and damage to the environment. Every year, this phenomenon causes significant economic losses worldwide. Moreover, according to CRED (2015), it is known that more people are affected by flooding than any other natural disaster. Ghana has experienced a sharp alteration in its weather that has recently resulted in widespread flooding currently reported in over five regions of which four are heavily affected. The floods have caused devastating impacts on people's health, safety, and the destruction of properties and livelihood. The reported flooding across the country is likely to cause devastation that communities and government are unable to cope with and or even recover from it as farmers are losing their investments in production, communities are helplessly being displaced and infrastructure such as roads and buildings are collapsing (Ghana Meteorological Agency, 2017). There is an urgent need to conduct an in-depth assessment to have a better overview of the situation (IFRC 2017). Runoffs in the major rivers and drainage systems are the main driver of flooding in Kumasi. When high rainfall occurs, runoff increases with an increase in velocity and width thereby being the resultant effect of flooding in communities within the banks of these rivers in Kumasi. Data collected between March 2016 to October 2018 from the National Disaster and Management Organization suggest that at least 10 houses are affected whenever there is an incidence of flood (NADMO 2018). The most recent flooding in Kumasi between the periods of May 2018 and October 2018 suggests that approximately Ghc1,04m in cost has been used for flood management by the National Disaster and Management Organization (NADMO 2018). Damage caused by floods in urban areas is extremely high because of the amount and value of the stakes at risk. The reduction of the risk of flooding will depend largely on the amount of information on floods that are available and knowledge of the areas that are likely to be affected during a flooding event. 1.1 Problem Statement Kumasi has been affected by flooding over the years which is mostly caused by heavy downpours and rivers overflowing its channels (KMA, 2018). The rapid increase in population and expansion of settlement in Kumasi will mean high vulnerability risk (GSS, 2018). Thus, an incidence of flooding will endanger a greater number of populations. The undulating nature of the terrain makes low lying areas vulnerable to flooding. In addition to this, river channels have been silted naturally by erosion and artificially by the dumping of both solid and liquid waste in them. The annual rainfall pattern of Kumasi causes major flooding which displaces a lot of houses and incurred cost in managing these floods (NADMO 2018). Roads and buildings constructed in flood-prone areas are exposed to flood. Kumasi is located within the Pra basin and is divided into five sub-basins: Kwadaso, Subin, Aboabo, Wiwi and Susan. The city is drained by a relatively dense network of rivers: Wiwi River, Subin River, Susan River, Suatem River and Daban River. The Subin River flows from Kumasi zoo to Kaase and the Wiwi River also flows from Nsenie into Susan River at Atonsu. The Suatem River flows through North Suntreso to join Kwadaso River at Dakwadwom to form the Daban River. The Subin, Daban and Susan River then joins Oda River, which is a tributary of the River Pra. Flood events that have inundated areas around the above-mentioned rivers have caused property destruction and claimed lives. Information about stream flow direction and its subsequent accumulation can help communities reduce their current and future vulnerability to floods (Fosu et al, 2012). 2 | Page There is little planning to mitigate flooding in Kumasi. There has therefore been the increasing need to research new methods to identify and map flood risk zones. This research will produce a flood risk map of Kumasi to provide information for prediction, mitigation and prevention measures of flooding in Kumasi. 1.2 Aims and Objectives 1.2.1 Main Objective: The overall aim of this research is to provide a flood risk map for Kumasi using Geographic Information System (GIS). 1.2.2 Specific Objectives: From the main objective of the study, we expect to achieve the following specific objectives: ● To map out flood risk zones in Kumasi and classifying them into very high, high, moderate, low and very low. ● To determine the influence of flood risk on population density in Kumasi. ● To demonstrate the risk of flooding in urban areas in Kumasi through flood simulation. 1.3 Research Questions: With reference to the specific objectives, the following aspects must be considered a) What areas in Kumasi are at higher risk of flood? b) How does flood influence the risk on the population in Kumasi? c) How can we visualize the flood risk in urban areas in Kumasi? 1.4 Thesis Outline This report consists of five chapters. Chapter One introduces the research by giving a brief background to flooding in Kumasi. The problem statement, aims and objectives and research questions are discussed in Chapter One. Chapter Two contains the literature review that discusses various themes on flooding, how attempts have been made to 3 | Page mitigate floods and the review of GIS on flood risk mapping and simulation. In the third chapter, the profile of the study area, data materials, software and methodology applied is explained. Chapter four shows the results obtained from the methods applied on the datasets used and the discussions on those results. Chapter six focuses on findings, conclusions and recommendations. CHAPTER TWO FLOOD, GIS & RISK MAPPING 2.0 Floods Floods has been the most recurring and destructive natural disaster which has affected its victims globally. Since it occurs naturally and frequently, mitigation and control measures has been challenging over the years. Floods can occur in different forms, sizes and at various locations with different vulnerabilities, their impacts differ strongly. The occurrence of flooding cannot be prevented but the impacts can however be minimized. A flood is an overflow or irruption of a great body of water over land in a built-up area not usually submerged (Oxford English Dictionary, 2018). This water temporarily covers land to a certain depth coming from overflown rivers caused by torrential rainfall and inundate areas within the floodplains. Urban flooding is one of world’s problems in recent times because of its frequent occurrence which results in loss of lives and properties (Kwang C., Osei E.M., 2017). In the Kumasi Metropolis, it is observed that the terrain is generally undulating, characterized by steady, steep rising areas and valleys. In 2018, the Metropolis reported deaths and displacement of properties due to the flooding in June and September (NADMO, 2018). Most affected areas were Atonsu, Buokrom, Nhyiaeso, Daban, Morshie Zongo, Bantama, Oforikrom, Anloga, Kronom and Kwadaso (NADMO, 2018). These communities are fast growing urban areas and human activities such as dumping of refuse into drains and building in flood plains are common. The areas severely affected such as Anloga-Junction in the figure below were realized to be relatively low terrain areas and mostly have valley-like landscapes therefore causing runoff from upstream to downstream in the cause of either heavy or continuous rainfall. 4 | Page Figure 2. 1 Anloga junction flooded after continuous rainfall (authors; 20th May, 2019) Kumasi Metropolis experience mostly river and flash flooding because of its high urbanization over the years causing most areas to be impervious to rain water especially in low terrains. Determination of flood risk requires the knowledge on the types of flood, causes of flood, flood chances of occurrence, how they can be modelled and mapped and the required datasets for producing a flood risk map. This chapter of the study therefore focuses on issues relating to flooding in the Kumasi Metropolis. Literature materials concern to the study have been reviewed to provide a basis for having an explanation to flooding and its causes, the identification and assessment of flood risk areas and management of floods. 2.0.1 Flood Types Flooding in urban areas can be caused by one or more of the following types and these are flash floods, coastal floods, river floods, groundwater flooding and drain/sewer flooding. ● A flash flood is a type caused by heavy and sudden rainfall; it happens when the ground cannot absorb as quickly as rain falls. The intensity and the occurrence of flash flooding can be determined by the intensity and the distribution of the rainfall, the nature of the topography and land use and soil type. ● Coastal flood is a type of flood that occurs at the coast of sea or low land areas near the sea due to extreme weather and high sea tides. ● River flooding is one of the most common type of inland flood; occurring when a body of water exceeds its capacity. This happens when the river overflows its 5 | Page banks due to high rainfall over a prolonged period of time. Localized flooding can cause considerable damage to surrounding properties. ● Ground water flood occurs when rain falls over an extended period and the ground becomes saturated with water until it cannot absorb anymore. When this happens, water rises above the ground’s surface and causes flooding. ● Drain/Sewer flood occur as a result of blockage or similar failure within the drainage system (FEMA-EMI.). 