University of Ghana http://ugspace.ug.edu.gh MONITORING SHORELINE CHANGE USING MEDIUM RESOLUTION MULTI-TEMPORAL SATELLITE IMAGERY: A CASE STUDY OF KETA, GHANA PHILIP-NERI JAYSON-QUASHIGAH (10328165) This Thesis is Submitted to the University of Ghana, Legon in Partial Fulfillment of the Requirement for the Award of MPhil Environmental Science Degree. July, 2011 University of Ghana http://ugspace.ug.edu.gh 6f 3 5 o 3 b S PI University of Ghana http://ugspace.ug.edu.gh D ECLA RATIO N This is to certify that this thesis is the result o f research undertaken by Philip-Neri Jayson-Quashigah under the supervision towards the Award o f the M .Phil Degree in the Environmental Science Programme, University o f Ghana. I j 02 /mO. Philip-Neri Jayson-Quashigah (Student) (Principal Supervisor) (Co-Supervisor) University of Ghana http://ugspace.ug.edu.gh ABSTRACT Shoreline change analysis provides important information upon which most coastal zone management and intervention policies rely. Such information is however mostly scarce for very large and inaccessible shorelines mostly due to expensive field work. This study investigated the reliability of medium resolution satellite imagery for mapping shoreline positions and for estimating historic rate of change. Both manual and semi-automatic shoreline extraction methods for multi-spectral satellite imageries were explored. Five shoreline positions were extracted for 1986, 1991, 2001, 2007 and 2011 covering a medium term of 25 years period. Two additional shoreline positions for 2010 and 2011 were extracted and used for accuracy assessment. Rates of change statistics were calculated using the End Point Rate and Weighted Linear Regression methods. Approximately 283 transects were cast at simple right angles along the enlire coast at 200m interval. Uncertainties were quantified for the shorelines ranging from ±4.1nt to± 5i accuracy of mapping the shoreline at 15m resolution estimated to be ±1 lm. The results show that the Keta shoreline is a very dynamic feature with average rate of erosion estimated to be about - 2m/year ±0.44m. Individual rates along some transect reach as high as -16m/year near the estuary and on the east of the Keta Sea Defence site. The study confirms earlier rates of erosion calculated for the area and also reveals the influence of the Keta sea defence on erosion along the eastern coast of Ghana. The research shows that shoreline change can be estimated using medium resolution satellite imagery. University of Ghana http://ugspace.ug.edu.gh DEDICATIO N I dedicate this work to my parents Joseph and Josephine Quashigah and to my brothers and sisters and all loved ones and friends who in diverse ways have contributed to my success. I Love you all and may God richly bless you. University of Ghana http://ugspace.ug.edu.gh A C K N O W LED G EM EN TS To my Lord and Saviour Jesus Christ who gave meaning to my life and has brought me this far, be all Glory, honour and adoration forever and ever, Amen. I am so grateful to the Lord for seeing me through this study for giving me strength and wisdom to complete this work successfully. I am also very grateful to my supervisors; Dr. K. Appeaning Addo and Mr. S.K. Kufogbe for their unflinging support throughout this work. My gratitude also goes to Prof. Scot M itchell and Prof. Doug King o f Carleton University for their immerse support whilst I was an exchange student in Canada. To all GEOG 5803 students (W inter, 2 0 1 1 ) and researchers at the GLEL laboratory at Carleton I am very much grateful. I would also like to thank USGS and Digital Globe for making data and software available free for my study and also the TALIF Remote Sensing and GIS laboratory for access to software and data as well as technical support. To Mr. Obodai o f Geography Department Map Library I say thank you for helping me with map data. N ot forgetting my colleagues (Environmental Science Programme, 200 9 //2 0 1 0) especially Yaw Agyeman Boafo who spent time to read through my draft. Finally to my parents for bringing me up and teaching me the way to go I am forever grateful. N ot forgetting my siblings for their financial support and help during fieldwork. Lastly, I would like thank my pastors and spiritual leaders for their encouragement and prayers. God Bless Us A ll!!! University of Ghana http://ugspace.ug.edu.gh TABLE O F C O N TE N T D ECLARATIO N.................................................................................................................................. i ABSTR A C T......................................................................................................................................... ii DEDICATIO N.................................................................................................................................... iii A C K N O W LE D G E M E N T S............................................................................................................. iv TABLE OF CO NTENT..................................................................................................................... v LIST OF FIG U R E S........................................................................................................................... x LIST OF T A B L E S........................................................................................................................... xii LIST OF A B BR E V IA T IO N S....................................................................................................... xiii CHAPTER O N E ................................................................................................................................ 1 INTRO DUCTIO N.............................................................................................................................. 1 1.1 O verview ................................................................................................................................................... 1 1.2 Significance o f Study............................................................................................................................. 5 1.3 General O bjective................................................................................................................................... 8 1.3.1 Specific objectives w ere:.............................................................................................................. 8 1.4 Justification o f Objectives..................................................................................................................... 8 1.6 Thesis O rganization............................................................................................................................. 10 CHAPTER T W O ............................................................................................................................. 12 v University of Ghana http://ugspace.ug.edu.gh DEFINITIONS AND CO NCEPTS................................................................................................ 12 2.1 Introduction.............................................................................................................................................12 2.2 Defining Coastal Zones........................................................................................................................12 2.3 Challenges o f Coastal Z ones.............................................................................................................. 13 2.4 Defining Shoreline................................................................................................................................ 14 2.5 Factors affecting Shoreline C hange................................................................................................. 15 2.5.1 Wave and Current a c tio n ............................................................................................................ 16 2.5.2 Sea Level R ise................................................................................................................................17 2.5.3 Sediment Supply........................................................................................................................... 19 2.5.4 Types o f C o a st.............................................................................................................................. 20 2.5.5 Anthropogenic Factors.................................................................................................................21 2.6 Shoreline M anagem ent........................................................................................................................22 2.7 Remote Sensing o f Coastal Environm ents..................................................................................... 23 2.7.1 Shoreline Extraction from Remote Sensing Im agery......................................................... 25 2.7.2 Sources o f Error in Shoreline M apping..................................................................................27 2.8 Shoreline Change A nalysis.................................................................................................................28 2.8.1 Net Shoreline Movement (N S M )..............................................................................................28 2.8.2 End Point Rate (E P R )..................................................................................................................29 2.8.3 Average o f R ates........................................................................................................................... 30 2.8.4 Linear Regression Rate (L R R ).................................................................................................. 31 2.8.5 Jackknifmg M eth od ..................................................................................................................... 32 2.8.6 Weighted Linear Regression (W L R ).......................................................................................33 2.8.7 Comparison o f M ethods.............................................................................................................. 34 vi University of Ghana http://ugspace.ug.edu.gh 2.9 Long Term and Short Term Rates o f C hange................................................................................35 CH APTER T H R E E ......................................................................................................................... 37 PROFILE OF STUDY A R E A ........................................................................................................37 3.1 Introduction............................................................................................................................................. 37 3.2 Location and C haracteristics...............................................................................................................37 3.3 G eology.................................................................................................................................................... 39 3.4 C lim ate..................................................................................................................................................... 40 3.5 W aves, Tides and C urrent................................................................................................................... 41 3.6 V egetation............................................................................................................................................... 42 3.7 Settlement and Economic A ctivities................................................................................................ 43 3.8 Land U s e ................................................................................................................................................. 44 3.9 Erosion along the Eastern C o ast....................................................................................................... 45 3 .10 Keta Sea Erosion: M itigation and C hallenges............................................................................ 46 3.4.1 Early M itigation M easures......................................................................................................... 47 3.4.2 Challenges...................................................................................................................................... 49 3.4.3 The Keta Sea Defence Project (K S D P ).................................................................................. 49 CH APTER FO UR............................................................................................................................ 53 M ET H O D O L O G Y .......................................................................................................................... 53 4.1 Introduction............................................................................................................................................ 53 4.2 M ateria ls.................................................................................................................................................53 4.2.1 Satellite Im agery...........................................................................................................................53 vii University of Ghana http://ugspace.ug.edu.gh 4.2.2 Other D ata .......................................................................................................................................56 4.3 Image Pre-processing........................................................................................................................... 57 4 .3 .2 GPS Data C ollection.................................................................................................................... 58 4.4 Shoreline extraction..............................................................................................................................59 4.5 Shoreline preparation and change analysis.....................................................................................61 4.5.1 Geodatabase Developm ent..........................................................................................................61 4.5.1 Baseline C onstruction..................................................................................................................62 4.5.3 Casting T ransects.......................................................................................................................... 62 4 .5 .4 Uncertainty Quantification..........................................................................................................63 4.6 Change Rate Calculation..................................................................................................................... 65 4.7 Accuracy A ssessm ent..........................................................................................................................66 C H A P T E R F I V E ........................................................................................................................................... 68 R E S U L T S ......................................................................................................................................................... 68 5.1 Introduction............................................................................................................................................ 68 5.2 Extracted shorelines............................................................................................................................. 68 5.2 Overall C hanges.................................................................................................................................... 69 5.3 Shoreline Change between 1986 and 2 0 1 1 .................................................................................... 74 5.4 Shoreline Change between 2001 and 2 0 1 1 .................................................................................... 75 5.5 Accuracy A ssessm ent..........................................................................................................................78 C H A P T E R S IX ...............................................................................................................................................81 D ISCU SSIO N O F R E S U L T S .................................................................................................................... 81 viii University of Ghana http://ugspace.ug.edu.gh 6.1 Introduction.............................................................................................................................................81 6.2 Erosion T rends....................................................................................................................................... 81 6.3 Factors Influencing E rosion................................................................................................................83 6.3.1 W aves Currents and T id e s ..........................................................................................................83 6.3.2 Construction o f the Akosombo D a m .......................................................................................84 6.2.3 Sea Level R ise................................................................................................................................85 6.2.4 Shoreline O rientation...................................................................................................................85 6.2.5 The Keta Sea Defence Project (K S D P ).................................................................................. 87 6.2.6 The Land Squeeze Factor............................................................................................................88 6.2.7 Sand mining and M angrove H arv estin g .................................................................................89 C H A P T E R SE V E N ........................................................................................................................................ 90 C O N C L U S IO N ............................................................................................................................................... 90 7.1 Introduction............................................................................................................................................ 90 7.2 Data Source and Approach to Study................................................................................................ 90 7.3 Summary o f R esults.............................................................................................................................91 7.4 C onclusion............................................................................................................................................. 