2.0.2 Causes of Flood in Kumasi Flooding is generally considered as an environmental hazard. It is quite a natural process and is simply the reaction of a natural or man –made system that allow the presence of too much water at a particular period of time. Natural causes of flooding refer to those causes that are not caused by human influence directly. An example of the natural causes of floods includes the nature of the terrain and high rainfall pattern which increases the volume of water in rivers and drains causing overflows of its banks resulting in floodplains being submerged. Human causes of flooding describe the flooding caused by the direct actions and inactions of humans. Below are the main contributors of flooding in Kumasi. Natural causes of flooding: 2.0.2.1 Terrain variation Flooding is common in low terrain areas. The topography of Kumasi in undulating mainly comprising of steep slopes. Rivers flow more slowly in lowland areas, if the water volume increases abruptly or suddenly, floods occur. There is a reduction in the amount of infiltration of water into the ground on steep slopes. This means water can easily flow down to rivers as overland flow. Steep slopes also make it easier for more through flow within the soil. These two situations can both raise river levels easily causing flooding. (Jackson, 2012). 2.0.2.2 High Rainfall trend The incidence of heavy rainfall is generally the cause of flooding in the Kumasi Metropolis. The river and drainage network system in Kumasi usually cannot carry all the water in its channels after a heavy downpour thereby causing flooding. Flooding in 6 | Page Kumasi is more common in the months of May, June and August since the annual rainfall pattern with these months are relatively high (Ghana Meteorological Agency, 2018). High rainfall causes rivers and drainages to overflow its banks leading to flooding in the nearby flood plains. Social or human causes of flooding: 2.0.2.3 Inadequate Drainage networks Most flood cases in Kumasi are influenced by improperly constructed and managed drainage networks. Most river channels in Kumasi are converted to drainage networks. These drains are also choked with refuse and dirt deposits which resist the free flow of moving water in case of heavy rains. There are also less culverts constructed in flood zones, leading to water accumulation at a depth in these areas causing flooding. 2.0.2.4 Waste in water bodies and drains Most rivers and drainage systems have deposits of refuse caused by human activities and silt from untarred roads caused naturally by gully erosion. Most of the rivers and drains are used as refuse damps which includes both solid and liquid waste. These sediments accumulate impedes the passage of water in the rivers and drains resulting in flooding as it can be seen in the figure below around Morshie Zongo area in Ku Figure 2. 2 Solid and liquid waste in a waterway 7 | Page 2.0.2 Negative Effects of floods in Kumasi Floods have numerous negative effects on its victims. This section tends to outline the negative effects of flooding in Kumasi Metropolis. 2.0.2.1 Population and Health Floods poses threat on the health of the population affected in Kumasi. In most cases of flood occurrences, health issues and sometimes deaths are recorded (NADMO, 2018). In the June 2018 flood in Kumasi, five people were reported dead and many others exposed to injuries. Flood waters mix with raw sewage and in effect increases the risk of outbreak of diseases such as cholera and malaria in these areas (KMA, 2018). 2.0.2.2 Financial Effects Flooding in Kumasi has posed a major financial threat on these management organizations and the mass population affected over the years. Businesses close down temporarily and sometimes permanently during and after the floods and can take up to months to recover. This is due to inability to access roads and pathways and the lack of access or failure of basic services (water supply, electricity and telecommunication and infrastructure). Data collected at the National Disaster and Management Organization shows that almost GHS 9.7m has being spent on managing flood in Kumasi between 2016 and October 2018 as shown in the table below. Table 2. 1 Flood Risk Data sheet for Kumasi Flood Data sheet for Kumasi Metropolitan Assembly Date of Occurrence Location 12/3/2016 Pankrono Estate 16/6/2016 Kwadaso North 24/6/2016 Oforikrom 15/7/2016 Adoato 8 | Page Houses Affected Estimated Cost GHS 8,000.00 GHS 3,000,000.00 GHS 178,920.00 GHS 29,000.00 11/9/2016 215 23/4/2016 Gyinase Milano 24/6/2016 Buokrom South Africa 4/6/2017 5/6/2017 Tafo Pankrono, Afful Nkwanta, Suame, Cultural Centre, Suntreso, Krofrom, Buokrom, Dichemso, Yennyawso and Abrepo Abrepo Junction 21/6/2017 Atonsu S-Line 22/6/2017 Kuwait-Abrepo 23/6/2017 5/7/2017 Chirapatre Area and Kentinkrono Suntreso and Abrepo Junction 23/8/2017 Kwadaso North 23/9/2017 17 6/10/2017 Atonsu, Kaase and Kwadaso 8/11/2017 Nsenie 22/11/2017 6/5/2018 Sokoben and Kwadaso Estate Anyinam Newsite and Apraman Nyiaeso Atasomanso Nyiaeso 7/5/2018 Bantama and Nyiaeso 5/5/2018 9 | Page GHS 1,275,000.00 GHS 4,000.00 GHS 50,000.00 GHS 1,075,000.00 GHS 39,000.00 GHS 15,000.00 GHS 40,000.00 GHS 1,805,000.00 GHS 50,000.00 GHS 25,000.00 GHS 1,000,000.00 GHS 61,000.00 GHS 5,000.00 GHS 10,000.00 GHS 30,000.00 GHS 10,000.00 GHS 20,000.00 9/5/2018 Nyiaeso 10/5/2018 28/6/2018 Dichemso, Adoato, Bantama, Ohwim Amanfrom and Abrepo Junction Dichemso and Buokrom 23/9/2018 Atafoa-Bantama 16/10/2018 Old Amakom and Morshie Zongo GHS 60,000.00 GHS 173,150.00 Total GHS 665,400.00 GHS 50,000.00 GHS 10,000.00 GHS 9,688,470.00 Source; (NADMO, 2018) 2.0.2.3 Buildings and Structures Flooding has a great impact on the built environment considerably, as many buildings and structures are affected by the floods. Some buildings and structures within flood prone areas are constructed with no permits. They are typically constructed with materials and techniques that cannot stand in case of extreme rainfall. These structures are totally or partially destroyed when hit by floods (KMA, 2018). A total of 215 houses were affected by flood that occurred on September, 2016 (NADMO, 2018). 2.0.2.4 Sanitation Flood impact on water and sanitation in Kumasi has been a recurring situation for the past years. Runoff water from flood carries a lot of liquid and solid waste which are deposited in open locations as shown in the figure below around Kronom in Kumasi. 10 | Page Figure 2. 3 Solid waste carried by Flood water 2.0.3 Flood Mitigation in Kumasi Flood mitigation involves the management and control of flood water movement, such as redirecting flood run-off through the use of floodwalls and flood gates, rather than trying to prevent floods altogether. In the Kumasi Metropolis, it was observed that most residents in flood prone areas have constructed embankments and walls to prevent intrusion of flood waters into their buildings. Bridges and embankments are also constructed in most of the major river passage ways to direct the free flow of runoff. However, these mitigation methods are not sufficient since high volumes of river runoff still overflows its bank leading to flooding in these areas. The figure below shows a bridge under construction on the Daban river. Figure 2. 4 A bridge being constructed at Daban river to control flood water 11 | Page 2.1 GIS in Flood Management Geographic Information System (GIS) is a system designed to capture, store, manipulate, analyze, manage and present spatial or geographic data. GIS applications are tools that allow users to create interactive queries (user created searches), analyze spatial information, edit data in maps and present the results of these operations. GIS can refer to a number of different technologies, processes and methods. It is attached to too many operations and has many applications related to engineering, planning, management, etc. For that reason, GIS applications can be the foundation for many locations enabled services that rely on analysis and visualization (Amoako E. O., Sun L., 2018). Advancement in GIS and Remote Sensing technology help in real time monitoring, early warning and quick damage assessment of flood disasters. A Geographic Information System is a tool that can assist floodplain managers in identifying flood risk areas. In GIS, geographical information is stored in a database that can be queried and graphically displayed for analysis. By overlaying or intersecting different geographical layers, flood risk areas can be identified and targeted for mitigation or flood management practices. The main advantage of using GIS for flood management is its ability to create means to further analyze and estimate probable impacts of flooding incidence aside generating a visualization of flooding. 2.2 Flood Risk Mapping GIS have over the years been one of the most sort after means for the production of flood risk maps. Many researches have been performed using different methods in GIS. Flood risk maps guarantees the first step in flood management. Flood risk mapping and simulation forms the foundation of the decision-making process by providing information which is essential to the understanding of the nature, risk and characteristics of flooding to areas that could be affected by the flood (B. Essel, 2017). Flood risk map provides information on the danger to people and infrastructure when exposed to floods (B. Essel, 2017). In 2013, the Associated Programme on Flood Management established the fact that, there are three different approaches to develop a flood risk map: 12 | Page ● Modelling approach provides simulation of floods in which hydrological and hydraulic models are applied. Challenges identified in this approach are often in the availability and the spatial resolution of the data. Availability and spatial resolution of data affect the accuracy of the prediction. However, one major merit of this approach is that it is used for mapping accurate flood inundation. It is known to provide detailed and accurate flood assessment among the other approaches. ● Historic approach which provides information on past flooded areas which is obtained from national and regional database, technical reports, newspapers and photographs. The preparation of risk map involves gathering all the flood data of an area and overlaying of all layers. An advantage associated with this approach is that the flood records can be used as a visual proof and reminder that the danger of flood inundation is real. Also, a major drawback of this approach is that the map needs to be updated periodically as when new flood inundation events are recorded. In addition, prediction of flood inundation using historic data is problematic since the conditions of the floodplain, land use or the landscape could change and as such, a similar flood inundation in future might have different characteristics. ● Geomorphologic approach provides information on flood inundation based on surface features. Generally, past floods events leave traces on a landscape. As such, it is possible to read and interpret the traces on the landscape. An advantage of this method is that it can often be used for validation of a flood risk map produced from a modelling approach. This study will apply the modelling approach for its flood risk mapping. Recent studies have been done using the modelling approach in Geographical Information Systems. (Ozkan S.P., Tarhan C., 2016) discussed that the possible means of overcoming flood risk is by simulations. (Ozkan S.P., Tarhan C., 2016) in their study established that the required factors for creating a flood hazard and risk map are elevation, land use, rainfall, flow accumulation and slope, although in some instances, there are modifications to the factors used in producing the flood risk map. An example is (Ismail Elkhrachy, 2015), in this study, the causative factors used were surface roughness, geology, land use, rainfall run-off, drainage density and distance from the river channel. (Amoako E. O., Sun L., 2018) in their study considered factors relating to elevation and topography, land use, flow accumulation and drainage basins. In that study, the Normalized Difference Water Index (NDWI) was used to derive possible flood inundation and land cover pattern. 13 | Page (Forkuo, 2011) applied topographical, land cover and demographic factors for flood risk mapping. (Ena Gamez, Tereza Cavazos, Luis Mendoza, 2018) applied hydrological tools of the ArcGIS spatial analysis combined with the Geospatial Hydrologic Modeling Extention (HEC-GeoHMS) tool to perform a flood risk assessment, they considered factors based on elevation and hydraulic models, land use information, soil types and precipitation. Generally, it can be said that creating of flood risk maps is influenced by the availability of input data and available tools. The use of input data is not constant; however, the Digital Elevation Model is one the most important input data to produce a flood risk map and simulate flood. In our study, we will use the Quantum GIS software tools for flood risk mapping. The choice is because it is a free package software, it is effective and easy to use and has the ability to accept different data formats. The GIS tool has several plugins and extensions available that allows graphic representation, visualization and simulation. 2.3 Flood Simulation Simulation of flood is a vital part in the assessment of flood risk. It provides a clear visualization of flood behavior and potential effects they may pose. This can help flood management authorities determine the extent of flood as it occurs to apply the necessary public awareness measures, future land planning and resource allocation. Most earlier researches have applied the one, two- and three-dimensional hydraulic models such as HEC-RAS and MIKE-2 and hydrologic models such as ArcHydro extension tools in ArcGIS. (C. Fosu, 2012) in their research used Geographic Information System, spatial technology and the HEC-RAS hydraulic model as tools for river modelling of the Susan river in Kumasi. In their research, a Digital Elevation Model (DEM) which is basic input for any effective flood modeling was created from contour data. The geometric data needed for the modeling process were extracted from the DEM, topographic map and field measurements. A remotely sensed image was classified into various land cover types which was used for estimating the roughness coefficient of the various cover types during the modeling process. The model results were displayed and analyzed in ESRI ArcGIS environment. The flooded area was geometrically overlaid on the topographic map to delineate the affected buildings. (E. K. Forkuo, 2013) in his research used a DEM to delineate drainage patterns and watershed of the study area using ArcGIS desktop and its ArcHydro extension tools. Using geometric data needed for the modeling process, a vector-based approach was used 14 | Page to derive inundation areas at various flood levels. A Google Earth image of the study area was used as input for determining properties and areas which will be inundated in a flood event and subsequently generating flood inundation maps. The hazard map produced clearly shows the spatial allotment of the flooded area, which is located at areas with relatively low relief. (Twumasiwaah, K.A., 2016) in his thesis used the same modeling approach adopted in (E. K. Forkuo, 2013). A simple vector-based method that only required the extents of flood levels in simulating flood extent based on the derived drainage lines, their depth, and their capacity to hold rainfall run-off was used. With the aid of the elevation measure, flood water levels were selected. The flood contours showed the extent of flood at a given flood level. The flood model extents were further overlaid on a geo-referenced Google Earth image of the study area which visibly demonstrated areas at risk in the event of floods. All stated approaches used digital elevation models as an important input for effective modeling and applied Geographic Information System and hydraulic models as tools at arriving at their aim. However, there are more rigorous and much simple methodologies that can simulate flood over a given area. An example is the use of QGIS GRASS 7 Commands raster tools - r.lake module (QGIS Documentation, 2019) which only require a digital elevation model of your area, the water level of the flood and a raster layer containing cells of the drainage (more or less like the DEM). Post-processing using this method yields an output map showing areas with their corresponding flood depth from the specified water level. 15 | Page CHAPTER 3 METHODOLOGY 3.0 Introduction This chapter of the research will show a case study of Kumasi and describe the processes involved in producing a flood risk map by GIS hydrological modeling approach. This chapter will further illustrate the data, materials, and methods used and their sources for the modeling. 3.1 Study Area Kumasi is the second-largest city in Ghana in terms of population. It is the capital of the Ashanti Region in Ghana. Kumasi is located in the central part of Ghana with geographic coordinates between Latitude 6⁰38’ and 6⁰45’ North and Longitude 1⁰41’ and 1⁰32’ West. It has a projected geometric area of approximately 210.26 square kilometers, a perimeter of 105.38 kilometers, and elevation ranging from 250m to 300m above mean sea level. It shares boundaries with Afigya Kwabre Municipal to the North, Atwima Kwanwoma Municipal to the South, Atwima Nwabiagya Municipal to the West and Asokore Mampong Municipal to the East (TCPD, 2018). Flooding in Kumasi is a frequent occurrence. The population of Kumasi has increased by 18.87% between 2010 and 2018 due to the expansion of settlements leaving more people 16 | Page vulnerable to flood risk (GSS, 2018). Most people in Kumasi have being building in flood areas and thus exposed to flood. Figure 3. 