92 7.4 Recom m endations.................................................................................................................................93 R E F E R E N C E S ................................................................................................................................................95 A P P E N D IC E S ...............................................................................................................................................108 ix University of Ghana http://ugspace.ug.edu.gh LIST OF FIGU RES Figure 2. 1 The Bruun Rule for Shoreline Response to Sea Level Rise (Bruun, 1 9 6 2 ) ................ 18 Figure 2. 2 End Point Rate (Him m elstoss, 2 0 0 9 ) .................................................................................... 30 Figure 2. 3 Linear Regression Rate (Him m elstoss, 2 0 0 9 ) ..................................................................... 32 Figure 2. 4 Jackknifing Method (Dolan et al., 1 9 9 1 )...............................................................................33 Figure 2. 5 W eighted Linear Regression Rate (Himmelstoss, 2 0 0 9 ) ................................................. 34 Figure 3. 1 Location O f The Study Area (A fter Ly, 1980; Ghana Survey D epartm ent)................ 38 Figure 3. 2 One O f The Largest Creeks At Atiteti (Author, June, 2 0 1 1 ) .......................................... 39 Figure 3. 3 Sandy Coast, Atiteti (Author, June, 2 0 1 1 ) ........................................................................... 40 Figure 3. 4 Vegetation: (A ) M angrove, (B ) Coconut (Author, June, 2 0 1 1 ) ..................................... 43 Figure 3. 5 Longshore Drift Along The Coast O f Ghana (Ly, 1 9 8 0 ) ................................................. 46 Figure 3. 6 Destruction O f Buildings At Keta (Gldd, 2 0 0 0 ) ................................................................ 47 Figure 3. 7 Plan O f The Keta Sea Defence Project (Gldd, 2 0 0 1 ) ........................................................ 50 Figure 3. 8 Aerial View O f The Keta Sea Defence (Gldd, 2 0 0 1 ) ........................................................ 51 Figure 3. 9 One O f The Groynes At Kedzi (Author, June, 2 0 1 1 )........................................................ 51 Figure 4. 1 Color Composites o f the Satellite Im ageries........................................................................ 55 Figure 4. 2 Tracking the Wet/Dry B oundary.............................................................................................59 Figure 4. 3 Transects and B aseline.............................................................................................................. 63 Figure 5. 1 Extracted Shorelines for a Section O f The Coast................................................................ 69 Figure 5. 2 Overall Erosion and Accretion Rates..................................................................................... 71 Figure 5. 3 Areas o f Erosion and A ccretion..............................................................................................73 Figure 5. 4 Erosion and Accretion Rates Between 1986 and 2001 ..................................................... 74 x University of Ghana http://ugspace.ug.edu.gh Figure 5. 5 Erosion and Accretion Rates For 2001 to 2 0 1 1 ..................................................................76 Figure 5. 6 Destruction by Erosion at (a) Blekusu and (b) A to rk o r.................................................. 77 Figure 5. 7 Overlay o f 1986 Vector Shoreline on Original Im a g e .....................................................78 Figure 5. 8 Overlays o f 1986 and 2011 on 2 0 1 0 Im ag ery .................................................................... 80 Figure 6. 1 The Shoreline Showing the Cape (G oogle Earth.................................................................86 XI University of Ghana http://ugspace.ug.edu.gh LIST O F TABLES Table 3. 1 Wave and W ind D ata..................................................................................................................42 Table 4. 1 Imagery Data Properties.............................................................................................................. 56 Table 5. 1 Average Erosion and Accretion R ates..................................................................................... 70 Table 5. 2 NSM Results used for Accuracy A ssessm ent........................................................................79 Table 5. 3 Uncertainty L evels........................................................................................................................79 X I 1 University of Ghana http://ugspace.ug.edu.gh LIST O F ABBREVIATIONS ASTER Advanced Spaceborne Thermal Emission and Reflection Radiometer DSAS Digital Shoreline Analysis System EPR End Point Rate ETM+ Enhanced Thematic M apper Plus GCP Ground Control Points GIS Geographic Information System GLDD Great Lakes Dredge and Docks GPS Global Positioning System HWL High Water Line ICZM Integrated Coastal Zone M anagement JK Jack Knifing KSDP Keta Sea Defence Project LiDAR Light Detecting and Ranging LR Linear Regression MS Multi-Spectral NOAA National Oceanic and Atmospheric Administration NSM Net Shoreline M ovement OLS Ordinary Least Squares RMS Root Mean Square SMP Shoreline M anagement Planning SLR Sea level rise TM Thematic Mapper University of Ghana http://ugspace.ug.edu.gh UTM Universal Transverse-M ercator VNIR Visible Near Infrared WGS World Geodetic System W LR W eighted Linear Regression XIV University of Ghana http://ugspace.ug.edu.gh C H A P T E R O N E IN TR ODU CTIO N 1.1 Overview As humankind move to the twenty first century, environmental changes are predicted to accelerate, with unknown and potentially devastating consequences (Lunetta, 1999). The coastal zone is the most dynamic interface between land and sea and represents a challenging frontier between human civilization and environm ental conservation. W orldwide, over 38% o f human population lives in the coastal zones and this population is on the increase (W ang, 2 0 1 0 ). Due to the wide range o f natural resources available, the zone is considered suitable for residential, communication, recreational and economic development. They are highly valued and greatly attractive as sites for resorts, vacation destinations as well as for port, harbor and industrial facilities. It also provides some o f the most productive and richest habitats on earth (Shan and Hussain, 2010; Korakandy, 2005; Charlier and Charlier, 1995). Currently coastal zones are facing intensified natural and anthropogenic disturbances including sea level rise, coastal erosion, over exploitation o f resources among others. Over 70% o f the worlds beaches are experiencing coastal erosion and this presents a serious hazard to many coastal regions (A ppeaning Addo et al., 2 0 0 8 ). According to Zhang (2 0 1 0 ), awareness o f the quality o f global coastal ecosystems being adversely impacted by multiple driving forces has accelerated efforts to assess, m onitor and mitigate coastal stressors. M onitoring spatio-temporal changes o f coastal environments 1 University of Ghana http://ugspace.ug.edu.gh can help understand among others, the spatial distribution o f erosion hazards, predicting their development trend and supporting the m echanism research on coastal erosion and its counter measures. For coastal zone monitoring, shoreline extraction from remotely sensed data in various times is a fundamental work. The shoreline, which is defined as the position o f the land- water interface at one instant in tim e ( Gens, 2 0 1 0 ) is a dynamic feature and is an indicator for coastal erosion or accretion. The processes o f erosion and accretion affect human life, cultivation and natural resources along the coast. Rapid shoreline changes can create catastrophic social and economic problems along populated strands. Design of viable land-use and protection strategies to reduce potential loss is necessary and this requires com prehension o f regional shoreline dynamics (B lodget et al., 1991; Chu et al., 2 0 0 6 ). Coastal m anagem ent and engineering design require information about where the shoreline is, where it has been in the past, and where it is predicted to be in the future (B oak and Turner, 2 0 0 5 ). According to Alves (2 0 0 7 ), the analysis o f historical shoreline data can be useful to identify the predom inant coastal processes operating in specific coastal locations using lateral change rates as an indicator o f shoreline dynamics. The real importance o f such studies is to avoid decisions based on insufficient knowledge, wrong assessments or arbitrary decisions, leading to losses in resources and infrastructure that could have been prevented. 2 University of Ghana http://ugspace.ug.edu.gh Shoreline changes occur over a wide range o f time scales from geological to short lived extreme events (A ppeaning Addo, 2 00 9 ). These changes are mainly associated with waves, tides, winds, periodic storms, sea-level change, and the geomorphic processes o f erosion and accretion and human activities (V an and Bihn, 2 00 8 ). W hile there is no doubt that shorelines are changing, the nature o f changes is complex and the m agnitude is uneven and vary from one point to another (Cam field and M orang, 1996). The detection and m easurem ent o f shoreline changes is therefore an important task in environmental monitoring and coastal zone m anagement (V an and Bihn, 2 0 0 8 ). According to Appeaning Addo (2 0 0 9 ) historic shoreline change information, which portray a cumulative outcome o f the processes that altered the shoreline for the periods analysed, facilitate form ulating effective coastal m anagement strategies and planning by revealing trends. The study o f changing shorelines has become more than a topic o f scientific curiosity with the increasing population in coastal areas (M oore, 200 0 ). Data sets spanning several years are desirable as a basis for such studies. Shoreline monitoring is however a challenging task. The conventional ground surveying methods used for monitoring shorelines can achieve high accuracy o f measurement, but is labour intensive, costly and time consuming (V an and Bihn, 2008; Kuleli, 2010; Appeaning Addo, 2 0 0 9 ). This limits the generation o f consistent data and monitoring large and inaccessible areas especially in developing countries. 3 University of Ghana http://ugspace.ug.edu.gh Satellite remote sensing techniques provide a synoptic vision o f the Earth that is not possible to obtain other than by exhaustive and expensive field evaluations. Data from remote sensors allow analysis o f a region with sufficient accuracy in an efficient, rapid and low-cost way (Berlanga-Robles and Ruiz-Luna, 2 0 0 2 ). It also helps in analysing areas that are poorly accessible or rapidly changing (C hu et al., 2 0 0 6 ). The use o f remote sensing data is therefore increasingly becom ing a more effective option for monitoring shoreline change. Over the years, geomorphologists, oceanographers and geologists have developed interpretation keys for mapping coastline geomorphic features using aerial photographs; however, few studies o f this type have used images generated by remote sensing orbital instruments (Kawakubo, 2 01 1 ). Though the use o f aerial photographs tends to be effective in this case, the frequency o f acquisition, cost and coverage presents a challenge. Furthermore, the spectral range o f these sources is minimal and may introduce errors in shoreline interpretation (A lesheikh et al., 2 0 0 7 ). M ulti-spectral remote sensing satellites provide digital imagery in various spectral bands, including the near infrared where the land-water interface is well defined. Furthermore this approach has advantages: not time consuming, inexpensive to implement, large ground coverage, and the capability for repeat data acquisition and monitoring (V an and Bihn, 2 0 0 8 ). The principal limitation o f satellite images is arguably their low spatial resolution when compared to photographs taken from aircraft (Kawakubo, 201 1 ). According to White and El-Asmar (1 9 9 9 ), the synoptic capability o f Landsat Thematic M apper (T M ) and Enhanced Thematic Mapper Plus (ETM +) imagery enables monitoring 4 University of Ghana http://ugspace.ug.edu.gh o f large sections o f coastlines at relatively coarse (3 0 m ) spatial resolution. Areas o f rapid change can be identified and targeted for more detailed monitoring in the field, or using higher resolution images. Rates o f erosion and deposition can be estim ated crudely, and areas where change appears to be accelerating can also be identified. Other sensors including SPOT (L e Systeme Pour d ’ Observation de la Terre), RADAR (Radio Detecting and Ranging), ASTER (A dvanced Spaceborne Thermal Emission and Reflection Radiom eter), provide similar capabilities. Though high resolution orbital data have become available only recently, it is difficult to perform multi-temporal studies for periods longer than lOyears (Kawakubo, 201 1 ). In developing countries like Ghana with a total shoreline o f about 540km and mostly inaccessible, it is expensive to m onitor using aerial photographs and high resolution imagery. There is the need to assess the effectiveness o f medium resolution satellite derived data for monitoring shoreline temporal differences especially where there is lack o f such data. W ith Landsat now freely available, there is the need for more research to explore its capability. 1.2 Significance of Study G hana’s coastal zone represents about 6 .5% o f the land area o f the country, yet houses 25% of the nation’s population. This small strip o f land now hosts 80% o f the industrial establishments in Ghana. Environmental degradation o f coastal areas has been identified 5 University of Ghana http://ugspace.ug.edu.gh as key issues in G hana’s Environmental Action Plan. Poverty, ailing human health and rapid urbanisation are very much evident (A rm ah and Amlalo 1998). Over 70% o f the 550km coastline is sandy (A rm ah and Amlalo, 1998). Coastal erosion, flooding and shoreline retreat are serious problems along the coast. According to Ly (1 9 8 0 ) the eastern coast has been identified as the most erodible stretch with rates as high as 4m/year prior to the construction o f the Akosombo Dam on the river Volta. The construction o f the Dam in the early 1 9 6 0 ’s has supposedly reduced sediment supply to this coast offsetting the balance between the sediment lost to longshore drift and replenishment (Ly, 1980). Erosion rates increased reaching as high as 8m/year around 1970. The high erosion rates have varied implications for this coast. According to Chen (1 9 9 8 ), shoreline variations have direct impact on economic development and land management. There is evidence o f loss o f land to the sea and lagoon due to erosion and flooding, among others, which affects land availability and the way in which land is used as well as the livelihoods o f the people along this coast (Fiadzigbey, 2005; Akyeampong, 2 001 ). The stretch which also serves as a nursing ground for a number o f endangered species is also being lost (Kofigah, 200 5 ). There have been interventions such as the Keta Sea Defence Project (K SD P) which involved stabilization of the shoreline with break water and groynes, construction o f a flood control structure and land reclam ation from the lagoon (GLDD, 2 0 0 1 ). These among others have influenced the accretion and erosion patterns along this coast (Appeaning Addo, 2009; Boateng, 200 9 ). 6 University of Ghana http://ugspace.ug.edu.gh Due to the economic and ecological importanc e o f this zone, there is the need for understanding these changes and the driving forces which will help in developing effective Integrated Coastal Zone M anagement strategies for the Zone. However, there is little data for example on shoreline change available on these processes. This has been attributed to the expensive and labour intensive nature o f field work, inadequate skilled personnel, among others (A rm ah and Amlalo, 1998; Appeaning Addo, 2009; Boateng, 200 9 ). Also the vast and inaccessible nature o f parts o f the coastline makes it difficult to study. Furthermore, Boateng (2 0 0 9 ) proposed the need for a holistic management o f G hana’s shorelines and this requires a general assessm ent o f the changing nature o f the shoreline over long time periods and for larger areas. It has also been recognized that there is the need to assess geomorphic processes at large scales for example, tens o f kilometres and over years to centuries in order to plan sustainable and spatial integrated measures (A ppeaning Addo, 2 00 9 ). The need for the current study therefore arose from the prevailing situations along this coast and the need to assess the influence o f the KSDP on shoreline change. Overall, this study was to investigate the capability o f using Landsat TM, ETM+ data and ASTER imagery for monitoring the dynamic shoreline o f Keta and for estimating change rates for the area which will provide relevant information for planning and intervention. 7 University of Ghana http://ugspace.ug.edu.gh 1.3 General Objective The main objective o f this study was to use medium resolution multi-temporal and multi- spectral remote sensing data to extract and detect shoreline changes from 1986 to 2011 and discuss the possible underlying factors and the implications o f such changes for the m anagement o f the zone. 1.3.1 Specific objectives were: 1. Identifying change in shoreline positions from 1986 to 2011 2. Statistically estimating historic shoreline rate o f change 3. Assessing the influence o f the KSDP on erosion 1.4 Justification of Objectives Effective m anagement strategies are required to deal with the risks arising from coastal erosion. These strategies rely on observations o f historic coastline locations and m ovement through time (D ar and Dar, 2 00 9 and Appeaning Addo, 2 00 9 ). Identifying the position o f the shoreline at any point in time is therefore important for change detection. The first objective will provide such information based on the imagery dates which span from 1986 to 2011. This would provide shoreline information that is particularly lacking for shoreline change analysis in the area. 8 University of Ghana http://ugspace.ug.edu.gh Based on the extracted shoreline positions, change detection was carried out and shoreline change rates calculated. This provided information on the nature o f shoreline change (erosion and accretion patterns). Hotspots could also be identified and targeted for planning and interventions as well as predicting future changes. According to Maiti and Bhattacharya (2 0 0 9 ), the study o f the rate o f change in shoreline position is important for a wide range o f coastal studies, such as development o f setback planning, hazard zoning, erosion-accretion studies, regional sediment budgets and conceptual or predictive modelling o f coastal morphodynamics. The third objective sought to discuss shoreline change looking at the underlying factors ( natural and anthropogenic) based on both literature and field work, focusing on the influence o f the KSDP (Figure 1.1). O bjective 1 Figure 1. 1 Flow Chart for Objectives 9 University of Ghana http://ugspace.ug.edu.gh Overall information about shoreline changes would provide basis for the implementation o f sound coastal zone m anagem ent strategies, coastal environmental protection policies and sustainable coastal development and planning schemes (A ppeaning Addo, 2009 ). Detecting shoreline changes over the 25 years period will therefore help analyse the changes that have occurred both before and after the completion o f the KSDP and also help in identifying the driving forces in order to enhance m anagement practices in the area. 1.6 Thesis Organization This thesis is organized into 7 chapters with a list o f references and appendices. Chapter 2 discusses the concepts behind the study including the coastal zone and shoreline change detection and analysis based on satellite imagery. Chapter 3 discusses the study area in terms o f location geology and hydrology. The problem o f erosion along this coast is discussed as well as major interventions such as the KSDP. Chapter 4 looks at the materials and methodology used in this study to achieve the set objectives. Chapter 5 presents the results for shoreline extraction and change detection and discuss the patterns of change with attention to the period before and after the KSDP. Chapter 6 discusses the possible factors affecting shoreline change in the study area based on previous works and current results. Chapter 7 concludes the research with summaries and recommendations for policy formulation as well as further research. 10 University of Ghana http://ugspace.ug.edu.gh 11 University of Ghana http://ugspace.ug.edu.gh C H A P T E R T W O DEFINITIONS AND C O N C EPTS 2.1 Introduction Various works have been done on coastal and shoreline monitoring. This chapter defines and discusses the various concepts related to the study based on literature review. It focuses on the shoreline, its definition and changing nature (erosion and accretion) as well as factors affecting change. It also discusses shoreline change detection and monitoring based on the application o f remote sensing and Geographic Information System (G IS ) tools. 2.2 Defining Coastal Zones The boundary between the land and ocean is generally not a clearly defined line, but occurs through a gradual transitional region. The name given to this transitional region is usually ‘coastal zone’ or ‘coastal area’ (Kay and Alder, 200 5 ). The concept of coastal zone is not defined with geographical precision. It encompasses generally the expanses on both sides o f the “land-sea boundary”; the inner part o f the coastal shell and its hinterland (Charlier and Charlier, 1995). There is therefore interaction between the two parts, the terrestrial and marine environments. It is also defined to include areas o f continental shelves, islands, or partially enclosed seas, estuaries, bays, lagoons, beaches, 12 University of Ghana http://ugspace.ug.edu.gh and terrestrial and aquatic ecosystems within watersheds that drain into coastal waters (W ang, 201 0 ). Defining the limits o f a coastal zone is limited by various factors. According to Kay and Alder (2 0 0 5 ), at policy level four possible criteria can be used; fixed distance definitions, variable distance definitions, definition according to use or hybrid definitions. The Coastal Zone Indicative M anagement Plan (C ZIM P) by the Environmental Protection Agency (E PA ) Ghana in 1990 defined the coastal zone in Ghana as the line joining the landward limits o f the lagoons, lagoonal depressions, marshes and estuarine swamps together with the intervening interfluve areas (A rm ah and Amlalo, 1998). On the average, this line approximates 10km extension from the coastline (except at river estuaries and some lagoons) and is enclosed by the 30m contour. The rationale behind the extended landward limits is to ensure the inclusion o f the catchment areas o f the coastal streams or parts thereof for the purposes o f effective land use, environmental planning and monitoring (A rm ah and Amlalo, 1998). 2.3 Challenges of Coastal Zones Coastal zones have been degraded by exploitative activities such as removal o f mangroves, sand mining, erosion from coastal development, poor land management, industrial pollution (W ong, 201 0 ). Currently the zone is facing intensified natural and anthropogenic disturbances including sea erosion, sand mining, forest/mangrove destruction and urbanization. 13 University of Ghana http://ugspace.ug.edu.gh Over 70% o f the w orld’s beaches are experiencing coastal erosion (Bird 1996; cited in Appeaning Addo, et al., 2 0 0 8 ) and this presents a serious hazard to many coastal regions. Sea level rise (SL R ), change o f storm climate and human interference has been identified as causes o f coastal erosion (Zhang et al., 2 00 4 ). Coastal erosion threatens installations and industries; contaminate water aquifers, sand bars and arable land. According to Karl et al. (2 0 0 9 ) global climate change imposes additional stress on coastal environments through sea level rise. Rising sea levels is associated with elevated tidal inundation, increased flood frequency, accelerated erosion, rising water tables, increased saltwater intrusion, and a suite o f ecological changes. These biophysical changes are expected to cause various socio-economic impacts including loss o f land infrastructure and coastal resources as well as decline in associated economic, ecological, cultural and subsistence values (D olan and Walker, 200 4 ). 2.4 Defining Shoreline According to Bird (1 9 8 5 ), the term shoreline denotes the w ater’s edge; it is usually equivalent to the high spring tide shoreline. The National Oceanic and Atmospheric Adm inistration (N O A A ) (n .d.), states the shoreline shown on nautical charts represents the line o f contact between the land and water at a selected vertical datum. In areas affected by tidal fluctuations, this is usually the mean high-water line. In confined coastal waters o f diminished tidal influence, a mean water level line may be used. It is also defined us the high waterline or wet dry boundary (Oertal, 200 5 ). Simply put the position 14 University of Ghana http://ugspace.ug.edu.gh o f the land-w ater interface at one instant in tim e defines the instantaneous shoreline (Gens, 201 0 ). Due to the dynamic nature o f the idealized shoreline boundary, investigators have typically adopted the use o f shoreline indicators to define the shoreline for practical purposes. A shoreline indicator is a feature that is used as a proxy to represent the “true” shoreline position (Boak and Turner, 2 0 0 5 ). These indicators are classified into two groups. The first group is made up o f visually discernable coastal features such as previous high tide line, the wet/dry boundary and the vegetation line. The second group is based on specific tidal datum for example mean high water or mean sea level (Boak and Turner, 2005; Alves, 200 7 ). In this way, shoreline definition and delineation depends on the selected shoreline indicator and are subjective (Alves, 2007 ). According to Moore (2 0 0 0 ), the line between wet and dry sand, which can usually be clearly seen on aerial photographs and images, is the most commonly used proxy for shoreline position. The wet/dry line closely approximates the high water line (H W L ) which in turn approximates the mean HWL. 2.5 Factors affecting Shoreline Change Historically, coastlines are in a continual state o f change; present shore locations are the result o f erosional and depositional forces. Shorelines are influenced by numerous factors including changes in sea level, tidal regime, sediment supply, periodic storms, action o f 15 University of Ghana http://ugspace.ug.edu.gh waves and winds as well as human modification. Recent studies in various geographic regions indicate that shoreline erosion has become a m ajor problem (Selvavinayagam, 2009; Hall et al., 1986). The transport o f material along the coast is linked to natural forces such as waves, tidal movements, long- and cross-shore currents, and wind. Anders and Byrnes (1 9 9 1 ) discussed five o f the primary factors that may change shoreline position: 1) wave and current processes, 2 ) sea level change, 3 ) sediment supply, 4 ) coastal geology and morphology, and 5 ) human intervention. 2.5.1 W ave and Current action The driving force behind almost every coastal process is wave action (Pethick, 1984). Ocean waves are energy travelling along the interface between ocean and atmosphere, often transferring energy from a storm far out at sea over distances several thousand kilometers (Tarbuck and Lutgens, 2 0 1 0 ). The breaking waves in the near-shore zone and the near-shore currents are responsible for the transportation o f beach sediments that results in shoreline change. The larger the waves, the more sediment will be moved (Davis and Fitzgerald, 2004 ). W aves approach the shoreline at an angle depending on the shoreline orientation and produce longshore currents. The larger the waves, the faster the longshore currents. Such highly energetic conditions will cause significant removal o f sediment or rock depending 16 University of Ghana http://ugspace.ug.edu.gh on the nature o f the coastal material (D avis and Fitzgerald, 2004; Tarbuck and Lutgens, 2010). As wave action causes sediments to become temporarily suspended currents serve as agents for moving them (Davis and Fitzgerald, 2 0 0 4 ). An ocean current is a continuous, directed horizontal m ovement o f ocean water generated by the forces acting upon this mean flow, such as breaking waves, wind, coriolis force, temperature and salinity differences and tides (D avis and Fitzgerald, 2 0 0 4 ). Strong longshore currents may move at a meter per second, a velocity that is capable o f transporting large volumes o f sand (D avis and Fitzgerald, 200 4 ). Tides are the periodic (occurring at regular intervals) variations in the surface water level o f the oceans, bays, gulfs, and inlets. They are the result o f the gravitational attraction o f the sun and the moon on the earth. Tides originate from the oceans and progress towards the coastline (NOAA, 201 1 ). The tidal current significantly affects the size, sorting and distribution o f sediment over most of the sea floor (“Go Metal Detecting”, 200 8 ). 2.5.2 Sea Level Rise Sea level rise has been identified as the principal forcing function in shoreline retreat along sandy coasts worldwide (Bird, 1996). It controls the type and magnitude o f all coastal processes: tidal range, breaker type, longshore current velocities, sedimentation rates, etc (Pethick, 1984). The Bruun (1 9 6 2 ) rule concept suggests that the entire beach 17 University of Ghana http://ugspace.ug.edu.gh profile will shift landward and upward in response to SLR (Figure 2 .1). On an equilibrium coast a sea level rise would result in a landward m igration o f the transverse shore profile, with coastline retreat, and the transference o f sand from the beach to the nearshore zone. As the shoreline is displaced landward, there is increased opportunity for erosion by waves and currents since the waves break more inland. This condition is the single most important factor in the wide spread erosion o f present shorelines (D avis and Fitzgerald, 2004; Allersm a and Tilsman, 1993). Predominant Nearshore And Offshore Material Figure 2. 1 The Bruun Rule for Shoreline Response to Sea Level Rise (Bruun, 1962) The net change o f sea level during the 5000 years o f Holocene is relatively small, but the present rate o f rise (over the last 100 years) is much greater than the average (Allersm an and Tilsman, 1993). According to Davis and Fitzgerald (2 0 0 4 ), annual rate o f global sea level rise is about 2.5m m and there are indications that this will increase over the next century. However, there are some locations where sea level rise is much higher up to about 10mm yr'1. Considering a common coast with a lm vertical change over a 18 University of Ghana http://ugspace.ug.edu.gh horizontal distance o f 50m , the 10mm yr'1 SLR (lm per century) would lead to a displacement o f the shoreline o f 50m landward. 2.5.3 Sedim ent Supply The main supply o f sediments comes from rivers and coastal erosion (Allersm a and Tilsman, 1993). Rivers supply over 90% o f the total marine sediment (Pethick, 1984). Deltas and estuaries are found where rivers contribute sediment to the coast. This positive sediment budget either infills the existing embayment or augments adjacent coastal sediment compartment (W oodroffe and Leon, 2 0 1 0 ). According to Bird (1 9 8 5 ), apart from sectors where the coast has advanced because of land reclam ation or sediment accumulation alongside artificial structures, the main sectors o f propagation have been on beaches supplied with sediment from river mouths. Elsewhere, rates o f progradation o f deltas and coastal plains have accelerated when fluvial sediment yields have increased as the result o f soil erosion due to deforestation, overgrazing, or cultivation o f steep hinterlands. There have also been examples o f the onset of erosion on deltaic coastlines following dam construction on rivers and consequent reductions in water flow and sediment yield to river mouths. Examples include the Nile delta after the construction o f the Aswan High Dam in 1964, the Volga and the Zambezi (Bird, 1985). 19 University of Ghana http://ugspace.ug.edu.gh The constant shifting o f sediment along shorelines presents a fundamental challenge to the prediction o f beach behaviour (Bam hardt, 2 00 9 ). Sediment budget is therefore very important in understanding shoreline changes. The premise behind coastal sediment budget is that, if more sediment is transported into an area than transported out o f an area, shoreline accretion results. Conversely, if more sediment is transported out o f an area than is transported in, shoreline erosion results (NOAA, n.d.). 2.5.4 Types of Coast Coasts are generally categorized as rocky, sandy, muddy coasts, coral reefs or estuaries and deltas and these are affected differently by the action o f waves and currents (W oodroffe and Leon, 2 01 0 ). Whereas cliffs are generally resistant and change only over long time scales, with evidence o f substantial changes over geological time scale in response to adjustments of sea level, sandy shorelines are much more responsive at instantaneous and event time scales, influenced by wave energy and antecedent conditions (W oodroffe and Leon, 201 0 ). Sandy coasts due to their soft geology are vulnerable to erosion. Erosion along one stretch o f sandy beach may be responsible for accretion or reduced erosion o f a nearby beach downdrift (M orton, 2004 ). Estuaries and deltas are predominantly depositional environments with sediments being supplied from upstream. This supply leads to deposition and hence accretion in and around the river mouth. Sediments are also redistricted through longshore drifts 20 University of Ghana http://ugspace.ug.edu.gh supplying sediments to adjacent shores. If the waves and currents remove more sediment than is being delivered to the shore, then retreat occurs (M orton, 2 0 0 4 ). Muddy coasts are commonly found along low-energy shorelines which either receives annual supply o f muddy sediments, or where unconsolidated muddy deposits are eroded by wave action (Flem m ing et al., 2 00 0 ). Coral reefs on the other hand supply sand to adjacent beaches and control the rates of beach erosion by reducing the energy o f incoming waves (W ielgus et al., 201 0 ). 2.5.5 Anthropogenic Factors The pressures o f increasing population, industrialization, land reclamation, sea defence projects, construction o f harbors and dams modify coastal processes. According to Bird (1 9 8 5 ), sea walls and other structures have been built to stop erosion on cliffs, beaches and delta coastlines. Many times, these lead to increased erosion along most shores (D avis and Fitzgerald, 200 4 ). Furthermore, human activities such as sand mining, harvesting o f mangrove and other vegetations as well as coral reefs that help stabilize shorelines have rendered most shoreline more vulnerable to erosion. Each o f the elements that contribute to shoreline change does not operate regularly, at constant rates. Its strength changes through time, sometimes in combination with other elements, or in ways that are contrary to the action o f other elements. Over time, the combined effect o f several factors o f variable strength may result in more complex 21 University of Ghana http://ugspace.ug.edu.gh patterns in the rates o f change, and sometimes it may even lead to abrupt shifts in drift direction or the reversal in the sediment m ovem ent process from deposition to erosion, or vice-versa (M oran, 2 00 3 ). 2.6 Shoreline M anagement The impact o f coastal erosion is a significant problem for coastal managers especially in the face o f increasing population along coastal zones. Coastal management programs must minimize loss o f life and property caused by erosion and sea level rise, while continuing to protect natural coastal resources. Therefore, the solution to shoreline erosion is not as simple as hardening shorelines with bulkheads, riprap, or groins to wall off the sea (Castellan, 2007 ). According to Boateng (2 0 0 6 ), management strategies in Ghana, both past and existing, have largely focussed on the provision o f hard protection at specific locations where risk levels to life and economic assets are high. In most cases, such ‘ad hoc’ management interventions classically tend to stabilise the shoreline at the protected section and aggravate the situation elsewhere along the shoreline ( “knock-on effects”). Due to the cumulative impacts and unsustainable nature o f such protection works, others have developed more holistic and sustainable approaches (Boateng, 2 0 0 9 ). In the United Kingdom (U K ), for example, shoreline management planning (S M P) have been developed and extensively used (Pontee and Townend, 1999). The principal aim o f SMP 22 University of Ghana http://ugspace.ug.edu.gh is to provide the basis for sustainable strategic coastal defence policies and to set objectives for the future m anagement o f the shoreline that take full account o f the interrelationships between the coastal dynamics and other environmental and planning policies o f co-operating authorities (Boateng, 2 006 ). According to Boateng (2 0 0 6 ), Ghana has the capability for implementing SMP based on similarities in physical, legislative and coastal management responsibilities. However, there are gaps in data on a number o f coastal variables that serve as bases for implementing such management approaches. This includes current rates o f erosion and accretion, hydrodynamic processes and sea level rise (Boateng, 2009; Armah and Amlalo, 1998). 