1 A map of the study area showing its geographic location in Ghana 3.2 Data materials and Sources The following datasets were selected to aid in the analysis process for the purpose of this study. Table 3. 1 Data materials and sources Data Type Spatial data Digital Elevation Model Date captured Spatial Resolution Source Data Format Purpose of Data 2014 30m Shuttle Radar Topography Mission (SRTM) – USGS Earth Explorer Town and Country Planning Department Georeferenced Tagged Image File Format (GeoTIFF) Production of Elevation, Slope, TWI and stream channels maps. Shapefile (shp.) Digitized to obtain Kumasi and submetro boundaries. Boundary 2018 map of Kumasi Non-spatial data 17 | Page Population 2010 2018 & Rainfall 2005 2015 - Flood occurrence 2016 2018 - 18 | Page Ghana Statistical Service Microsoft docx. Determination of format Population Density within the risk zones of the study area. Ghana Microsoft docx. Determination of Meteorologica format highest recording l Agency mean monthly rainfall of study area. National Paper format Provide Disaster and information on Management recent flood Organization occurrences and estimated cost involved in flood management of study area. 3.4 Flow Chart of Methodology No Yes Figure 3. 2 Flowchart of Methodology 19 | Page 3.5 Software and Data Geoprocessing 3.5.1 Quantum GIS Analysis and creation of maps were performed on the spatial and non-spatial data acquired. The Quantum GIS software, an open source Geographic Information System (GIS) software licensed under the GNU General Public License was used for the analysis. QGIS provides a continuously growing number of capabilities provided by core functions and plugins. You can visualize, manage, edit, analyze data, and compose printable maps (QGIS Documentation, 2019). An advantage of QGIS is its ability to integrate data in different formats and its easiness in processing data compared to packages like ArcGIS. All datasets acquired were projected to the UTM coordinates system. 3.5.2 GRASS GIS GRASS (Geographic Resources Analysis Support System), is a free and open source Geographic Information System (GIS) software suite used for geospatial data management and analysis, image processing, graphics and maps production, spatial modeling, and visualization. GRASS GIS is currently used in academic and commercial settings around the world, as well as by many governmental agencies and environmental consulting companies. GRASS GIS has evolved into a powerful utility with a wide range of applications in many different areas of applications and scientific research. It is incorporated into Quantum GIS (source: https://grass.osgeo.org) 3.5.3 Microsoft Word and Excel 2016 Microsoft Excel and Word 2016 was used for drawing charts and for statistical computations. 3.5.1 Digital Elevation Model DEM is regular grid of elevation of earth surface in a raster form. DEM data has wide range of application in studies related to topography. Topographic data is essential for the visualization and modelling of the flood risk areas. The evaluation of the vertical accuracy of a DEM with regards to hydrological processes is very important. 20 | Page A study by G. Forkour and B. Maathuis, was done to compare the vertical accuracy of SRTM and ASTER derived Digital Elevation Models over two regions in Ghana by comparing them to a reference topographic data of higher accuracy. Root mean square error (RMSE); a common measure of quantifying vertical accuracy in DEMs, was calculated for each error map. Horizontal profiling was created on the DEM’s and compared. Correlation scatter plots was performed to assess the level of correlation between the DEM’s. It was realized that the SRTM has a higher vertical accuracy (in terms of RMSE) than ASTER GDEM for both sites. RMSE’s ranged between 4.9 and 5.5 (site 1) and 14.5 and 18.8 (site 2) for SRTM and ASTER respectively. The SRTM was therefore established to be good in terms of vertical accuracy for hydrological modelling since it depicts a better terrain than the ASTER GDEM. The vertical accuracy of a DEM is affected by the terrain morphological characteristics and terrain roughness. The accuracy of SRTM data decreases with increase in slope and elevation (S. Mukherjee P. K., 2013). However, small variations like 50m are barely significant. Shuttle Radar Topography Mission (SRTM) digital elevation data at a resolution of 1 arc-second (30 meters) of the study area was derived from earth explorer USGS / NASA website which was captured in 2014. The dataset was in Georeferenced Tagged Image File Format (GeoTIFF) which is a TIFF file of embedded geographic coordinates with horizontal datum WGS84 and vertical datum EGM96 (Earth Gravitational Model 1996). It was projected to the Universal Transverse Mercator (UTM) zone 30N projection system and clipped to the extent of the study area using the clip by polygon layer tool in QGIS 2.18 environment. 3.5.1.1 Filling of Sinks / Removal of Void Before a Digital Elevation Model can be used for hydrological analysis, all depressions or voids are to be filled. Sinks (also known as depressions) are that part of the DEM which impedes the flow of water to an outlet at the edge of the DEM. Fill sinks (wang & lui) tool under the terrain analysis – hydrology tools in the QGIS SAGA 2.1.2 geo-algorithm toolbox was used to fill the sinks. 21 | Page 3.5.1.2 Catchment Area Catchment area also known as flow accumulation shows the cells of the DEM raster where water catches or accumulates as it flows downwards. Cells with a high catchment values are areas of concentrated flow and may be used to identify the channel network. This was done using the catchment area tool under terrain analysis - hydrology in the QGIS SAGA geo-algorithms toolbox. 3.5.1.3 Flow Direction, Channel Network and Drainage Basins The flow direction shows the possible direction of water run-off on the elevation model. The drainage basins were delineated to determine basins and watershed boundaries. The Channel network and drainage basin tool in terrain analysis-hydrology was used to create the flow direction, stream network and drainage basins using the filled DEM as input and maintaining a threshold of 5.0000. 3.6 Modelling of hydrological factors 3.6.1 Digital elevation Elevation information is useful for many environmental applications including hydrological modelling. They provide the opportunity to model, analyze and display phenomenon related to topography and other surfaces (McDougall et al, 2008). Elevation is useful for dictating where the water will flow and accumulate allowing easy identification on areas with risk of flooding. The filled DEM of the study area was classified into five classes into equal interval discrete classification. The minimum point’s elevation was 206m and the maximum with an elevation of 315m. High elevation points have minimum risk of flood while low elevation points have maximum risk of flood. Figure 4.4 shows output data obtained from the classified DEM. 3.6.2 Slope Slope is an important element in flooding and as such the danger in floods increases as the slope decreases. The raster terrain analysis tool was used to create the slope map with the filled DEM as input using equal intervals. The output (slope) was classified into five classes with high values depicting high slope gradient and low values showing low slope 22 | Page gradient. Figure 4.6 shows the results of the output slope map into equal interval discrete classification. 3.6.3 Topographic Wetness Index (TWI) Topographic wetness index estimates the soil wetness and ability to retain runoff water based on the topography. The index combines slope and catchment area, two parameters that influence the soil wetness. If the catchment area value is high, this means that more water will flow into the cell, thus, increasing its soil wetness. A low value of slope will have a similar effect because the water that flows into the cell will not flow out of it quickly. The r.watershed – watershed basin analysis program raster tool under the GRASS GIS 7 command geoalgorithm was used to produce the topographic wetness index using the filled DEM as elevation input. The output raster data was classified into five classes with high values depicting areas with higher tendency to hold or accumulate runoff water. Low values show areas that has low tendency to accumulate water or are depicting dry areas within the study area. Figure 4.8 shows the results of the output TWI map using equal interval discrete classification. Mathematical formula for computing Topographic wetness index: α 𝐼𝑛( tan𝑡𝑎𝑛 (β) ) where α is the cumulative upslope area draining through a point per unit contour length (catchment area). β is the local slope angle converted to radians. The TWI reflects the tendency of water to accumulate at any point in the catchment and the tendency for gravitational forces to move that water downslope (Quinn et al. 1991). This value will be negative if α / tan(β) < 1. 