2.7 Remote Sensing o f Coastal Environments Coastal zones consist o f highly productive and sensitive ecosystems, such as estuaries and lagoons. Due to the dynamic nature o f these systems, monitoring conditions by means o f remote sensing is essential for sustainability (Berlanga-Robles and Ruiz-Luna, 2 002 ). It is difficult to implement conventional field survey, requiring huge workload, high costs and long periods. Meanwhile, remote sensing techniques have been applied in coastal monitoring and environmental management for several years, with the characteristics o f large acquisition area, huge amount o f information, short period of operation and the suitability for comparative analysis (Cracknell, 1999). Moreover, 23 University of Ghana http://ugspace.ug.edu.gh integrated use o f remotely sensed data and GIS techniques provides powerful tools for monitoring and analyzing coastal tem poral-spatial changes (Zhang, 201 0 ). According to Alesheikh et al. (2 0 0 7 ), from 1807 to 1927, all coastline maps have been generated through ground surveying using the plane table and rod. From the 1920s, the aerial photogrammetric survey method became the primary shoreline mapping technique. Until the 1980s, aerial photographs were known as the sole source for coastal mapping and remain the most common data source for determining past shoreline positions (Liu, 2009; Alesheikh et al., 2007; Boak and Turner, 2 0 0 5 ). Several studies still rely on aerial photographs for shoreline analysis due to the availability o f archived photographs which is favourable for long term change detection. However, the number o f aerial photographs required for coastline mapping, even at a regional scale, is enormous. This is also challenged by difficulties in determining the shoreline and various forms o f distortion such as tilt and scale differences between photos (Boak and Turner, 2005; Dellepiane et al., 2004; Crowell et al., 1991). The launch o f the first Earth Resources Technology Satellite (E R T S -1) in 1972 (later renamed Landsat 1), enabled the United States to initiate its technological capability for the monitoring o f environmental resources and the study o f ecosystems processes from space (Lunetta, 1999). Since that time, the utility o f space-based remote sensing has been demonstrated and their potential applications for long term monitoring recognized (Lunetta, 1999). Currently there are a wide range o f sensors in space capturing data o f the surface o f the earth. These include the Landsat TM and ETM+, SPOT, RADAR, SAR 24 University of Ghana http://ugspace.ug.edu.gh (Synthetic Aperture Radar), ASTER, QuickBird, W orldView 1 and 2, among others (W ang, 2010; Lunetta, 1999). These have varied capabilities for shoreline mapping. There are various sources o f data for shoreline mapping; however, the choice o f data or sensor is influenced by factors such as availability o f data, coverage, cost o f acquisition, and resolution. In principle, the accuracy o f shoreline detection depends on the spatial resolution o f the data source. The higher the spatial resolution o f the imagery, the higher the accuracy o f the detected shoreline. Furthermore, active sensors such as SAR and airborne LIDAR allow acquisition independent o f day light conditions and are increasingly becom ing products o f choice (Gens, 2010; Liu, 2009; Dellepiane et al., 2 00 4 ). For example, the advent o f the airborne LIDAR technology has led to a more cost effective and better accuracy for shoreline mapping (Liu, 2 00 9 ). It must be noted that, high resolution satellite imagery was developed more recently and hence long term analysis cannot be carried solely based on such imagery. 2.7.1 Shoreline Extraction from Remote Sensing Imagery Various studies have been carried out using satellite imagery for shoreline extraction and change detection. Manual delineation has been the common approach. Chu et al. (2 0 0 6 ) and Cui and Li (2 0 1 0 ) used Landsat TM and Multi-Spectral Scanner (M SS) to study the pattern o f erosion and accretion at the Yellow River delta. The mean high tide line as the shoreline proxy was delineated manually by the same person at the same scale to ensure 25 University of Ghana http://ugspace.ug.edu.gh high accuracy. This approach is definition subjective and relies on individual skills (Boak and Turner, 2 00 5 ). Marfai et al. (2 0 0 8 ) used data from topographic maps, Landsat and IKONOS to monitor shoreline dynamics in Semarang, Indonesia. Visual interpretation and band analysis was used for extracting the shoreline. Bo et al. (2 0 0 0 ) employed fuzzy connectivity analysis for the semi-automatic extraction o f the shoreline using Landsat, SAR and aerial photographs. Ghanavati et al. (2 0 0 8 ) used Landsat TM and ETM+ to monitor geomorphological changes at Hendijan River Delta in south-western Iran using both band analysis and manual digitization. Maiti and Bhattacharya (2 0 0 9 ), applied thresholding, edge detection and manual digitizing to extract shoreline from Landsat MSS, TM, ETM+ and ASTER imagery to m onitor shoreline change and prediction on the East coast o f India. Other studies have employed automatic methods for extracting and analysing shorelines. These include, Van and Binh, 2008; Chand and Acharya, 2 0 1 0 and Kuleli, 2010; Dewidar and Frihy, 2007; El-Asmar and White, 2002; Frihy et al., 1994; Li and Damen, 2010; Liu and Jezek, 2004. M ethods used include, band ratio, threshholding and edge detection, classification, among others. It is evident that there is no single acceptable method for extracting shoreline from imagery. The most common method used is the visual interpretation even though it’s mostly subjective (Gens, 201 0 ). However, researchers are focusing on developing automatic 26 University of Ghana http://ugspace.ug.edu.gh methods for shoreline extraction to overcome the problems associated with this method. For example, specialized tools like the BeachTools, an extension o f ArcView™ and algorithms im plemented using ArcGIS1M are being explored for automatic extraction of shorelines (G ens, 2010; Liu, 2 009 ). 2.7.2 Sources of Error in Shoreline M apping Erosion rates can only be as accurate as the data from which they are derived and the methods by which they are calculated. Potential sources o f error are those introduced by data sources and measurement methods (M oore, 2 0 0 0 ). Determining and digitizing the high water line introduces errors as well as registering maps in different coordinate systems to the same coordinate system and projection. According to Camfield and Morang (1 9 9 6 ), environmental variables such as sun angle and haze add complexity and ambiguity to identification o f the true shoreline. Some mapped shoreline changes are therefore just artifacts o f differences in water levels rather than actual change. Though less literature is available on accounting for errors in mapping shoreline from low to medium resolution imagery, it is obvious that the resolution o f such data introduces major errors in mapping shorelines. Furthermore, atmospheric conditions and type o f sensor also affects the imagery quality and hence the information derived from it. Such images according to Gens (2 0 1 0 ) are however less affected by tidal ranges. 27 University of Ghana http://ugspace.ug.edu.gh According to Crowell et al. (1991), regardless o f how thoroughly the data have been scrutinized and corrected, some degree o f error will remain. Therefore these must be accounted for when analysing shoreline change. 2.8 Shoreline Change Analysis Change detection is a process used to identify differences o f the state o f an object or phenomenon on successive images observed at different times (Singh 1989). Shoreline change analysis is either a linear measure o f the horizontal shift in shoreline position landward or seaward or as the quantity o f material added to, or lost from, the coast (Bird, 1985). The analysis o f shoreline variability, erosion and accretion trends is fundamental to a broad range o f investigations undertaken by coastal scientists, coastal engineers, and coastal managers. Shoreline change can be calculated through various methods, including net shoreline movement (NSM), end point rate (EPR), average o f rates, linear regression, weighted linear regression and Jack-knifing among others (Cowart et al., 2010; Genz et al., 2007 Himmelstoss, 2009). 2.8.1 Net Shoreline Movement (NSM) The NSM reports a distance, not a rate. The NSM is associated with the dates o f only two shorelines. It reports the distance between the oldest and the youngest shorelines for each transect. This represents the total distance between the oldest and youngest shorelines. If 28 University of Ghana http://ugspace.ug.edu.gh this distance is divided by the number o f years elapsed between the two shoreline positions, the result is the end point rate (Himmelstoss, 2009). 2.8.2 End Point Rate (EPR) This method calculates rate o f change by using the earliest and most recent shoreline positions (Figure 2.2). The distance between the two shorelines (mostly the oldest and most recent) is measured and divided by the number o f years that have elapsed. This can be mathematically represented as: Dm Where R | is the rate, Dm is the distance in meters between the two dates and T is the time between the two shoreline positions. The EPR is the most commonly used due to the computational ease and because only two shorelines are required (Dolan et al., 1991). 29 University of Ghana http://ugspace.ug.edu.gh Given that only the end points are used, the information contained in the other data points is entirely omitted. A major drawback is if one or both end points are erroneous, the calculated erosion rate will be inaccurate (Cowart et al., 2010; Genz et al., 2007; Crowell et al., 1993). 2.8.3 Average of Rates With this method individual EPRs are calculated from the shoreline positions where more than two are available. Foster and Savage (1989) developed an equation that incorporates the accuracy o f the shoreline position data and magnitude o f rate-of-change to determine if any given EPR meets a minimum time criterion (Tm;n). „ _ V ( E i ) 2 + ( £ z ) 2 1 min KD i 30 University of Ghana http://ugspace.ug.edu.gh Where Ei and E2 are measurement errors in the first point and second point, respectively and Ri is the EPR o f the longest time span for a particular transect. EPRs are determined between all data point pairs and are removed if the time interval is less than the specified minimum. All EPRs that pass the criterion are averaged to determine the shoreline change rate (Genz et al., 2007; Dolan et al., 1991). An advantage o f this method is all ‘good’ data, that is the all EPRs that survive the minimum time span are utilized and short term variability are also filtered. A drawback however is that the minimum time criterion can be affected by large errors or small EPRs. The method is also biased, giving more influence to EPRs o f short time span (Genz et al., 2007; Dolan et al., 1991). 2.8.4 Linear Regression Rate (LRR) The linear regression a pproach also referred t o as the ordinary least squares (OLS) statistic is determined by fitting least-squares regression lines to all shoreline points for a particular transect. The least squares regression assumes independent Gaussian errors and estimates the trend o f shoreline data by minimizing the sum o f the squared residuals between the data and line. The slope o f the line is an estimate o f the shoreline rate-of- change (Figure 2.3). The assumption o f Gaussian errors is usually valid, since the sum of many sources o f error tends towards Gaussian distribution (Genz et al., 2007; Dolan et al., 1991). 31 University of Ghana http://ugspace.ug.edu.gh Shoreline Date (years) Figure 2. 3 Linear Regression Rate (Himmelstoss, 2009) A major advantage o f LR is that all data available is used and the method is purely computational. The calculation is also based on accepted statistical concepts and it is easy to employ (Genz et al., 2007; Dolan et al., 1991). 2.8.5 Jackltnifing Method The jackknifing method (JK) uses multiple OLS fits to determine the shoreline change rate (Figure 2.4). The method uses all possible combinations o f LRs given by omitting one point for each iteration. A different point for each line is omitted, resulting in a different slope for each line. The slopes are averaged to provide a shoreline change rate (Genz et al., 2007, Dolan et al., 1991). 32 University of Ghana http://ugspace.ug.edu.gh 0 E ■LR fil without 1850 point z •25 ■LR fit without 1938 point 9 -50 h ; ■LR fit without 1950 point t o -75 O CL -100 LLI z ■125 i LLI -150 OC o -175 X 00 -200 1840 1860 1860 1900 1920 1940 1960 1980 2000 DATE Figure 2. 4 Jackknifing Method (Dolan et al., 1991) The JK method as the LR is purely computational. It has the advantage o f decreasing the influence o f clustering data and extreme data points. However, computing all possible linear trends is not efficient (Genz et al., 2007, Dolan et al., 1991). 2.8.6 W eighted Linear Regression (WLR) In computing weighted linear regression, more reliable data (i.e. shoreline positions with smaller uncertainty) are given greater emphasis or weight towards determining a best-fit line (Figure 2.5). The weight (w) is defined as a function o f the variance in the measurement (e). w = l / ( e 2) Where e is the shoreline uncertainty value 33 University of Ghana http://ugspace.ug.edu.gh Since uncertainty o f the shoreline feature is used to calculate a weight, this approach requires that uncertainties in shoreline positions be identified; however this is difficult in most cases. 120 100 ?■ o E BO Eo o 40 oc (A O 20 0 1930 1940 1950 I960 1970 19S0 1990 2000 2010 Shoreline Date (years) Figure 2. 5 Weighted Linear Regression Rate (Himmelstoss, 2009) This method is also sensitive to outliers even if their weights are small. Furthermore if the calculated uncertainties do not accurately express the real deviations, then the resulting rate may under estimate or overestimate the true rate (Genz et al, 2007). 2.8.7 Comparison of Methods Aside the methods discussed in the previous subsections, other statistical methods are available and new ones being developed. Selecting the method to use is influenced by 34 University of Ghana http://ugspace.ug.edu.gh factors such as availability o f data and uncertainties involved in mapping shoreline positions. If only two shoreline positions are available the obvious option is to use the EPR. Where more than two shoreline positions are available, which is mostly the case, the linear regression methods are employed. According to Genz et al. (2007), when uncertainties are mostly understood, weighted methods are recommended; conversely, if uncertainties are poorly understood, LRR and the JK and their related methods ar e recommended. 2.9 Long Term and Short Term Rates of Change Shoreline change rates can be based on long term data or short term. Boateng (2009), defined the period up to 20 years as short term, 20 to 50 years as medium term and anything above that as long term. According to Crowell et al. (1993), data sets for obtaining consistent long term rates o f change should cover at least 60-80 years in order to span short term storm events and natural decadal scale variability. Studies such as Hapke et al. (2010) used data spanning decades (200years) for calculating long term rates o f change. Also studies by Appeaning Addo et al. (2008) used 98 years span o f data to detect and estimate shoreline rate o f change at the regional scale. Such data also provide more reliable input for future projections. However access to such long term data is lacking especially in developing countries and for inaccessible shorelines. In such cases rates are calculated for shorter terms. Periods varying from a few years to about 50years may be considered a short term. Hapke et al. 35 University of Ghana http://ugspace.ug.edu.gh (2010) considered the period up to 30 years as short term in analysing shoreline change rates along the New England and Mid Atlantic Coast. 36 University of Ghana http://ugspace.ug.edu.gh C H A PTER TH R E E PROFILE OF STUDY AREA 3.1 Introduction This chapter discusses the profile o f study area in terms o f location, geology, hydrology and economic activities. It also looks at the problem o f sea erosion along the eastern coast with emphasis on Keta and some o f the interventions adopted to mitigate the situation with focus on the KSDP. 3.2 Location and Characteristics The coastal zone o f Ghana is generally divided into three sections, the western, central and eastern (Ly, 1980) (Figure 3.1). The Eastern coast, which is about 149km, stretches from Aflao (Togo Border) in the East to the Laloi Lagoon west o f Prampram. The shoreline studied covers about 52 km o f this stretch, from the eastern side o f the Volta estuary to Blekusu on the east o f Keta. This shoreline generally falls within the Keta Municipality. The area falls roughly between latitudes 5 25' and 6 20' North and between longitude 0 40' and 1 1 O' East. The landscape consists o f a large shallow lagoon, named the Keta Lagoon, surrounded by marshy areas with a sandbar (sand spit) separating the lagoon from the G ulf o f Guinea and a number o f creeks along the coast. 37 University of Ghana http://ugspace.ug.edu.gh Ptempram Ttema Winneba Cape Coast Sekondl Axim Legend 60m Contour G U LF OF GUINEA Legend • Tow n s ------------ R ivers ---------- R oads I______ I M u n ic ip a lity B o u n t i i r y 12 |___ | S tudy Area La go on Figure 3 .1 Location of the Study Area (after Ly, 1980; Ghana Survey Department) 38 University of Ghana http://ugspace.ug.edu.gh Figure 3. 2 One of the largest Creeks at Atiteti (Author, June, 2011) The sand spit is narrow; barely more than 2.5km at its widest point with a general elevation up to 2m above mean sea level (Awadzi et al., 2008; Boateng 2009). 3.3 Geology The study area basically falls within the geological setting referred to as the Keta basin. The basin is filled with 870m o f Paleozoic marine and non-marine sediment deposits. This soft geology generally comprises quaternary rocks and unconsolidated sediments made up o f clay, loose sand and gravel deposits (Akpati, 1978). Recent deposits rest on a series o f continental beds o f M iddle tertiary age. The rocks are unconsolidated limonitic argillaceous sands and gritty sands with persistent gravelly beds at their base. The tertiary sands rest on Cretaceous and Eocene age marine shale, glauconitic sandstone and limestone (Boateng, 2009; UNEP, 2004). 39 University of Ghana http://ugspace.ug.edu.gh Figure 3. 