3.6.4 Proximity (raster distance) of rivers The proximity (raster distance) analysis tool under the GDLA/ODR geoalgorithms generates a raster proximity map indicating the distance from the center of the nearest pixel identified as a target pixel. The river vector data was rasterized and used as input layer for the raster distance. High values indicated areas within the low risk or within the 23 | Page target pixel zone. Low values showed areas with high risk of inundation. The lower the value and the smaller the distance to the river pixels, the higher the risk. Figure 4.10 shows the results obtained from the raster distance analysis. 3.6.5 Overlaying of maps using the raster calculator 3.6.5.1 Reclassification of maps Reclassification is the process of assigning new values to a range or group of values of a raster or vector data. Before weighted overlay is performed, it is important that all criteria maps possess these same properties: ● ● ● ● ● ● Scale Resolution Spatial extent Map unit Coordinate and projection system File format The elevation, slope, TWI, and proximity maps were reclassified to continuous values (0 to 1) to the pixels using the raster calculator. All the maps were ranged such that higher values (pixel values nearer to 1) will depict greater influence on flood risk. Results show high-risk areas with high values and low-risk areas with low values. The Formula below indicates the general classification scoring procedure: Where 𝑋𝑖 = pixel value 𝑋𝑚𝑖𝑛 = minimum pixel value 𝑋𝑚𝑎𝑥 = maximum pixel value 24 | Page 𝑋𝑖−𝑋𝑚𝑖𝑛 𝑅 = ( 𝑋𝑚𝑎𝑥−𝑋𝑚𝑖𝑛 ) Table 3. 2 Formula table in raster calculator Raster maps Elevation Slope TWI Proximity Formula in raster calculator Explanation 1–R Low values of elevation have high risk of flooding. Inverse of formula is used and high values are given to high risk zones. 1–R Low values of slope have high risk of flooding. Inverse of formula is used and high values are given to high risk areas. R High values of TWI have risk of flooding. Actual formula is used and high values represents high risk areas. 1–R Low proximity has high risk of flooding. Inverse of formula is used and high values are given to high risk areas. 3.6.5.2 Weighting of reclassified maps Equals influence on flood risk were adopted for all the maps, and therefore equal weights were assigned to all four map layers. The weights were multiplied by the corresponding maps and the weighted sum overlay was applied to produce the flood risk map using the raster calculator. The formula below shows the raster calculation used for the weighted sum overlay: 𝑇𝑊𝐼 * 0. 25 + 𝐸𝑙𝑒𝑣𝑎𝑡𝑖𝑜𝑛 * 0. 25 + 𝑆𝑙𝑜𝑝𝑒 * 0. 25 + 𝑃𝑟𝑜𝑥𝑖𝑚𝑖𝑡𝑦 * 0. 25 = 𝑓𝑙𝑜𝑜𝑑 𝑟𝑖𝑠𝑘 𝑚𝑎𝑝 25 | Page Table 3. 3 Weighting of reclassified maps Factor map Weight influence (%) TWI 25 Elevation 25 Slope 25 Proximity 25 The output flood risk map was reclassified into five discrete equal interval classes. 3.7 GPS Field Survey A GPS survey was conducted in order to validate and test the accuracy of the output data. A handheld Garmin GPS with a positional accuracy of 1 m – 2 m was used. GPS Coordinates of 5 historically known flood areas within Kumasi were picked and overlaid on the flood risk map to assess the accuracy of the output flood risk map. 3.8 Population Density map Population density map from the most recent 2018 population census of the 9 sub-metros was prepared to determine the population per squared kilometer in exposure to risk of flood. Higher population density areas with very high flood risk are exposed to the greatest damage since more people will be affected, and emergency response for evacuation will be slower. 3.9 Rainfall Pattern and Trend 3.9.1 Regression Analysis Trend in rainfall at a particular station can be examined by applying the regression analysis with time as the independent variable and annual rainfall as the dependent variable. A linear equation, y = mt + c, defined by c (the intercept) and trend m (the slope), which represents the rate of increase or decrease of the variable, can be fitted by regression and t is time in years. The regression analysis for the annual rainfall pattern was performed to determine the rate of increase or decrease of annual rainfall in Kumasi. Statistical analysis on rainfall data was used to determine the monthly rainfall pattern in the study area. 26 | Page 3.10 Field Survey Verbal surveys were done to establish the socio-environmental causes and effects of flood on the people in exposure. Historic flood data from organizations was also analyzed to determine areas which experience flooding most often, the damages associated and the flood mitigation measures adopted. 3.11 Flood Simulation Flooding can be easily simulated by r.lake module in GRASS GIS 7 commands raster geoalgorithm in QGIS environment which fills a lake to a target water level from a given start point or seed raster. The resulting raster map contains cells with values representing lake depth (NULL for all other cells beyond the lake). The seed (starting) point can be a raster map with at least one cell value greater than zero, or a seed point can be specified as an E, N coordinate pair (Maris Nartiss, 2015). Water depth is reported relative to specified water level (specified level = 0 depth). This module uses a 3x3 moving window approach to find all cells that match three criteria and to define the flood depth: ● ● ● cells are below the specified elevation (i.e., water level); cells are connected with an initial cell (seed or coordinates pair value); cells are not NULL or masked. The accuracy of the output is also dependent on the accuracy of the digital elevation model. High accurate digital elevation models provide accurate flood depth estimates. A flowchart below defines the methodology of the flood simulation. Figure 3. 3 Flowchart of flood simulation methodology 27 | Page CHAPTER 4 RESULTS In this study, the mapping and assessment of flood risk were based on four hydrological factors discussed in the methodology. The results of hydrological factors and flood risk maps have been graphically represented and discussed in this chapter. 4.0 Flood Risk Mapping 4.0.1 Filling of Sink (Depressions) Sinks are areas in the DEM which do not allow the flow of water to an outlet at the edge of the DEM. All sinks were filled as shown in Figure 4.1 to make it possible for hydrological modelling. The dark portions marked areas of low elevation and the white portions with high elevation. Figure 4. 1 Map of Depressionless DEM 28 | Page 4.0.2 Catchment Area The catchment area shows how much surface water catches in each cell. Cells with high pixel values represents stream or river channels. Figure 4.2 shows the cells with the highest amount of water catchment flowing on the surface of the filled sinks DEM. Figure 4. 2 Map of Catchment area 4.0.3 Stream Network The stream network represents the cells with the highest catchment area values. The stream network indicated the path of formation of large body of flowing surface water in the occurrence of rainfall in the study area and they represent the rivers within the study area. Figure 4.3 shows a vectorized river from raster cells of the study area. 29 | Page Figure 4. 3 Map of Stream Networks in Kumasi 4.0.4 Digital Elevation Map The digital elevation model created from an SRTM DEM dataset shows areas classified into five classes as shown in Figure 4.4. It can be observed that, the lower elevation areas are all around the river channels. These areas inundate whenever the river overflows its banks leading to very high risk of flooding. The result of the total area of each class and its corresponding percentage determined is shown in Table 4.1. Table 4. 1 Area of Digital Elevation map classes Elevation Very Low Low Moderate High Very High Total 30 | Page Area in sq.km 30.6612 54.9567 53.0802 43.8102 18.1998 200.7081 Percentage Area (%) 15.28 27.38 26.45 21.83 9.07 100 Figure 4. 4 Digital Elevation Model of Kumasi The digital elevation map was reclassified and inversed to continuous values from 0 to 1 with lower elevation areas having the highest risk of flooding as shown in Figure 4.5. The nearer the pixel value to 1, the higher the risk and vice versa. 31 | Page Figure 4. 5 Classified and inversed Digital Elevation Model of Kumasi 4.0.5 Slope Map Slope calculated the maximum rate of change from a cell to its neighbors. The slope map was classified to five classes as shown in Figure 4.6. The results show areas with very high to very low gradient in the terrain. The higher the value, the higher the slope angle. Areas with higher slope have lower risk of flooding. This indicates that these areas will have increase in run-off since slope increases intensity of flow of runoff water. The result of the total area of each class and its corresponding percentage determined is shown in Table 4.2. Table 4. 