3 Sandy Coast, Atiteti (Author, June, 2011) The Volta River System, the main source o f sediment supply to this basin, consists o f a larger drainage basin, broad delta plain, narrow shelf, steep upper slope, and a large basin floor. Recent mapping o f the sea bed topography reveals the presence o f numerous canyons (valleys) from the shelf all the way to the deepwater. This shows that active erosion is taking place at the sea floor (Manu et al., 2005). 3.4 Climate The climate is dry equatorial with an average annual rainfall below 1000mm and unevenly distributed over the year. There are two maxima. The main season occurs between May and July when the south westerly monsoon winds dominate with a minor occurring between late August and early October. From November to February the north 40 University of Ghana http://ugspace.ug.edu.gh eastern harmattan winds dominate giving rise to a long dry season (Awadzi et al., 2008; A llersm aand Tilsman, 1993). The winds are generally weak. The monthly average wind speed in Ada ranges between 1.7 m/s and 2.6 m/s (Sorensen, 2003). Temperatures are quite high with mean monthly temperature o f about 30°C in the warmest month, March and about 26°C in the coldest month, August. The average minimum diurnal temperature is about 25°C and average maximum is about 33°C (Dickson and Benneh, 1995). 3.5 Waves, Tides and Current Two types o f wave approach this coast, the seas generated by the weak, local monsoon and the swell generated by storms in the southern part o f the Atlantic Ocean. The global wave model data from NOAA (Table 3.1) shows that average wave height for the area between 1997 and 2006 is 1.39m but may reach a height o f about 3m. They normally arrive from the direction between south and south west with an average period o f 10.91s but may reach a maximum o f 19.68s. 41 University of Ghana http://ugspace.ug.edu.gh Table 3. 1 Wave and Wind Data Wave Wind Wind Hs (m) Tp (s) Direction Speed Direction (° from) (m/s) ( from) Average 1.39 10.91 194.21 4.65 213.04 Max 2.82 19.68 330.64 11 358.94 Min 0.43 3.11 46.37 0.03 1.1 Source: Svasek Hydraulics, (2006) This coast is influenced by a semi-diurnal tide with an average range o f about lm . The tidal currents caused by this tides are weak. Stronger currents occur in inlets and estuaries. The Guinea current flows offshore from west to east with velocity between 1 m/s (max 1.5 m/s) in summer and 0.5 m/s (max 0.7 m/s) in winter (Allersma and Tilsman, 1993; Sorenson et al., 2003). 3.6 V egetation The zone falls within the coastal savanna zone o f Ghana and is relatively dry. Vegetation normally comprises o f coastal strands and mangrove. At the fringes o f the Lagoon, the vegetation consists o f small and scattered clumps o f short trees and mangrove. Both red (Rhizophora sp.) and white (Aveicennia sp.) mangroves are found especially near the estuary (Figure 3.4a). Patches o f the grass Paspalum sp. and the herb Sesuvium portulacastrum as well as the sedge Cyperus articulates and the cat-tail Typha domingensis are common (Sorensen et al., 2003; Kufogbe, 1997). Coconut plantations (Figure 3.4 b) are also found along the entire coast especially around the Cape St. Paul 42 University of Ghana http://ugspace.ug.edu.gh and the estuary. Some o f this coconut vegetation is being destroyed by the erosion around Atorkor. Vegetation along the coast helps keep sediments in place. (a) (b) Figure 3. 4 Vegetation: (a) Mangrove, (b) Coconut (Author, June, 2011) 3.7 Settlem ent and E conom ic A ctivities Generally, the population o f Keta has been growing at a relatively low rate o f 0.5% since 1970 and 1.3 between 1984 and 2000 (Ghana Statistical Service, 2005). The average population density for the area is 164 persons/km 2 with figures reaching 500 persons/km 2 for areas between Keta and Anloga which compares favourably to densities in Greater Accra (609.7 persons/km2) (Ghana Statistical Service, 2005). Population o f the area shows an increasing trend until the onset o f the recent sea erosion when people began to migrate to other towns (Keta Municipal Assembly, 2011). 43 University of Ghana http://ugspace.ug.edu.gh The main economic activity o f the people living in the area is fishing, both from the sea and the lagoon. Marine fishing is carried out along the 75km stretch o f coastline from Aflao in the east to the Volta River at Atiteti/Anyanui (Kufogbe, 1997). Communities around the Keta lagoon are also involved in fishing from the lagoon. Shallot and other vegetable farming occur extensively on the sand spit throughout the year. The zone is also noted for coconut and mango plantation farming. Poultry farming is also an important contributor to the local economy (Nukunya, 1997). Wood cutting (Mangrove harvesting) is an important economic activity. This is intensive around Anyanui, Atorkor and Salo for domestic and commercial use. Both the red and the white mangrove are harvested (Keta Municipal Assembly, n.d). The mining o f sand within the Keta Municipality until recently was an established economic activity. Places like Dzita, Atorkor, Dzelukope, Tegbi and Woe are extensively mined. Salt is also mined along the banks o f the lagoon though this is restricted to the dry season (Keta Municipal Assmbly, n.d, Ahiawodzi, 1997). 3.8 Land Use Keta inherits a system o f land ownership that is common to the people o f the sandbar. The allodial title to land is vested in the clan heads but often a piece o f land is associated with an individual rather than a whole clan, lineage or family. Such a land is usually one o f the numerous patches, which were not claimed by any o f the clans during the time of their early settlement in the area because the sites were considered unproductive or 44 University of Ghana http://ugspace.ug.edu.gh useless at the time. They therefore remained no m an’s land until much later when they were claimed by the said individuals (Fiadzigbey, 2005). In terms o f land use, Keta is noted for its commercial and residential land uses with minimal agricultural use due to scarcity o f land and poor soils and climate (Akyeampong, 2001; Nukunya, 1997). Land was also lost to the lagoon through flooding as well as to intensive sea erosion putting great pressure on existing land. Reclaiming o f land from the shores o f the lagoon was an important solution causing the Town to expand towards the lagoon. Residents also began to relocate to Dzelukope on the west o f Keta initiating a linear growth. Government buildings were also rebuilt at Dzelukope (Akyeampong, 2001 ). 3.9 Erosion along the Eastern Coast The entire coast o f Ghana is influenced by longshore transport o f sediment from west to east (Figure3.5). According to Ly (1980), shoreline retreats along the eastern coast is due to the removal o f sand from the unconsolidated Quaternary sediments exposed at the shoreline to the littoral zone to compensate the sand loss caused by longshore transport. Moderate erosion has been experienced at the frontage o f Keta due the loss o f littoral sediment into lagoon inlet (Boateng, 2009). Sediment supply from the Volta River is an important balance for this removal. However, shortage o f littoral sediment was created (i.e. from about 71 million m3/annum to about 7 45 University of Ghana http://ugspace.ug.edu.gh million m3/ annum since the construction o f the Akosombo Dam on River Volta beginning 1961 (Boateng, 2009; Ly, 1980). In effect there is a sediment deficit contributing to increased sea erosion along this coast. The rates reached as high as 8 m/year between 1965 and 1980 (Ly, 1980). 3.10 Keta Sea Erosion: Mitigation and Challenges Between 1784 and 1907 there is evidence o f accretion o f at least 600 feet at the Keta shore. Between 1907 and 1932, coastal erosion reversed the coastline at Keta to its 1784 position and coastal erosion had assumed alarming proportions. Thus the onset of sustained erosion at the Keta coast started around 1907 (Akyeampong, 2001). Infrastructures such as the Evangelical Presbyterian Church and the AME Zion Church as well as government buildings have been eroded (Akyeampong, 2001). The Danish built 46 University of Ghana http://ugspace.ug.edu.gh Fort Prinzenstein has also suffered from the sea erosion. It is estimated that 70% o f the Keta Town now lies under the sea (Fiadzigbey, 2005). Figure 3. 6 Destruction of buildings at Keta (GLDD, 2000) As a result o f erosion, the sand spit became narrower especially between Keta and Havedzi cutting o ff the road linking these communities. The Keta market was badly affected since traders from the eastern towns could not have access to the market. The Keta St. M ichael’s Catholic Church was also affected. 3.4.1 Early Mitigation Measures From 1907 the colonial government began to measure and record high water marks at Keta in order to assess the progress o f sea erosion. By 1923 the destructive nature o f the sea forced the government into action and plans were made to erect groynes along part o f 47 University of Ghana http://ugspace.ug.edu.gh the seashore. Timber groynes were sunk from 9th October 1923, with the aim o f breaking the force o f the sea waves and facilitating the deposition o f sand carried by the sea in its longshore movements. This worked for some time with accretion o f up to about 2 feet but by August 1926 the groynes had collapsed and erosion worsened (Akyeampong, 2001). The Anlo State Council in 1938 built a retaining wall along the shore at Keta using iron rails and coconut trees. A wall o f railway lines and palm trees was built near the Presbyterian Mission but this was washed away. However considerable accretion took place near the Roman Catholic Church (Akyeampong, 2001). In 1951 the new African government took a major initiative to check sea erosion. The actual work started in January 1952 with a number o f pile-driven groynes established. By May 1952, one o f the groynes was washes away and erosion resumed especially near the Roman Catholic Church (Akyeampong, 2001). Later, in the 1960’s metals were used as sea Defence structures to curb the situation. This failed because the metals quickly eroded and the sea continued to advance. Between 1972 and 1978, there were further attempts to intervene and boulders were placed at vantage points some o f which remained till recently (H. James-Ocloo, Personal Communication, 8 th June, 2011). There had also been other attempts from individuals and groups using coconut stems, timber, sand bags and steel-sheet piles and concrete walls. However, none o f these attempts were able to solve the problem (Dordor, 2005). As part o f efforts to protect the Roman Catholic Church for example, there was the use o f sand and silt from the lagoon 48 University of Ghana http://ugspace.ug.edu.gh which was used to fill sacks to protect the shoreline. Old tyres were also woven together and used to fill the depression around the church to reduce the impact waves (J. Quashigah, personal communication, June 2, 2011). 3.4.2 Challenges Generally, the sandy nature o f the Keta beach accelerated coastal erosion since mobile sand presents no resistance to the sea. The area is also subject to land subsidence which makes it vulnerable to transgression by the sea (Akyeampong, 2001). The heavily populated shoreline also makes the impacts o f sea erosion very devastating. Attempts to curb the problem were also less effective because, the erosion points were highly unpredictable; they shifted constantly and were not confined to any definite area. Furthermore the cost o f undertaking Defence works was high and government was reluctant in investing in such a project and there was less support from external sources (Akyeampong, 2001). 3.4.3 The Keta Sea Defence Project (KSDP) As a major attempt to address the sea erosion problem at Keta, the Keta Sea Defence Project (KSDP) initiated in the early 1990’s. The project commenced on December 14, 1999 and was to last a period o f 50 months. The project was executed by the Great Lakes 49 University of Ghana http://ugspace.ug.edu.gh Dredge and Docks Company (GLDD), USA in association with W.F. Baird and Associates (Conterra Limited, 2005). The project had four major components which included (a) sea Defence works to limit further erosion (groynes, revetments, beach nourishment), (b) land reclamation from the lagoon adjacent to the town o f Keta, providing an area for local inhabitants to rebuild homes that were lost to erosion; (c) construction o f a road between Keta and Havedzi, re­ establishing a road link between these townships lost to erosion; and (d) flood control for Keta Lagoon, providing relief from extreme flooding conditions for the inhabitants around the lagoon (GLDD, 2001). Figure 3.7 shows a section o f the plan for the KSDP and Figure 3.8 is an aerial view o f a section o f the completed project. B Mjbiui t l l j r d i Figure 3. 7 Plan of the Keta Sea Defence Project (GLDD, 2001) 50 University of Ghana http://ugspace.ug.edu.gh Reclaimed Land Revetm ent Figure 3. 8 Aerial View of the Keta Sea Defence (GLDD, 2001) Figure 3. 9 One of the Groynes at Kedzi (Author, June, 2011) 51 University of Ghana http://ugspace.ug.edu.gh A total o f 11 million cubic meters o f sand were dredged from the channel for lagoon reclamation and beach fill and a total o f one million tonnes o f rock were produced in a quarry at M etrikasa 40km away for building roads and groynes and revetments. The project was completed on schedule in February, 2004. A flood control structure comprising o f 20 gates, 80.5m long is in place to regulate lagoon levels. An 8.5km bituminous surfaced road between Keta and Havedzi was also completed and 240 hectares o f land had been reclaimed (Conterra Limited, 2005). Additional benefits from the Project include the creation o f an 8 km long dredged channel (11m deep) from Kedzi to Havedzi to be used for fishing, water sports and other tourism activities. 52 University of Ghana http://ugspace.ug.edu.gh C H A PT E R FO U R METHODOLOGY 4.1 Introduction Various approaches to shoreline mapping and change detection have been discussed in Chapter 2. The current chapter discusses the specific approach used in this study. It looks at the materials used and their sources and discusses the data processing as well as their use for change detection. Uncertainties involved mapping the shorelines and the level o f accuracy was also discussed. 4.2 Materials Materials used in this study range from medium resolution (15m-30m) to high resolution (2m) satellite imagery and GPS data. The images were collected from different sensors with varied levels o f processing and quality. Other data also include wave data, photographs and literature. This section discusses the materials used, their sources and characteristics. 4.2.1 Satellite Imagery The data used for this project included Landsat TM, ETM+ and ASTER images spanning a period o f 25 years (from 1986 to 2011). Six dates o f images (1986, 1991, 2001, 2007, 53 University of Ghana http://ugspace.ug.edu.gh 2010 and 2011) were acquired from the United States Geological Survey, Earth Resources Observation and Science Centre and Digital Globe. In addition, shoreline was estimated from GPS coordinates taken in 2011. The study area fell within one scene therefore only one scene was acquired for each year for the Landsat and ASTER imagery and a tiled W orldView-2 image covering a portion o f the study area (Table 4.1.). The Landsat images were processed to level L IT which implies radiometric, geometric and precision corrected using the same ground control points and projected to UTM Zone 3 IN (USGS, 2009; Image Metadata File). The six visible and Near Infrared (NIR) bands at 30m were used in addition to the panchromatic band at 15m o f the Landsat 7 ETM+ data. The ASTER imagery was however acquired at level LI A which implies reconstructed, unprocessed instrument data at full resolution. The Visible Near Infrared (VNIR) consisting o f 3 bands at 15m spatial resolution were processed for the analysis. The images were acquired between November and January which coincides with the dry season o f the area. Though there were some clouds present, the region o f interest (shoreline) was relatively cloud free. Furthermore due to the failure o f the Scan Line Corrector (SLC) o f the ETM+ in 2003 there were gaps in the 2011 data (Figure 4.1e). The World View-2 image covered only a portion o f the entire coast under study. The 8 multi-spectral bands at 2m spatial resolution were used in this study. The image was already tiled and projected to the UTM Zone 3 IN using the WGS 84 datum. It was relatively cloud free and o f good quality (Figure 4. If). 54 University of Ghana http://ugspace.ug.edu.gh W orldView S a te ll ite i m a g e r y Comp°s‘tes Figure 55 University of Ghana http://ugspace.ug.edu.gh 4.2.2 Other Data Field data included ground reference points and tracking o f shoreline using GPS. It also included photography, observation and interviews o f some community members. Wave and wind data for the area based on the global wave model was also collected from NOAA for the period from January 1997 to January 2006. The location for the data is latitude 4°N and longitude 1°W (Svasek Hydraulics, 2006). Table 4. 1 Imagery Data Properties Data Path/Row Acquisition Bands Resolution Level o f date Processing Landsat TM 192/56 1986-01-13 6 MS 30m LIT Landsat TM 192/56 1991-01-03 6 MS 30m LIT Landsat 6 MS 30m LIT ETM+ 192/56 2001-01-30 1 Pan 15m ASTER 192/56 2007-11-06 3 VNIR 15m L1A WorldView-2 Tiled 2 0 1 0 -1 1 - 1 0 8MS 2 m - IPan 0.5m Landsat 6 MS 30m LIT ETM+ 192/56 2 0 1 1 -0 1 - 1 0 1 Pan 15m Source: Author’s Construct 56 University of Ghana http://ugspace.ug.edu.gh 4.3 Image Pre-processing Landsat images were basically ready for use in shoreline extraction. The 1986, 1991, 2001 and 2011 Landsat images were in the same projection UTM Zone 31 using the WGS 84 reference datum and GLS 2000 elevation data. However, to ensure consistency in spatial resolution, the Landsat TM data was resampled using nearest neighbour and 1st order polynomial transformation to 15m. This however does not add any spatial information to the new data and increases the uncertainty for the shoreline position. For the Landsat ETM + data the panchromatic band with a resolution 15m was used to sharpen the six multi-spectral bands to obtain a new image at 15m. The Gram-Schmidt pan sharpening algorithm (Laben and Brover, 1998) in ENVI which is based on principal component analysis was used. Studies by Yuhendra and Hiroaki (2011) showed this method is very good at colour recovery and sharpness. The pan-sharpening improved the spatial information o f data and hence reduces the uncertainty. The ASTER VNIR bands at 15m but were acquired at LI A. Basically at its raw stage, there was a need for geometric correction/rectification. The VNIR bands were co­ registered (Image to image) to the Landsat 2001 ETM+ data using 30 visually interpreted Ground Control Points (GCP). These GCPs were used to warp and resample the ASTER using first order polynomial and nearest neighbour transformation. The total RMS error was 0.35m. 57 University of Ghana http://ugspace.ug.edu.gh The resulting images were combined to form visible and infrared color composites using Red-Green-Blue (RGB) display (refer to Figure 4.1). For the Landsat images the bands 5, 4 and 3 were combined and bands 3, 2 and 1 for ASTER. This combination enhanced the land-water boundary and highlight subtle details not readily apparent in the visible bands alone (“Web PDX”, 2001). 