2 Area of Slope map classes Slope Very Low Low Moderate High Very High Total 32 | Page Area in sq.km 46.3716 62.3169 54.3231 31.1121 7.5465 201.6702 Percentage Area (%) 22.99 30.90 26.94 15.43 3.74 100 Figure 4. 6 Slope map of Kumasi The slope was reclassified and inversed to continuous values from 0 to 1 with lower slope having the highest risk of flooding. The nearer the pixel value to 1, the higher the risk and vice versa. Figure 4.7 shows the reclassified and inversed slope map. 33 | Page Figure 4. 7 Classified and inversed Slope map of Kumasi 4.0.6 Topographic Wetness Index Map TWI reflects the tendency of water to accumulate at any point in the catchment and the tendency for gravitational forces to move that water downslope. Results from the TWI map shows different classes of the TWI map with the high-risk areas having higher values in Figure 4.8. The result of the total area of each class and its corresponding percentage determined is shown in Table 4.3. Table 4. 3 Area of Topographic wetness index classes TWI Very Low Low Moderate High Very High Total 34 | Page Area in sq.km 9.9189 151.1928 18.4059 18.7380 2.5839 200.8395 Percentage Area (%) 4.94 75.28 9.16 9.33 1.29 100 Figure 4. 8 Topographic Wetness Index map of Kumasi The reclassified TWI map with values nearer to 1 show areas with the highest risk of flooding in Figure 4.9 since these areas accumulates more rainfall or runoff water in the study area. Moreover, these areas retain more water when there is high rainfall which accumulates for a period leading to water flooding at a depth in these parts of the study area. 35 | Page Figure 4. 9 Classified Topographic Wetness Index map of Kumasi 4.0.7 Proximity/Raster distance of river The raster distance function measured the straight-line distance from each cell to the closest source. Areas within the study area with higher risk of flooding are much closer to the river channels. High risk areas have lower values. Figure 4.10 shows the reclassified and inversed proximity map with higher values nearer to 1 showing areas with high risk of flooding. 36 | Page Figure 4. 10 Proximity/Raster distance map of Kumasi 4.0.8 Results on Weighted Overlay - Flood Risk Map The four reclassified elevation, slope, TWI and proximity layers where overlaid by multiplying each by its weights and summing them together by the weighted sum method to produce the final flood risk map. Equal weights were applied because equal influence to flooding was adopted for each of the four layers. The final output layer was classified into five equal interval classes showing five flood risk zones in the study area in Figure 4.12. Figure 4.14 shows the points picked during the field works showing historically known high risk areas within the study area overlaid on the flood risk map to verify the accuracy of the output dataset. Figure 4.14 also shows some towns in the study area indicated in the five flood risk areas. Tables 4.4 shows the total area and percentage for the five flood risk areas in the study area. Table 4. 4 Area of Flood risk map classes Flood Risk Area Very Low Low Moderate High Very High Total 37 | Page Area in sq. km 20.4120 55.6533 63.4419 39.2022 19.0638 197.7732 Percentage Area (%) 10.3 28.1 32.1 19.8 9.6 100 Figure 4. 11 Quantitative chart of Flood risk zones in Kumasi Figure 4. 12 Flood risk map of Kumasi 38 | Page Figure 4. 13 Vectorised Flood risk map of Kumasi From the flood risk map, it can be identified that the southern and north-western part of Kumasi is seen to be at very high risk to flooding. They have very high potential of been inundated and covers 9.6% of the study area. 19.8% of the study area is identified to be at high risk to flooding. Areas within the moderate risk of flooding covers the most in the study area with 32.1%. Also, from the flood risk map, there are some areas which has lower probability to flood. The low risk areas cover a total of 28.1% and the very low risk areas cover 10.3% of the study area. These areas are mosthly located in the North eastern part of Kumasi. Overall, areas identified to be highly susceptible to flooding covers 61.5% including the moderate zones of the entire study area. This suggests that more areas can be considered to be at danger of flooding in the study area. 39 | Page Figure 4. 14 Flood risk map with towns and flood points in Kumasi Figure 4. 15 Chart showing percentage area of flood risk zones for Kumasi submetros 40 | Page From Figure 4.15, Nhyiaeso, Asokwa, Bantama, Kwadaso and Oforikrom are the sub-metros at greater risk of flood since the high susceptibility areas (very high, high and moderate) covers a larger percentage compared to Suame, Tafo and Manhyia, where the low susceptibility areas (low and very low) covers a larger percentage of the total area. Subin has an equal percentage of area lying within the high susceptibility and low susceptibility areas. However, due to its large population density, flood occurrence in this area will endanger a greater number of people and properties especially in the high susceptibility areas. During the field work and flood data analysis, it was noticed that most of the worst affected areas where concentrated at the southern and north-western portion of the study area. All these areas lie within the channels of the major water bodies. This is in context with the modelled flood risk map. Table 4. 5 Area of Flood risk and Population Density classes Flo od Ris k Label Very Low Low Moderate High Very High Total 41 | Page Very Low 3.98 17.58 23.5 15.52 7.57 68.2 Population Density (% Area) Low Moderate High 0.9 2.06 1.77 3.05 2.1 2.05 4.34 1.25 0.84 2.69 0.5 0.35 1.34 0.2 0.12 12.3 6.1 5.1 Very High 1.53 3.33 2.19 0.81 0.45 8.3 Total 10.2 28.1 32.1 19.9 9.7 100 Figure 4. 16 Chart showing the percentage population density area for each Flood risk zone 4.1 Population Density Map The population density map provides insight on the population distribution per squared km for each sub-metro. The results showed that the two sub metros: Subin and Tafo have the highest population density, averaging, 19173 to 22535 persons per squared km. This indicates a greater population to be affected when there is the occurrence of flood mostly in the very high-risk zones. Manhyia was identified to have high population density averaging 15812 to 19173 persons per square km. Suame had a moderate population density of 12450 to 15812 persons per squared km. Also, from the Figure 4.17, it can be seen that there are some sub-metros namely: Nhyiaeso, Kwadaso, Asokwa and Oforikrom which have the lowest population density of 5727 to 9089 persons per squared km. In an incidence of flooding, these areas are likely not to affect a greater number of people. In general, Kumasi is highly dense with about 52% (1.07m) of people living in a highly dense area. These areas are depicted in the central and Northern part of Kumasi. 42 | Page Figure 4. 17 Population density map of Kumasi 4.2 Rainfall Pattern and Trend Results from the statistical analysis of the monthly rainfall trend in Kumasi shows that June has the highest monthly average rainfall of 242.99mm, July with 199.40mm, May with 177.31mm, September with 175.26mm and October with 172.05mm. This indicates that there is the high probability or likelihood of flood occurrence during these months, especially June. Areas within the very high and high flood risk zones will be exposed to more damage since much runoff water will accumulate more at a depth in these areas. November, December, January, and August have the lowest average monthly rainfall. This suggests that flooding is not likely to occur during low or dry weather in these months. 43 | Page Figure 4. 18 Mean monthly rainfall graph of Kumasi Figure 4.19 shows the result of the annual rainfall trend for Kumasi from the least squares regression linear analysis method. The graph shows an upwards trend with a slope of 2.4938. This indicates that rainfall in Kumasi is likely to partially increase with time and therefore more damage may be the resultant in the high flood risk zones in the study area. Figure 4. 19 Annual rainfall pattern of Kumasi Table 4. 6 Rainfall data of Kumasi from 2005 to 2015 44 | Page 4.3 Field Survey Verbal interviews conducted between some residents in flood affected areas and correspondents of flood management organizations gives some of the major socio-environmental causes and effects of flood in the study area. Some pictures where acquired during our field work showing some of the flood mitigation measures adopted by the communities in these areas. Figure 4. 20 shows an embankment built around Morhie-Zongo area to direct run-off water. However, most of these embankments becomes useless when high rainfall causes a collection of high volume of run-off water that overflows onto their banks causing flooding in these areas. Figure 4. 