4.3.2 GPS Data Collection Three sections along the entire coastline were selected for extraction o f current shoreline position. Sections around the sea defence, Cape St. Paul and Atorkor were tracked. In all, a total of about 16km o f shoreline was mapped. Handheld Garmin eTrex GPS was used for tracking transects and the instrument accuracy was 3m throughout the tracking. The unit o f collection was set to WGS 84 to be consistent with the other data projections. The dry/wet boundary which was also visible from the imageries was used as a reference for tracking the shoreline positions (Figure 4.2). 58 University of Ghana http://ugspace.ug.edu.gh Figure 4. 2 Tracking the Wet/Dry Boundary 4.4 Shoreline extraction Both manual and semiautomatic methods were explored for the extraction o f the shorelines. Band ratio between the mid infrared (band 5 [b5]) and the green (band 2 [b2]) was used to identify the water-land boundary for the Landsat images except the 2011 image due to the gaps in the data. This was used to reduce the level o f subjectivity in delineating the shoreline. For this study band ratio was implemented using the band ratio model in the ENVI software (i.e. b5/b2). The resulting image with ratio values between 0 and 3 was sliced and segmented to form a binary image with values less than 1 being classified as water and values greater than 1 being classified as land thereby delineating the boundary between the water and the land 59 University of Ghana http://ugspace.ug.edu.gh as the shoreline. The water class was then converted from raster to vector and exported as shapefiles for overlay in ArcMap. In ArcMap, the extracted shorelines were overlaid on the Landsat image. The output vector however consisted o f other water/land boundaries such as those o f creeks and lagoons and could not be directly used for change detection. To extract the target sections, the extracted vector shoreline were overlaid on the colour composites and was used as guide to digitize the target shoreline. Due to the differences in sensor platforms (wavelengths), the presence o f clouds and cloud shadows over portions o f the shoreline o f the 2007 ASTER image, band analysis was considered ineffective. The 2007 shoreline was therefore manually digitized directly from the colour composite (231) o f the ASTER imagery. The shoreline was digitized at the same position as for the 1986, 1991, 2001and 2011 images and at the same scale (1:5000). For the ASTER imagery and the 2011 ETM+ data which had problems with cloud and gap, respectively, sections o f the shoreline were not digitized. Shoreline data for 2007 and 2011 therefore had gaps. The 2010 high resolution imagery was also digitized. For the tracked shoreline, the data was downloaded from the GPS using the Map source software where the tracks were extracted as line feature. The data was then saved in the ‘.d x f format and opened in ArcMap and then exported as a shapefile. 60 University of Ghana http://ugspace.ug.edu.gh 4.5 Shoreline preparation and change analysis The Digital Shoreline Analysis System (DSAS 4.2) updated by the USGS (Himmelstoss, 2009) was used for shoreline change detection. The software is an extension for ArcGIS and computes rate o f change at user specified interval along the shoreline using different methods. This was chosen due to the various statistical methods available for change rate estimation. The DSAS requires that the input data be consistent and in one geodatabase for change statistic calculation. 4.5.1 G eodatabase D evelopm ent The DSAS software requires that all input data be managed within a geodatabase which also serves as the storage location for the program-generated transect feature class and related statistical output tables (Himmelstoss, 2009). A geodatabase was therefore created for the extracted shoreline positions. Each shoreline had attributes which included date, length, ID, shape and uncertainty. The date o f acquisition for each image was entered for the date column while the length, ID and shape were automatically generated. Uncertainties were also quantified (refer to section 4.4.4) and entered as integers for the uncertainty column. The five shoreline positions were then appended to one shapefile and were ready for rate calculation. 61 University of Ghana http://ugspace.ug.edu.gh 4.5.1 B aseline C onstruction The DSAS uses measurement baseline method to calculate rate o f change statistics for a time series o f shorelines. The baseline is constructed to serve as the starting point for all transects cast by the DSAS application. For this analysis the baseline was constructed by manually digitizing about 500m onshore away from the closest shoreline (Figure 4.1) taking into consideration the general orientation o f the shoreline. The data was also projected to the UTM Zone 3 IN projection. The attributes include object_id, shape and length with an additional user generated ID column. 4.5.3 C asting Transects Once all the inputs were ready in the database, transects were constructed. A total o f 284 transects were cast along the entire stretch o f coastline from east to west at a specified interval o f 200m. As discussed in section 3.3, the geology for the shoreline is generally homogenous. In this case, little variation is expected for distances lesser than this. The transects were cast at simple right angles from the baseline offshore (Figure 4.3). The length o f the transect was set to 900m to ensure it intercepts all the 5 shoreline positions 62 University of Ghana http://ugspace.ug.edu.gh Figure 4. 3 Transects and Baseline 4.5.4 U ncertainty Q uantification For this study, uncertainties were quantified using estimates based on studies such as Crowell et al. (1991) and Moore (2000) and Hapke et al., (2010). Additional enrors, which were associated with the imagery used for this study, were estimated. Four main sources o f error were identified to account for the uncertainties. Errors resulting from image registration, digitization o f the shoreline, position o f HWL and differences in resolution were considered. As discussed in section 4.3, resampling the 1986 and 1991 63 University of Ghana http://ugspace.ug.edu.gh images from 30m to 15m did not add any spatial information. A pixel with the same spatial information is split into four to get 15m. The uncertainty estimates for this study are displayed in Table 4.2. The tidal range (lm ) o f the area was negligible and therefore was not accounted for as a source o f uncertainty due the resolution o f the imagery used. A total shoreline positional error for each epoch (Ex) was therefore calculated using the following equation: = vW+£p + £ r ) where Es is the error occurring from scale difference, Ep is the photogrammetric error and Er is the registration error. This approach carries the assumption that component errors are normally distributed (Dar and Dar, 2009). The total uncertainties were used as weights in the shoreline change calculations. The values were annualised to provide error estimation for the shoreline change rate at any given transect and expressed as: „ V ( £ 12 + £ 2 + £ | + E | + £ 52) Ea = -----------------------^ ----------------------- 64 University of Ghana http://ugspace.ug.edu.gh where E\, £ 2 , £ 3 , £ 4 and £ 5 are the total shoreline position error for the various years and T is the 25 years period o f analysis. 4.6 Change Rate Calculation For this study the W LR was used since there were more than two shoreline positions and uncertainties were also quantified. The NSM which reports the distance between two shoreline positions was also used for accuracy assessment purposes. For the purpose o f assessing accuracy o f the mapped shoreline positions (discussed in section 4.6) the net shoreline movement was used to determine the distance between the 2010 satellite image extracted shoreline and 2011 GPS mapped shoreline, a period less than a year. The distance between where the two shoreline positions intercept with a transect is reported as the net shoreline movement taking into consideration uncertainties. This distance was used as measure to determine the ability o f mapping the shoreline at 15m resolution (see section 4.7). The EPR was employed where only two shoreline positions were available as was the case for the period between 2001 and 2007. The distance between the two points where the shoreline intercept a transect was calculated and this distance was divided by the number o f years that elapsed in this case 7years, to give the end point rate. This was done along all transects and the EPR was reported for all. 65 University of Ghana http://ugspace.ug.edu.gh For the entire period under study five shoreline positions were available (i.e. 1986, 1991, 2001, 2007 and 2011) and uncertainties were also quantified. W LR method was therefore used for calculating the rates. The method was also used to calculate changes for shorter periods thus between 1986 and 2001 (period before the KSDP) and between 2001 and 2011 (the period during and after the KSDP). Both periods had three shoreline positions mapped. Here shoreline positions with smaller positional-uncertainty values had more influence in calculating the regression. The slope o f the repression line between the shoreline positions at each transect was reported as the rate o f change. This method proved to produce better rates o f change statistics. 4.7 Accuracy Assessment To assess the accuracy o f the band ratio at mapping the water/land boundary, the extracted shorelines were overlaid on the colour composites and visually interpreted. The 1986 and 2 0 1 1 shoreline positions were also overlaid on the 2 0 1 0 image to further assess their accuracy through visual interpretation. Furthermore, the accuracy o f mapping the shoreline at 15m resolution was also assessed by comparing the 2010 November shoreline to the January 2011 shoreline positions. This covers a period o f only three months and little change was expected for the shoreline positions aside from tidal effects which for the area is only about lm . 66 University of Ghana http://ugspace.ug.edu.gh The 20 10 shoreline was then compared with the 2011. The movement between the November 2010 and June 2011 lb ’ was taken as a short term movement o f the shoreline position in the study area. This was then subtracted from the movement between the November 2010 and the January 2011 a ’. The result was assumed as the error at mapping the shoreline using 15m resolution imagery. Thus: a — b = x where x is the resulting error in meters The change between these two dates was considered as an error for mapping the shoreline at 15m resolution. 67 University of Ghana http://ugspace.ug.edu.gh C H A PTER FIVE RESULTS 5.1 Introduction This chapter discusses the results from the shoreline mapping and change analysis, looking at the pattern o f change and the rates o f accretion and erosion along this coast. Changes are presented for the period before the KSDP thus from 1986 to 2001 and the period after thus from 2001 to 2011. Overall changes were presented for the 25 years period thus between 1986 and 2011. Results for a shorter period between 1986 and 2007 were also presented as well as results for the accuracy assessment. 5.2 Extracted shorelines A total o f 6 shorelines were extracted from satellite imageries and additional shorelines from GPS field survey were used for shoreline analysis and accuracy assessment. The shoreline positions were for 1986, 1991, 2001, 2007, 2010 and two shoreline positions for 2011 as shown in Figure 5.1. The shoreline positions from satellite imagery for 2007 and 2 0 1 1 had gaps. 68 University of Ghana http://ugspace.ug.edu.gh Figure 5.1 Extracted Shorelines for a Section of the Coast 5.2 Overall Changes Change rates were calculated for the period between 1986 and 2007 (Figure 5.2a) and then for 1986 to 2011 (Figure 5 .2b). This was to cross the check the effects o f the gaps in the 2011 image on the rates. The results show that there have been significant changes along the entire coast for the 25 years period under study. For the period between 1986 69 University of Ghana http://ugspace.ug.edu.gh and 2011, about 40% o f transects were ignored due to the gap in the 2011 shoreline positions. This affected change rates especially near the estuary. However, overall rates were not much affected, showing little variation in rates o f changes. The averages o f the calculated rates are shown in Table 5.1. Table 5. 1 Average Erosion and Accretion Rates Period Erosion Rate Accretion Rate Ave(m/year) (m/year) 1986-2011 1.91 2.04 1986-2007 2.38 2.77 1986-2001 3.10 5.17 2001-2011 4.52 5.59 2001-2007 4.68 10.04 Overall rates per transect varied from between -12m/year to 18m/year where negative values represent erosion and positive values represent accretion (Figure 5.2). Using the 1986 to 2007 results as a reference, about 45% o f the entire shoreline experienced erosion while the remaining have mostly accreted. No change was recorded for only one transect. Accretion rates per transect ranged from 0.1 m/year to 19m/year with an average 2.5m/year while erosion rates were between 0.1 to 9.3m/year with an average of 2.38m/year. Both rates were significantly high for the area. 70 University of Ghana http://ugspace.ug.edu.gh (a) ij . . . . . . . . . — i— i— i— i— 20 40 60 80 1 00 1 20 1 40 1 60 1 80 200 220 240 260 280 Transects 12 11 10 9 8 7 6 5 4 3 2 1 0 -1 -3 -4 -5 -6 -7 -8 -9 20 40 60 100 120 140 160 180 200 220 240 Transects Figure 5. 2 Overall Erosion and Accretion Rates 71 WLR WLR University of Ghana http://ugspace.ug.edu.gh The Keta area, as shown in Figure 5.3 (a) has seen much o f accretion with rates reaching about 18m/year while the area between Keta and Adina (Figure 5.3 b) has seen much erosion with rates at an average o f 3.5m/year with some sections recording as high as 9m/year. Near the estuary there are extreme cases o f erosion and accretion over the period (Figure. 5.3c) and rates are as high as -11 m/year and 17m/year respectively. For the entire shoreline erosion and accretion rates averaged at 2 m/year. 72 University of Ghana http://ugspace.ug.edu.gh (c) Figure 5. 3 Areas of Erosion and Accretion 73 University of Ghana http://ugspace.ug.edu.gh 5.3 Shoreline Change between 1986 and 2011 This period revealed that erosion dominates the entire shoreline (Figure. 5.4) with about 70% o f the cast transects recording erosion. Erosion rates ranged from 0.1 to 15.4 m/year with an average o f 3m/year and accretion rates ranging from 0. lm /year to 21m/year with an average o f 5.90m/year. The higher erosion rates occurred between Keta and Adina and Atorkor and Anyanui while the area between Keta and Anloga experienced significant accretion. Transects Figure 5. 4 Erosion and Accretion Rates between 1986 and 2001 Close to the estuary there is evidence o f both erosion and accretion over the period. Here erosion rates were as high asl5m /year and accretion rates also at a high o f ]4m/year. 74 University of Ghana http://ugspace.ug.edu.gh 5.4 Shoreline Change between 2001 and 2011 The period between 2001 and 2011 showed a reversal o f situations with the entire coast experiencing more accretion (about 80%) than erosion (Figure 5.5). However, erosion rates remained high, ranging from 0.1 to 17m/year with an average as high as 4.5m/year. Accretion rates also were high ranging from 0.1 to 26m/year and an average of 5.6m/year. The area between Keta and Blekusu and the area near the estuary remained high points o f erosion over this period with rates reaching as high as 16m/year. Figure 5.6 shows the impacts o f the high erosion rates around Blekusu and Atorkor. It is also evident that most areas that experienced erosion between 1986 and 2001 had accreted between 2001 and 2007. The immediate vicinity o f the Keta Township continues to accrete as well as areas around the estuary with values reaching 17m/year. Most portions o f the Cape have also accreted. 75 University of Ghana http://ugspace.ug.edu.gh (a) 2 0 0 1 to 2 0 1 1 f 20 40 60 80 100 120 140 160 180 200 220 240 Transects (b) 2001 to 2007 ;0 40 60 60 100 120 140 160 1d0 200 220 240 260 Transects Figure 5. 5 Erosion and Accretion Rates for 2001 to 2011 76 University of Ghana http://ugspace.ug.edu.gh (a) Figure 5. 6 Destruction by Erosion at (a) Blekusu and (b) Atorkor 77 University of Ghana http://ugspace.ug.edu.gh 5.5 Accuracy Assessment Visual interpretation o f the overlay o f the automatically extracted shoreline positions (Figure 5.7) show that the method was efficient in delineating the water land boundary. Additionally, three o f the extracted shorelines were used for accuracy assessment thus the 2010 and the two 2011 shoreline positions. The average o f the NSM between the November 2010 and June 2011 shoreline positions was 20.12m while that o f November 2010 and January 2011 was 31.43m (Table 5.2). From the application o f the method discussed in section 4.7, the total error for mapping the shoreline at 15m resolution was estimated to ± 11.3 lm. Figure 5. 7 Overlay of 1986 Vector Shoreline on Original Image 78 University of Ghana http://ugspace.ug.edu.gh Table 5. 2 NSM Results used for Accuracy Assessment TransectID NSM (m) NSM (m) (11/10/2010-01/10/2011) (11/10/2010-06/06/2011) 109 29.43 31.53 1 1 0 39.53 32.32 111 17.25 28.77 1 1 2 17.86 25.51 113 22.74 10.48 114 29.47 10.78 115 33.73 6.83 119 21.48 1.90 1 2 0 16.93 4.33 121 2 1 . 8 8 0.53 1 2 2 16.20 1.13 2 1 2 66.52 42.26 213 75.54 41.32 214 92.20 44.00 Average 31.43 20.12 Based on estimates (Table 5.3) the maximum annualized uncertainty for this study is ±0.44m/year. Table 5. 3 Uncertainty levels Shoreline Registration Digitizing HWL Scale Total Year Uncertainty Uncertainty Uncertainty Uncertainty Uncertainty 1986 - ± 1 ,0 0 m ±4.5m ±3m ±5.5m 1991 - ± 1 .0 0 m ±4.5m ±3m ±5.5m 2001 - ± 1 .0 0 m ±4.5m - ±4.61 m 2007 ±0.35m ± 1 .0 0 m ±4.5m - ±4.62m 2011 - ± 1 ,0 0 m ±4.5m - ±4.6 lm Annualized Transect Error (Ea) (m/year) ±0.44m Furthermore, visual interpretation o f the overlay o f the 1986 and 2011 shoreline positions on the 2 0 1 0 high resolution imagery revealed a fairly accurate mapping o f the shoreline University of Ghana http://ugspace.ug.edu.gh by the 15m resolution imageries. As shown in Figure 5.8, the area covered by circle B' was a large creek which was still part o f the shoreline during the late 1980 and early 1990’s as confirmed by community members. The circle ‘A’ also shows part o f the community around the Roman Catholic church which was also affected around the same time. The 2011 shoreline position was also approximately in line with the 2010 one which is a period o f about only three months. Figure 5. 8 Overlays of 1986 and 2011 on 2010 Imagery 80 University of Ghana http://ugspace.ug.edu.gh C H A PT E R SIX DISCUSSION OF RESULTS 6.1 Introduction Shoreline change is evident along the entire coastline o f Keta and various factors contribute to such change. This chapter discusses the erosion and accretion trends and the possible factors that are influencing the observed trends along this coast with emphasis on erosion. Natural factors including wave and current actions, shoreline orientation as well as anthropogenic factors including land, soil mining and mangrove harvesting as well as the influence o f Keta Sea Defence Project is discussed. 6.2 Erosion Trends According to Ly (1980), erosion rates along the eastern coast have increased after the construction o f the Akosombo Dam in 1962. The rates reached as high as 8 m/year as compared to the 4m/year high rates before the construction. The result o f this study revealed high rates o f erosion along the entire coast for the period under study from 1986 to 2011 thus an average rate o f about 2m/year ±0.44m. The period before the construction o f the KSDP marked intense erosion along the entire coast with rates reaching as high as 15m/year and an average of about 3.10m/year for the area near Keta and the Volta estuary. This has led to destruction o f many coastal facilities and homes especially at 81 University of Ghana http://ugspace.ug.edu.gh Keta (see section 3.4). It also confirms the assertion that the Cape has been retreating since the construction o f the Akosombo Dam (Boateng, 2009). As part o f efforts to curb the situation, the Keta Sea Defence project was initiated in 2001. Thereafter, the rates indicate more accretion to the west o f the site and erosion to the east. Erosion rates remain high, averaging 4.52m/year while accretion rates were also as high as 5.5m/year. Currently, the people around Keta are relieved as the shoreline has been stabilized and the community protected from the impacts o f erosion. However, high erosion rates in the east have led to the destruction o f houses at Blekusu and its surrounding communities. This situation is responsible for the generally higher erosion rates recorded after the KSDP. Further down to the west (near the estuary) erosion rates had also increased leading to the destruction o f homes and schools. As well the road linking Anyanui in the far west to Keta was completely cut off at Atorkor. Efforts are underway to protect this area from further erosion. Previous studies estimated erosion rates at 1.13m/year ±0.17 for the Accra shoreline (Appeaning et al., 2008) and Ly (1980) estimated for the eastern coast between 4­ 8 m/year. The current rates reflect this general trend. 82 University of Ghana http://ugspace.ug.edu.gh 6.3 Factors Influencing Erosion Natural factors continue to be a major contributor to shoreline change. Climate change which leads to sea level rise, wave, current, tides and shoreline orientation influence the movement o f shoreline at any point in time. However anthropogenic influences such as construction o f harbours, dams and coastal Defence works greatly affect coastal processes. A combination o f these factors is responsible for shoreline change along the Keta coast. 6.3.1 W aves Currents and Tides As discussed in section 3.5, waves are active in the study area and it is considered a high energy beach. The prevailing southwesterly wind causes an oblique wave approach to the shoreline. This wave approach generates an eastward littoral transport. Shoreline retreat is therefore due to removal o f sand from the unconsolidated Quaternary sediments exposed at the shoreline to the littoral zone to compensate the sand loss caused by longshore transport. Since there are no major headlands to act as obstacles to the littoral transport rates o f retreat along this coast is high (Ly, 1980; Boateng, 2009). Furthermore, due to the generally sandy nature o f the Keta beach, there is accelerated coastal erosion since mobile sand presents less resistance to wave action. The sand is easily removed from the coast and carried away by the drift. Reduction in sediment supply from other sources such as the Volta River to compensate for this loss increases the risk o f erosion along this coast (Ly, 1980). 83 University of Ghana http://ugspace.ug.edu.gh According to M anu et al. (2005) recent mapping o f the sea bed topography o f the estuary area reveals the presence o f numerous canyons (valleys) from the shelf all the way to the deep water. Waves reaching this point may behave as in deep waters. The waves will therefore break at a higher speed on suddenly reaching the shoreline. The impact is stronger and may partly explain the high erosion rates along the Keta shoreline. 6.3.2 Construction of the Akosombo Dam Sediment supply to the region from the Volta River is important. According to Ly (1980), prior to the construction o f the Akosombo Dam sediment supply from the Volta River was estimated to be 71 million m3/a. This was carried along the wave-induced littoral drift to the east. In effect there was natural accretion at Cape St. Paul. With the construction o f the Dam, this supply reduced to only 7 million m3/a leading to shortage o f littoral sediment and hence reduction in accretion along the eastern strip since the late 1960s (Boateng, 2009). Accretion was now occurring further west o f the Cape near the estuary (i.e. Atiteti) as a result o f reduced ebb tidal energy from the Volta which is caused by the reduction o f the flow of the river. Further away from the estuary the shortage o f sediment supply has led to the removal o f sediments from the shoreline to compensate for this loss leading to increased erosion. Ly (1980) confirmed that this reduction led to an increase in erosion rates from an initial high o f 4m/year to 8 m/year. As confirmed by the results o f this study, erosion dominates this shoreline prior to the construction o f the Keta Sea Defence. The erosion rates have remained fairly high. 84 University of Ghana http://ugspace.ug.edu.gh 6.2.3 Sea Level Rise Rising sea levels as a result o f climate change in concert with coastal erosion is contributing to gradual submergence o f communities along the W est African nation’s coast. Coastal systems react to changes in mean sea level by redistributing sediments. As supply o f sediments reduce along this coast the shoreline retreat (Allersma and Tilsman, 1993). Keta has been identified as highly vulnerable to increased erosion associated with sea level rise (Boateng, 2009). For the coast o f Ghana, the local sea level is rising in conformity with the global trend at a historic rate o f approximately 2 mm/yr. This is expected to increase, potentially up to as much as 6 mm/yr (Appeaning Addo, 2009). Higher sea levels exposes previously out o f reach land to waves and currents increasing the vulnerability to erosion. This partially explains the high erosion rates found in this area. It has been identified that all the frontage o f the Keta strip could be submerged by lm rise in sea level, and 2m rise may result in inundation o f the whole frontage (Boateng, 2009). 6.2.4 Shoreline Orientation Cape St. Paul is a dominant feature along this coast projecting seaward and giving a convex shaped coastline. The shoreline roughly runs in the south-east direction (Figure 6.3). As sediments are supplied and transported from the Volta estuary greater part is deposited between its mouth and a point eastward o f Cape St. Paul, where the littoral 85 University of Ghana http://ugspace.ug.edu.gh transport almost ceases. This is because the south-westerly waves cannot reach this area. Active erosion occurs here (point A, Figure 6.3), where the transport capacity increases and feeds the coast further east (Allersma and Tilsman, 1991). Figure 6. 1 The Shoreline Showing the Cape (Google Earth) This phenomenon is responsible for accretion around the Cape St. Paul prior to the construction o f the Akosombo Dam. As discussed in section 6.2.2 the construction o f the Dam reduced sediment supply to this coast. The less sediment depletes quickly before reaching the Cape leading to the removal from the Cape onwards to compensate for the loss, hence the recession o f the Cape after the construction o f the Dam. 8 6 University of Ghana http://ugspace.ug.edu.gh 6.2.5 The Keta Sea Defence Project (KSDP) As discussed in section 3.4, various attempts were made to halt the shoreline recession in the Keta area. The KSDP was the largest and was aimed at intercepting the reduced yet significant present littoral sediment drift (Boateng, 2009). The effect o f this project on shoreline change was determined by comparing shoreline change before and after the construction. Prior to the construction o f the Sea Defence, erosion was dominant along the entire coast especially around Keta. With the completion o f the project in 2004, erosion was greatly reduced as the shoreline between Keta and Havedzi was stabilized. There is evidence o f accretion along most sections o f the coast especially west o f the defence between 2001 and 2007 as a result of the construction. However, the construction o f site specific hard structures such as the Keta Sea Defence tends to stabilise a specific section o f the coastline and cause a “knock on effect" down drift (Boateng, 2009). As confirmed by this study, to the immediate east of the Sea Defence erosion is occurring at high rates leading to the destruction o f properties. Furthermore there is increased erosion closer to the Volta estuary leading to massive destruction o f coastal establishments. As part o f efforts to curb the current situation, 2.5km long gabion revetment structure is being constructed along the shoreline at Atorkor as well as the reconstruction o f the road linking the community. About 500m of the revetment has been constructed and the rest will be completed by the end o f the year 87 University of Ghana http://ugspace.ug.edu.gh 2011 (Ayivor, 2011). This is expected to stabilize the shoreline around this area but may shift focus to another point. As discussed in section 2.6, there is the need for the development o f SMPs for the management o f Ghana’s shoreline. 6.2.6 The Land Squeeze Factor Land use patterns have significant influence on erosion patterns. Increase in population along the coast puts much pressure on the natural land, making it more vulnerable to erosion. Rapid development along the coast o f Ghana has been identified as a driving force for coastal erosion in Ghana. As discussed in section 3.8, land is, and remains a scarce commodity in the Keta area and as a result there high pressure on existing land. The sandbar on which the Townships are located is barely more than 2.5km at its widest point. The land is bounded in north by the Keta Lagoon and the south by the G ulf o f Guinea. According to Schleupner (2008), coastal development prevents coasts from adapting to increased erosion rates by shifting landward. Since the sandbar is limited in the north by the lagoon, developments cannot be moved further inland. As well, the land cannot adapt to sea level rise by migrating inland. In effect the available land is competed for by wave action as well as developments leading to squeeze situation. Furthermore, most o f the settlements, historic and tourism sites and industries are within 2 0 0 m radius from the shoreline.The impact o f erosion is therefore felt in this area (Boateng 2006). 8 8 University of Ghana http://ugspace.ug.edu.gh 6.2.7 Sand mining and M angrove Harvesting As discussed in section 3.7 sand mining even though banned is being carried out in the area. Mensah (1997) has established that sand mining plays an important role in coastal erosion. The sand deposit ensures that sediment is available for littoral transport. Its removal reduces the beach volume and hence increases erosion. This has been identified as a major contributor to erosion along the Keta coast especially near Atorkor, Dzita, Dzelokope and Woe. Also, mangroves prevent erosion by stabilising sediments with their tangled root system and also trapping sediments originating from inland thereby stabilising the shoreline. Over exploitation o f mangroves has also been identified as a driving factor in increased coastal erosion. The intensified harvesting o f red and white mangroves growing around Anyanui, Atorkor and Salo for domestics and commercial use has further aggravated the soil erosion problem (Keta Municipal Assembly, n.d.). 89 University of Ghana http://ugspace.ug.edu.gh C H A PTER SEV EN CONCLUSION 7.1 Introduction The main objective o f this study used medium resolution multi-temporal and multi- spectral remote sensing data to extract and detect shoreline changes from 1986 to 2011 and discussed the possible underlying factors, particularly the influence o f the KSDP and the implications o f such changes for the management o f the zone. It involved the extraction o f historical and current shoreline position, shoreline change analysis and assessment o f factors influencing the changes observed. This chapter summarizes the approach to the study, the results and draws conclusions. Recommendations are then made both for further research and for policy formulation and management o f the area. 7.2 Data Source and Approach to Study The data used for this study were from different satellite sensor platforms including WorldView-2, Landsat TM and ETM+ and ASTER, with resolution varying from 2m to about 30m. All data, except the high resolution were re-sampled to 15m resolution. A total o f seven shoreline positions were extracted using semi-automatic and manual methods. The extracted shorelines represented the wet/dry boundary for 1986, 1991, 2001, 2007, 2010 and 2011. Both the semi-automatic and the manual delineation proved efficient at extracting this boundary. 90 University of Ghana http://ugspace.ug.edu.gh Uncertainty values for the shoreline positions used vary ±4.1m to ± 8 .8 m. These were used as weights in linear regression analysis. The NSM, EPR and the WLR methods available in the DSAS were used change analysis and rate estimation. NSM results were used for assessing the accuracy o f mapping the shoreline at 15m. The EPR was used to calculate the rate o f change when only two shoreline positions were available while the WLR was used when more than two shoreline positions were available. Rates were calculated along a total o f 283 transects cast at right angles to the constructed baseline. Overall rates o f change for the entire period (25year) were calculated as well as for the periods 1986 to 2001 (period before the KSDP) and 2001 to 2011 (the period after the KSDP). 7.3 Summary o f Results Overall the average rate o f erosion along the coast was estimated at 2m/year. The area east o f the KSDP site, between Keta and Blekusu have experienced high erosion over the entire period. Near the estuary, there is the evidence o f accretion and erosion with the area recording the highest rates. The area immediately south o f the KSDP has seen much accretion. The middle section around the Cape St. Paul changes has not been that abrupt with evidence o f accretion and erosion. Prior to the inception o f KSDP, erosion was dominant on the entire coast with average rate o f 3m/year. Since the completion o f the project there has been evidence o f accretion 91 University of Ghana http://ugspace.ug.edu.gh dominating the shoreline. Change around the Cape St. Paul has not been that abrupt. Results show that the Cape was retreating gradually prior to the Sea Defence. After the construction o f the Sea Defence the entire Cape accreted at an average rate o f 5m/year. However, erosion rates for the entire shoreline remained high (4.5m/year). 7.4 Conclusion Results o f this study have been useful in revealing the trends in shoreline change both erosion and accretion along the coast o f Keta. Although aerial photographs are traditionally the main sources o f data for shoreline monitoring, the study has shown that medium resolution multi-spectral satellite imagery can be used to map and monitor the large and dynamic shoreline o f Keta. Though the study was only for the Keta coast, it is clear the methodology can be applied elsewhere. The first objective o f the study was to identify change in shoreline positions from 1986 to 2011. To achieve this shoreline positions was extracted from the images for 1986, 1991, 2001, and 2011. The extracted shoreline positions were overlaid and visually interpreted to indentify the changes that have occurred. The results indicated significant variations in the positions o f the shoreline over the period. Hence the first objective was achieved. Using the extracted shorelines, the DSAS was using in calculating the shoreline rate o f change. The rates were calculated along transects that were cast along the entire shoreline using weighted linear regression. The findings generally confirmed the high rates 92 University of Ghana http://ugspace.ug.edu.gh reported for this area after the construction o f the Akosombo Dam. Average erosion rates were estimated to be 2 m/year with the sections to the extreme east and west experiencing higher rates. The effect is the destruction o f infrastructure along the coasts as evident at Atorkor and Anyanui. The second objective was hence achieved. Natural factors including wave action, sea level rise and shoreline orientation are major contributors to shoreline change. The comparison o f rates before and after the KSDP as discussed in section 6.2.5 has shown that, the structure is currently playing a major role in the erosion and accretion patterns in the area. Erosion is now taking place down drift (Blekusu and beyond). The shoreline around Keta and Cape St. Paul has been experiencing accretion since the completion o f the KSDP. Other factors such as increasing pressure on the scarce land in the area, sand mining and mangrove harvesting in the area have been blamed for making the coast more vulnerable to erosion. The third objective has thus been achieved. 7.4 Recommendations Based on the results obtained from this study, the area east o f the KSDP (Blekusu) and near the estuary (Atorkor) were identified as hotpots for coastal erosion. The timely intervention for a sea defence structur e at Atorkor is therefore justified and timely. However further east, the Blekusu stretch is also eroding very fast and there is the need for assessment and intervention. 