21 shows an image of buildings which are situated in floodplains. The river had dried out during the date of capture; however, it can be seen from afar that buildings area lying in very close proximity to the river channel. Residents of these places have built embankments and walls to reduce the intrusion of high rising waters from the nearby river which arises during high and continuous rainfall in such areas. From field observations, it was observed that traces of high rising waters have being left on these walls. Interviews performed on some residents tells that these flood mitigation measures adopted by the residents are not enough to curb the effects that floods possess in these areas. 45 | Page Figure 4. 20 Embankments built to direct runoff water Figure 4. 21 Buildings within river banks 4.4 Flood Simulation The flood depth derived from the r.lake.layer raster module in QGIS shows areas that are inundated in the event of flood at the specified flood depths shown in Figure 4.24. The flood depth obtained from the resampled digital elevation model was overlaid on the georeferenced Google Earth satellite image of the area to show areas at risk in the event of floods. A contour interval of 1m was also extracted from the elevation model and overlaid on the satellite imagery to determine contours falling within the flood levels. Figure 4.23 and 4.22 shows the resampled map and non-resampled map respectively. Figure 4.24 shows the classified digital elevation model of Susan river, Atonsu Junction and its geographic location on the map of Kumasi. Field survey information suggest that areas that lies within the various flood depths are known for their flood vulnerability during heavy rainfalls. Most commercial and residential such as the gas station and the main road can be found in Figure 2.4. Areas lying outside the depth zone are due to their high elevations. The map was exported to the web browser using the Qgis2threejs plugin in QGIS environment to help in 3d visualization of the area as shown in Figure 4. 24. Qgis2threejs plugin exports terrain data, map canvas image and vector data to your web browser. You can view exported 3D objects on web browser which supports WebGL. This plugin uses three.js library (QGIS Documentation, 2019). 46 | Page Figure 4. 22 Extracted DEM of Susan River, Atonsu-Junction 47 | Page Figure 4. 23 Resampled DEM of Susan river, Atonsu-Junction Figure 4. 24 DEM of Susan river, Atonsu-Junction 48 | Page Figure 4. 25 Flood simulation map of Susan river, Atonsu-Junction Figure 4. 26 3d visualization of flood depth simulation on Susan river, Atonsu-Junction with a vertical exaggeration of 4.5 using Qgis2threejs plugin in QGIS 2.18 software. 49 | Page DISCUSSIONS 4.5 Discussion on Results of Flood Risk Map The results of the flood risk mapping shows that there are areas in Kumasi that are at higher vulnerability to flooding. The results of this research have established the fact that these areas require immediate flood mitigation measures to be mounted and awareness strategies made to residents. Kumasi has a total land size of 210.26 square kilometers of which a total of 58.266 sq. km falls within the highly vulnerable area – very high risk and high risk. The moderate zone covered an area of 63.4419 sq. km and the least vulnerable areas – low and very low covered a total of 76.0653 sq. km from the flood risk map in Figure 4.12. Very high and high-risk zones are most likely to get flooded with more resulting damage while Very low and low risk zones are least likely to get flooded and with less resulting damage. The spatial variability of the classified risk zones in Kumasi varies across the study area. However, flood incidences are often likely to affect the North-Eastern and Southern parts as the results in Figure 4.13 tends to show that they have a higher risk to flooding. Research works by (Amoako E. O., Sun L., 2018), (Ena Gamez, Tereza Cavazos, Luis Mendoza, 2018), (Ozkan S.P., Tarhan C., 2016) and (Ismail Elkhrachy, 2015) has established the fact that, a composition of different hydrological factors is useful for the mapping of flood risk areas. In our research, four hydrological factors derived from the Digital Elevation Model was used in the production of the flood risk map. Figure 4.4 shows the elevation map where it can be observed that, the lower elevation areas are all around the river channels. These areas inundate whenever the river overflows its banks leading to very high risk of flooding. The results from the slope map in Figure 4.6 shows that areas at lower slope steepness or gradients will be at higher risk to flood than areas at higher slope steepness since run-off water in these areas will drain faster compared to the otherwise. Figure 4.8 show the results for the Topographic Wetness Index reflecting the tendency of water to accumulate at any point in the catchment and the tendency for gravitational forces to move that water downslope. Areas within the high TWI zones possesses the greatest flood risk since these areas are wetter than low TWI zones. The results obtained in Figure 4.10 for the river proximity shows that, areas within the study area with higher risk of flooding are much closer to the river channels. It was further deduced from Figure 4.14 that areas in the very high and high-risk zones such as Atonsu-Junction, Anloga Junction, Morshie-Zongo and Daban are all characterized by low elevation, low slope, high proximity to river channels and high 50 | Page topographic wetness index. Areas such as Tafo, Nsema, Ehwemase and Amanfrom which are characterized by high elevation, high slope, low proximity to river channels and low topographic wetness index lies within the very low and low risk zones. The flood risk map in Figure 4.12 then shows that areas with low slope, at low elevation, in close proximity to rivers and has a higher tendency to retain run-off water possesses the highest risk of flooding. These conditions are however subject to change over time due to the nature of settlements and human activities in the study area. These factors are likely to increase or decrease the vulnerability of the areas to flooding. 4.6 Discussion on the Influence of Flood Risk on Population Density Since flood risk is associated with the degree of its occurrence and the impacts on its victims, a population density data of the study area overlapped on the flood risk map shows each percentage population density associated with each flood risk zone in the study area in Table 4.5. Results showed a higher percentage for lower population density areas in the very high and high-risk zones than the high population density areas in the very high and high-risk zones. This is due to the fact that 4 of the 9 sub-metros in Kumasi lied in the very low population density zones in Figure 4.17. The 1.26% of the total residents within the very high and high-risk zones will face the highest effect in the occurrence of flooding. Thus, awareness and immediate measures will be necessary in this situation. 4.7 Discussion on Rainfall pattern in Kumasi The results from the Table 4.6 reveals a fluctuating average annual rainfall. Figure 4.19 shows a slight increase in slope from the regression graph indicating an increasing annual rainfall pattern of the study area. Also, the results from Figure 4.18 shows that the highest mean amount of rainfall recorded where all in the rainy season (May to October, except August). This could back the frequent occurrence of flooding in the rainy seasons and therefore measures should be taken in these months. 4.8 Discussion on Flood Simulation QGIS incorporated extensions and GRASS 7 Command raster module-r.lake was used in the determinations of flood potential and flood depths simulation using a resampled DEM as the main input which was extracted by raster clipping a drepressionless SRTM 30m 51 | Page DEM of the Susan river, Atonsu-Junction in Kumasi. The results in Figure 4.25 shows areas that falls within a given flood depth above the river surface. This module depends primarily on digital elevation raster data. The challenges we encountered in using this module is the accuracy of our DEM which was resampled to 2m but however did not change the accuracy of our DEM. More cells where added by using a bilinear interpolation method to interpolate new elevations for these cells. This method however is very simple and accurate provided you are using a much accurate and high-resolution DEM which is likely to yield more accurate results. However, field survey indicated that the areas that were inundated within the predicted flood depths are well known to be at risk in the event of flooding. It was identified that the affected elements included buildings; mainly residential and commercial structures like roads. 52 | Page CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS 5.0 Conclusions Flooding is a continuous problem in Kumasi. This research has helped explain a GIS approach with the aid of QGIS software and GRASS tools for the generation of a flood risk map for Kumasi using hydrological factors obtained from digital elevation model. This has enabled us understand the relationship that exist between these factors in identifying areas of very high, high, moderate, low and very low flood risk zones in Kumasi which indicates that our objectives were achieved. The findings of this research show that high probability of very high and high – risk areas were concentrated in the southern and north-eastern part of the study area, reason being low elevations in those areas. This was a proof that Digital Elevation Model has a great impact on flood risk mapping and as such, elevation can be considered an essential dataset for flood risk mapping. The reclassified elevation, slope, river proximity and topographic wetness index maps where combined in the QGIS environment using numerous tools and modules to produce the flood risk map for Kumasi Metropolis. A flood simulation was further done using GRASS 7 Command geoalgorithm raster module-r.lake in QGIS to demonstrate and visualize the risk of flooding along the Susan river, Atonsu-Junction which lies in one of the very-high flood risk zones. In this contest, we draw further conclusions that: a) The production of flood risk map has helped in the identification of very-high, high, moderate, low and very low risk areas in Kumasi where a total of 30% are at high vulnerability (very high and high-risk) to flooding. b) Flooding in Kumasi is likely to occur during the rainy seasons due to high rainfall volume causing increase in run-off water. There is high probability of flooding to occur in May, June, July, September and October since these months collects large volumes of rains which causes inundation. c) Flood risk influence on population is the most vital part of flood risk assessment and should be taken into consideration since population constitutes the major victims associated with such disasters. 53 | Page d) Flood simulation using QGIS and GRASS extension tools is one of the accurate methods of demonstrating and visualizing risk of flood in various risk zones. This helps in easy allocation of resources by flood managers and aid residents on awareness and insurance of properties. 5.1 Recommendations Results obtained in our research proofs that, flooding in Kumasi is a problem and new measures should be adopted not compromising old ones in managing it. Information acquisition on flood occurrence, estimation and its assessment should be made available to individuals and co-operate bodies concerned with flood. Future flood mapping should make use of Digital Terrain Models in flood simulations since the presence of buildings and other structures on the ground have direct influence on the risk associated with flooding (flood depth and velocity). Modern and more accurate datasets and methods for flood mapping and assessment should be adopted such as the use of dense cloud data, very high-resolution remote sensing imagery, terrestrial surveying and photogrammetric means. Flood management authorities and land use planners should invest in these fields and make data readily available for researchers. Furthermore, flood management authorities should create awareness and educate residents on how to deal with flood to reduce flood effects on victims in case of its occurrence. They should be enlightened on the need to insure properties to reduce the financial effects floods poses on the flood management authorities. 54 | Page (n.d.). References Retrieved from FEMA-EMI.: https://www.training.fema.gov/hiedu/docs/fmc/chater 2-types of flood (2018, October 17). Retrieved from http://en.m.wikipedia.org/wiki/Flood Amoako C., Boamah E.F. (2014). The three-dimensional causes of flooding in Accra, Ghana. Incremental Learning and Flood Responses in Informal Communities in Accra, Ghana. Amoako E. O., Sun L. (2018). Application of Remote Sensing Technique and Geographic Information Science for Flood Risk Mapping-A Case Study of the Offinso District, Kumasi-Ghana. J Remote Sensing & GIS 7: 224, Vol 7(1). doi:10.4172/2469-4134.1000224 Associate Programme on Flood Management. (2013). Retrieved from APFM Web site. C. Fosu, E. K. (2012). River Inundation and Hazard Mapping – a Case Study of Susan River – Kumasi. Proceedings of Global Geospatial Conference 2012. C.Y. Okyere. Y. Yacouba, D. Gilgenbach. (2012). The problem of annual occurrences of floods in Accra: An integration of hydrological, economic and political perspectives. Theoretical and Empirical Researchers in Urban Management. Retrieved October 2018, from http://www.researchgate.net/publication/288201856 CRED. (2011). The OFDA/CRED International Disaster Database. Centre for Research on Epidemiology of Disasters–CRED. E. K. Forkuo, V. A. (2013, March). The Use of Digital Elevation Models for Watershed and FLood Hazard Mapping. International Journal of Remote Sensing & Geoscience (IJRSG) Volume 2, Issue 2. Retrieved May 2019, from www.ijrsg.com Ena Gamez, Tereza Cavazos, Luis Mendoza. (2018). Application of ArcGIS tools for Flood Risk Assessment. Centro de Investigacion Cientifica y de Educacion Supeior de Ensenada, B. C. (CICESE), Mexico. Forkuo, E. K. (2011). Flood Hazard Mapping using Aster Image with GIS. International Journal of Geomatics and Geosciences Volume 1, No. 4. Retrieved October 10, 2018 G. Forkuor, B. M. (n.d.). Comparison of SRTM and ASTER Derived Digital Elevation Models over Two Regions in Ghana – Implications for Hydrological and 55 | Page Environmental Modeling . Studies on Environmental and Applied Geomorphology . (2018). Ghana Meteorological Agency. Rainfall data between 2005 to 2015. GSS. (2018). Population data of Kumasi. International Conference on Computer Simulation. (2010). Risk Analysis and Hazard Mitigation. Algarve: Southampton: WIT, (2010). Retrieved November 2018 Ismail Elkhrachy. (2015). Flash Flood Hazard Mapping Using Satellite Images and GIS Tools: A case study of Njran City, Kingdom of Saudi Arabia (KSA). The Egyptian Journal of Remote Sensing and Space Science Volume 18, Issue 2. Retrieved from http://doi.ord/10.1016/j.ejrs.2015.06.007 J. O. Olusina, C. J. (2018). Visualisation of Uncertainty in 30m Resolution Global Digital Elevation Models: SRTM v3.0 and ASTER v2. doi:http://dx.doi.org/10.4314/njtd.v15i3.2 Jonkman S.N. (2005). Global perspectives on loss of human life caused by floods. Natural Hazards. Retrieved November 10, 2018 Julian Eleutério. (2013). Flood risk analysis: impact of uncertainty in hazard modelling and vulnerability assessments on damage estimations. Economics and finances. Université de Strasbourg, 2012. <NNT: 2012STRAB014>. <tel-00821011>. Retrieved November 10, 2018, from https://tel.archives-ouverted.fr/tel-00821011 KMA. (2018). Flood data of Kumasi. Konadu D.D., Fosu C. (2009). Digital Elevation Models and GIS for Watershed Modelling and Flood Prediction - A Case study of Accra Ghana. Retrieved November 3, 2018 Korah P.I., Cobbinah P.B. (2016). Juggling through Ghanaian urbanization: flood hazard mapping of kumasi. GeoJounak 82 (6). Retrieved November 2018 Kwang C., Osei E.M. (2017). Accra Flood Modelling through Application of Geographic Information Systems (GIS), Remote Sensing Techniques and Analytical Hierarchy Process. J Remote Sensing & GIS. Retrieved October 12, 2018 Messner F. and Meyer V. (2006). Flood damage, vulnerability and risk perception challenges for flood damage research. (J. Z. Schanze, Ed.) in: Flood risk management: Hazards, vulnerability and mitigation measures. 56 | Page Mirza. M, D. A.-G. (2011). Accuracy and relevance of Digital Elevation Models for Geomatics applications A case study of MakkahMunicipality, Saudi Arabia . International Journal of Geomatics and Geosciences Volume 1, No 4, 2011 . NADMO. (2018). Flood data of Kumasi. Nyarko, B. K. (2000). Flood Risk Zoning of Ghana - Accra experience. International Archives of Photogrammetry and Remote Sensing, XXXIII,( B7). Nyarko, E. K. (2002). Application of a Rational Model in GIS for Flood Risk Assessment in Accra, Ghana. Journal of Spatial Hydrology, 2(1). Ozkan S.P., Tarhan C. (2016). Detection of Flood Hazard in Urban Area Using GIS: Izmir Case. Retrieved November 2018, from http://www.creativecommons.org/licenses/by-nc-nd/4.0/ QGIS Documentation. (2019). Retrieved from http://www.qgis.com S. Mukherjee, P. K. (2013). Evaluation of Vertical accuracy of open source Digital Elevation Model (DEM). International Journal of Applied Earth Observation and Geoinformation 21 (2013). doi:205-217 S. N. Jonkman and J. K. Vrijling. (2008). Loss of life due to floods. doi:10.1111/j.1753-318X.2008.00006.x TCPD. (2018). Boundary and submetro map of Kumasi. Twumasiwaah, K.A. (2016). Urban Flood Risk Management: A case Study of Aboabo, Kumasi. Kumasi: Unpublished thesis presented to Dept. of Geomatic Engineering, KNUST. 57 | Page APPENDIX Appendix A: Coordinates of Very High-Risk Zones from GPS Survey (Garmin 62c handheld GPS) Location Eastern (X) Northern (Y) Morshie-Zongo Mosque Area 654250 742824 Anloga Junction 654995 739593 Atonsu-Junction 655299 736187 Daban river 652732 735041 Kronom river 649718 747245 58 | Page