93 University of Ghana http://ugspace.ug.edu.gh Since high resolution images are expensive and lack very large coverage’s, medium resolution such as used for this study can be used for assessing overall changes in shoreline for the entire eastern coast o f Ghana and better still the entire coastline of Ghana to assess current trends and target intervention. There is the need to study sediment budgets in the area to further understand the patterns o f shoreline change in the area especially near the estuary. There is also the need for further research on the tidal conditions as well as the local influence o f sea level rise on erosion in the area. These were outside the scope o f this study. As has been indicated by this study, the implementation o f hard structures such as the KSDP leads to knock off effects in other areas which also then need other forms o f intervention. There is therefore the need for integrated shoreline management for the entire shoreline o f the Country so as to reduce these effects. There is the need for policies by the local and municipal Assembly to completely halt sand mining and mangrove destruction near the estuary. Development along the coastal margin also needs to be properly planned to reduce the impacts o f erosion. 94 University of Ghana http://ugspace.ug.edu.gh REFERENCES Akpati, B.N. (1978). Geologic Structure and Evolution o f the Keta Basin, Ghana West Africa. Geological Society o f America Bulletin, 89, 124-132. Akyeampong E.K. (2001). Between the Sea and the Lagoon: An Eco-Social History o f the Anlo o f Southeastern Ghana, c l 850 to Recent Times. Athens: Ohio University Press. Alesheikh, A.A, Ghorbanali, A. and Nouri, N. (2007). Coastline Change Detection Using Remote Sensing. International Journal o f Environmental Science and Technology, 4 (1), 61-66. Allersma, E. and Tilmans, W.M.K. (1993). Coastal Conditions in W est Africa: A Review. Ocean and Coastal Management, 19, 199-240. Alves, M.V.M. (2007). Detection o f Physical Shoreline Indicators in an Object-based Classification Appraoch: Study case: Island o f Schiermonnikoog, The Netherlands (M aster’s thesis). International Institute for Geo-Information Science and Earth Observation, Enschede, The Netherlands. Amlalo, D.S. (n.d.). The Protection, M anagement and Development o f the Marine and Coastal Environment o f Ghana. Retrieved February 19, 2011 from http://www.fig.net/pub/figpub/pub36/chapters/chapter 10.pdf Anders, F.J. and Byrnes, M.R. (1991). Accuracy o f Shoreline Change Rates as Determined from Maps and Aerial Photographs. Shore and Beach, 16:17-26. Appeaning Addo, K., Walkden, M., and Mills, J.P. (2008). Detection, Measurement and Prediction o f Shoreline Recession in Accra, Ghana. Journal o f Photogrammetry & Remote Sensing, 63, 543-558. University of Ghana http://ugspace.ug.edu.gh Appeaning Addo, K. (2009). Detection o f Coastal Erosion Hotspots in Accra, Ghana. Journal o f Sustainable Development in Africa, 1 1(4), 253-265. Armah, A.K. and Amlalo, D.S. (1998). Coastal Zone Profile o f Ghana. G ulf o f Guinea Large Marine Ecosystem Project. Accra, Ghana: Ministry o f Environment, Science and Technology. Awadzi, T.W., Ahiabor, E. and Breuning-Madsen, H. (2008). The Soil-Land Use System in a Sand Spit Area in the Semi-arid Coastal Savanna region o f Ghana- Development, Sustainability and Threats. West African Journal o f Ecology, 13, 132-143. Ayivor, N. (2011). Atorkor Sea-Defence Wall to be Completed. Ghana News Link. Retrieved June 6 , 2011 from http://www.ghananewslink.com/?id= 14306 Bamhardt, W.A. (2009). Coastal Change along the Shore o f Northeastern South Carolina: The South Carolina Coastal Erosion Study. U.S. Geological Survey Open File Report 2008-1206. Berlanga-Robles, C. A. and Ruiz-Luna, A. (2002). Land Use M apping and Change Detection in the Coastal Zone o f Northwest Mexico Using Remote Sensing Techniques. Journal o f Coastal Research, 18(3), 514-522. Bird, E.C.F. (1985). Coastline Changes: A Global Review. Chichester: John Wiley and Sons Ltd. Bird, E.C.F. (1996). Beach management. Chichester: Wiley and Sons Ltd. Blodget, H.W., Taylor, P.T. and Roark, J.H. (1991). Shoreline Changes along the Rosetta-Nile Promontory: Monitoring with Satellite Observations. Marine Geology, 99, 67-77. 96 University of Ghana http://ugspace.ug.edu.gh Bo, G., Dellepiane, S., Laurentiis, R. (2000). Semiautomatic Coastline Detection in Remotely Sensed Images. Proceedings o f the International Geoscience and Remote Sensing Symposium, 5, 1869-1871. Boak, E.H. and Turner, I. (2005). Shoreline Definition and Detection: A Review. Journal o f Coastal Research, 21(4), 688-703. Boateng, I. (2006). Shoreline Management Planning: Can it benefit Ghana? A Case Study o f UK SMPs and their Potential Relevance in Ghana. Proceedings o f the International Federation o f Surveyors Regional Confence, M arch 8 th- 11th. Accra, Ghana. Retrieved August 11, 2011 from http://www.fig.net/pub/accra/papers/tsl6 / t s l 6 04 boateng.pdf Boateng, I. (2009). Development o f Integrated Shoreline Management Planning: A Case Study o f Keta, Ghana. Proceedings o f the International Federation o f Surveyors Working Week, May 3rd-8th, Eilat, Israel. Retrieved August 11, 2010 from http://www.fig.net/pub/fig2009/papers/ts04e/ts04e boateng 3463.pdf Braun, P. (1962). Sea-Level Rise as a Cause o f Shore Erosion, Journal o f Waterways and Harbours Division, 8 8 (1): 117-130. Camfield, F.E. and Morang, A. (1996). Defining and Interpreting Shoreline Change. Ocean and Coastal Management, 32(3), 129-151. Castellan, A. (2007). Shoreline Management: Alternatives to Hardening the Shore. Ocean and Coastal Resource Management. Retrieved July 4, 2011 from http://coastalmanagement.noaa.gov/shoreline.html Chand, P. and Acharya, P. (2010). Shoreline Change and Sea Level Rise along Coast o f Bhitarkanika Wildlife Sanctuary, Orissa: An Analytical approach o f remote 97 University of Ghana http://ugspace.ug.edu.gh sensing and statistical techniques. International Journal o f Geomatics and Geosciences, 1(3), 436-455. Charlier, R.H. and Charlier, C.C.P. (1995). Sustainable Multiple-Use and Management o f Coastal Zone. Environment management and Health, 6 (1 ), 14-24. Chen, L. C. (1998). Detection o f Shoreline Changes for Tideland Areas Using M ulti­ temporal Satellite Images. International Journal o f Remote Sensing, 19 (17), 3383-3397 Chu, Z.X., Sun, X.G., Zhai, S.K. and Xu, K.H. (2006). Changing Patter o f Accretion/Erosion o f the M odem Yellow River (Huanghe) subaerial delta, China: Based on Remote Sensing Images. International Journal o f Marine Geology, Geochemistry and Geophysics, 227, 13-30. Conterra Limited. (2005). Keta Sea Defnce Project; A Situation Report. The Volta Monitor, 1, 40-42. Cowart, L., Walsh, J.P. and Corbett, D.R. (2010). Analyzing Estuarine Shoreline Change: A Case Study o f Cedar Island, North Carolina. Journal o f Coastal Research, 26(5), 817-830. Cracknell, A.P. (1999). Remote Sensing Techniques in Estuaries and Coastal Zones - An Update. International Journal o f Remote Sensing, 19(3), 485-496. Crowell, M., Leatherman, S. P. and Buckley, M .K. (1991). Historical Shoreline Change: Error Analysis and Mapping Accuracy. Journal o f Coastal Research, 7 (3), 839­ 852. Crowell, M., Leatherman, S.P. and Buckley, M.K. (1993). Erosion Rate Analysis: Long Term Versus Short Term Data. Shore and Beach, 61 (1), 13-20. 98 University of Ghana http://ugspace.ug.edu.gh Cui, B.L. and Li, X.Y. (2010) Coastline Change o f the Yellow River Estuary and its Response to the Sediment and Runoff (1976-2005), Geomorphology, doi: 10.1016/j .geomorph.2 0 1 0 . 1 2 . 0 0 1 Dar, I.A. and Dar, M.A. (2009) Prediction o f Shoreline Recession Using Geospatial Technology: A Case Study o f Chennai Coast, Tamil Nadu, India. Journal o f Coastal Research, 25(6), 1276-1286. Davis, R.A. and Fitzgerald, D.M. (2004). Beaches and Coasts. Blackwell Publishing, Oxford, UK. Dellepiane, S., De Laurentiis, R. and Giordano, F. (2004). Coastline Extraction from SAR Images and a Method for the Evaluation o f Coastline Precision. Pattern Recognition Letters, 25, 1461-1470. Dewidar, K.M. and Frihy, O.E. (2010). Automated Techniques for Quantification o f Beach Change Rates Using Lansat Series Along the North Eastern Nile Delta, Egypt. Journal o f Oceanography and Marine Science, 1(2), 28-39. Dickson K. B. and Benneh G. (1995). A New Geography o f Ghana. 3rd edition. Longmans Book Company. Dolan, A.H. and Walker, I.J. (2004). Understanding Vulnerability o f Coastal Communities to Climate Change related Risks. Journal o f Coastal Research, 39 (Special Issue). Dolan, R., Fenster, M.S., and Holme, S. J. (1991). Temporal Analysis o f Shoreline Recession and Accretion. Journal o f Coastal Research, 7(3), 723-744. Dordor, M.S.K. (2005). The Genesis o f the Keta Sea Defence Project. The Volta Monitor, 1,28-30. 99 University of Ghana http://ugspace.ug.edu.gh El-Asmar, H.M. and W hite, K. (2002). Changes in Coastal Sediment Transport Processes due to Construction o f New Damietta Harbour, Nile Delta, Egypt. Coastal Engineering, 46, 127-138. Fiadzigbey, M. (2005). The Story o f Keta: The Land Factor. The Volta Monitor, 1, 20-22. Flemming, B.W., Delafontaine, M.T. and Liebezeit, G. (Eds.). (2000). Muddy Coast Dynamics and resource Management. Amsterdam, The Netherlands: Elsevier Science. Foster, E. R. and Savage, R.J. (1989). Methods o f Historical Shoreline Analysis. Proceedings o f Coastal Zone '89, New York, USA, 4434-4448. Frihy. O. E., Nasr, S. M., Hattab, M. M. El. and Raey, M. El. (1994). Remote Sensing of Beach Erosion Along the Rosetta Promontary, Northwestern Nile Delta, Egypt. International Journal o f Remote Sensing, 15(8), 1649-1660 Gens, R. (2010). Remote Sensing o f Coastlines: Detection, Extraction and Monitoring. International Journal o f Remote Sensing, 31(7), 1819-1836. Genz, A.S., Fletcher, C.H., Dunn, R.A., Frazer, L.N. and Rooney, J.J. (2007). The Predictive Accuracy o f Shoreline Change Rate Methods and Alongshore Beach Variation on Maui, Hawaii. Journal o f Coastal Research, 23(1), 87-105. Ghanavati, E., Firouzabadi, P.Z., Jangi, A.A. and Khosravi, S. (2008). Monitoring Geomorphologic Changes using Landsat and ETM+ data in the Hendijan River delta, Southwest Iran. International Journal o f Remote Sensing, 29(4), 945-959. GLDD. (2001). Keta Sea Defence: Protecting a Fragile Ecosystem and the Communities it Supports. Retrieved December 10, 2010, from http://www.gldd.com/images/Static/QurProiects 1 2-2.pdf 1 0 0 University of Ghana http://ugspace.ug.edu.gh Go Metal Detecting. (2008). Facts About Ocean Tide. Retrieved June 15, 2011 from http://gometaldetecting.com/ocean-tides.html Ghana Statistical Service. (2005). 2000 Population and Housing Census Data, Accra. Ghana. Hall, S.L., W ilder, W.R., and Fisher, F.M. (1986). An Analysis o f Shoreline Erosion along the Northern Coast o f East Galveston Bay, Texas. Journal o f Coastal Research, 2(2), 173-179. Hapke, C.J., Himmelstoss, E.A., Kratzman, M.G., List, J.H. and Thieler, E.R. (2010). National Assessment o f Shoreline Change: Historical Shoreline Change along the New England and Mid-Atlantic Coasts. U.S. Geological Survey, Open-File Report, 2010-1118. Harbors Division, 8 8 , 1 17-130. Himmelstoss, E.A. (2009). DSAS 4.0 Installation Instructions and User Guide. In Thieler, E.R., Himmelstoss, E.A., Zichichi, J.L. and Ergul, A. (Eds.) Digital Shoreline Analysis System (DSAS) Version 4.0 -An ArcGIS Extension for Calculating Shoreline Change. U.S. Geological Survey Open-File Report, 2008-1278. Karl, T.R., Melillo, J.M. and Peterson, T.C. (Eds.). (2009). Global Climate Change Impacts in the United States. Cambridge University Press. Kawakubo, F.S., Morato, R.G., Nader, R.S. and Luchiari, A. (2011) Mapping Changes in Coastline Geomorphic Features Using Landsat TM and ETM+ Imagery: Examples in Southeastern Brazil. International Journal o f Remote Sensing, 32(9) 2547-2562. Kay, R.C. and Alder, J. (2005). Coastal Planning and Management. London; New York: 10 1 University of Ghana http://ugspace.ug.edu.gh Keta M unicipality (n.d.). Environmental Situation. Retrieved April 5, 2011 from http://keta.ghanadistricts.eov.gh/?arrow=atd& =121&sa=3950 Kofigah, F. (2005). Tourism Development Potential o f Keta District. TheVolta Monitor, 1, 59-62. Korakandy, R. (2005). Coastal Zone Management: A Study o f the political Economy of Sustainable Development, 1, India: Kalpaz Publication. Kufogbe, S.K. (1997). The Natural Resources o f the Southern Ewes. In Agbodeka, F. (Ed.) A Handbook o f Eweland: The Ewes o f Southeastern Ghana, 1(293-319). Accra, Ghana: Woeli Publishing Services. Kuleli, T. (2010). Quantitative Analysis o f Shoreline Changes at the Mediterranean Coast in Turkey. Environmental Monitoring Assessment, 167, 387—397. Laben, C.A. and Brower, B.V. (2000). Process for Enhancing the Spatial Resolution of Multispectral Imagery Using Pan-sharpening. Technical Report, US Patent No. 6,011,875, Eastman Kodak Company. Li, X. and Damen, M.C.J. (2010). Coastline Change Detection with Satellite Remote Sensing for Environmental Management o f the Pearl River Estuary, China. Journal o f Marine Systems, 82, S55-S61. Liu, H. (2009). Shoreline Mapping and Coastal Change Studies Using Remote Sensing Imagery and LIDAR Data. In Yang, X. (Ed.). Remote Sensing and Geospatial Technologies fo r Coastal Ecosystem Assessment and Management, Lecture Notes in Geoinformation and Cartography, Springer, 297-322. 10 2 University of Ghana http://ugspace.ug.edu.gh Liu, H. and Jezek, K.C. (2004). Automated Extraction o f Coastline from Satellite Imagery by Integrating Canny Edge Detection and Locally Adaptive Thresholding Methods. International Journal o f Remote Sensing, 25: 5, 937-958. Lunetta, R.S. (1999). Applications, Project Formulation, and Analytical Approach. In Lunetta, R.S. and Elvdge, C.D. (Eds.). Remote Sensing Change Detection: Environmental Monitoring Methods and Applications (1-20). London: Taylor and Francis. Ly, C.K. (1980). The Role o f the Akosombo Dam on the Volta River in Causing Coastal Erosion in Central and Eastern Ghana (West Africa). Marine Geology, 37, 323­ 332. Maiti, S. and Bhattacharya, A.K. (2009). Shoreline Change Analysis and its Application to Prediction: A Remote Sensing and Statistics Based Approach. Marine Geology, 257, 11-23. Manu, T., Botchway, I.A. and Apaalse, L.A. (2005). Petroleum Exploration Opportunities in the Keta Area. TheVolta Monitor, 1, 56-57. Marfai, M. A., Almohammad, H., Dey, S., Susanto, B. and King, L. (2008). Coastal Dynamic and Shoreline Mapping: Multi-sources Spatial Data Analysis in Semarang Indonesia. Environmental Monitoring Assessment, 142, 297-308. Mensah, J. V. (1997). Causes and Effects o f Coastal Sand Mining in Ghana. Singapore Journal o f Tropical Geography, 18(1), 69-88. Moore, L.J. (2000). Shoreline Mapping Techniques. Journal o f Coastal Research, 16(1), 111-124. 103 University of Ghana http://ugspace.ug.edu.gh Moran, C.A.A. (2003). Spatio-temporal Analysis o f Texas Shoreline Changes Using GIS Technique (M aster’s thesis). Texas A&M University, USA. Morton, R.A. (2004). An Overview o f Coastal Land Loss: With Emphasis on the Southeastern United States. U.S. Geological Survey, Open-File Report, 03-337 NOAA. (2011). Tides are caused by Gravitational Pull o f the Moon and the Sun. National Ocean Service. Retrieved June 15, 2011 from http://oceanservice.noaa.gov/facts/tides.html NOAA. (n.d.). Shoreline. United States Office o f Coast Survey. Retrieved January 03, 2 0 1 1 from http://www.nauticalcharts.noaa.gov/mcd/leanmc shoreline.html Nukunya, (1997). Economic Activities. In Agbodeka, F. (Ed.) A Handbook o f Eweland: The Ewes o f Southeastern Ghana, 1(250-280). Accra, Ghana: Woeli Publishing Services. Oertal, G. F. (2005). Coasts, Coastlines, Shores and Shorelines. In Schwartz, M.L. (Ed.) Encyclopedia o f Coastal Science (323-327), Springer, The Netherlands. Pethick, J. (1984). An Introduction to Coastal Geomorphology, London, Great Britain: Edward Arnold Ltd. Pontee, N.L. and Towsend, I.H. (1999). The Development o f a Framework for Estuary Shoreline Management Plans. Presented at the Institute o f Coastal Engineers Coastal Management Conference, 22-23 September, Bristol, UK. Retrieved 5 July, 2011 from w w w .estuarv-guide.net/pdfs/estuan' snip fram ew ork .pdf Schleupner, C. (2008). Evaluation o f Coastal Squeeze and its Consequences for the Caribbean Island Martinique. Ocean and Coastal Management, 51, 383-390 104 University of Ghana http://ugspace.ug.edu.gh Selvavinayagam, K. (2009). Shoreline Change Monitoring in Tuticorin Coast Using Remote Sensing and GIS Tools. Retrieved January 1, 201 Ifrom http://stesalit-inc.com/coastalengg.pdf Shan, J. and Hussain, E. (2010). Object-Based Data Integration and Classification for High-Resolution Coastal Mapping. In Wang Y. (Ed.). Remote Sensing o f Coastal Environments (209-234). Boca Raton, FL: Taylor and Francis Group. Singh, A. (1989). Digital Change Detection Techniques using Remotely Sensed Data. International Journal o f Remote Sensing, 10(6), 989-1002. Sorensen, T.H., Volund, G., Armah, A.K., Christiansen, C., Jensen, L.B., and Pedersen, J.T. (2003). Temporal and Spatial variations in Concentrations o f Sediment Nutrients and Carbon in the Keta Lagoon, Ghana. West African Journal o f Applied Ecology, 4, 91-105. Svasek Hydraulics, (2006). Measured Wave data, Rotterdam, the Netherlands. Tarbuck, E.J. and Lutgens, F.K. (2010) Earth: An Introduction to Physical Geology. Prentice Hall. UNEP. (2004). Assessment o f Pollution Status and Vulnerability o f Water Supply Aquifers o f Keta, Ghana. Activity Report, Retrieved July 5th 2010 from http://www.unep.org/groundwaterproiect/Archives/Ghana-Bulletin.pdf USGS. (2009). Landsat Thematic Mapper (TM) Level 1 (L I) Data Format Control Book (DFCB). Retrieved March 20, 2011 from http://landsat.usgs.gov/documents/LS DFCB 20 TM Level 1 DFCB.pdf Van, T.T. and Binh, T.T. (2008). Shoreline Change Detection to Serve Sustainable Management o f Coastal Zone in Cu Long Estuary. Proceedings o f the 105 University of Ghana http://ugspace.ug.edu.gh International Symposium on Geoinformatics fo r Spatial Infrastructure Development in Earth and Allied Sciences. Retrieved December 27, 2011 from http://wgrass.media.osaka-cu.ac.ip/gisideas08/viewpaper.php?id^247 Wang, Y. (2010). Remote Sensing o f Coastal Environments: An Overview. In Wang Y. (Ed.). Remote Sensing o f Coastal Environments (1-21). Boca Raton, FL: Taylor and Francis Group. White, K. and El Asmar, H.M. (1999). Monitoring changing position o f coastlines using Thematic M apper imagery, an example from the Nile Delta, Geomorphology, 29 (1-2), 93-105. Wielgus, J., E. Cooper, R. Torres and L. Burke. 2010. Coastal Capital: Dominican Republic. Case Studies on the Economic Value o f Coastal Ecosystems in the Dominican Republic. Working Paper, Washington, DC: World Resources Institute. Retrieved June 20, 2011 from http://www.wri.org/coastal-capital Wong, P.P. (2010). Adaptation Policies and the Coastal Zones o f the Indian Ocean Region: Challenges, Opportunities, and Strategies. In Michel, D. and Pandya, A. (Eds.). Coastal Zones and Climate Change (60-83). Washington: The Henry L. Stimson Center. Woodroffe, C.D. and Leon, J. (2010). The Coastal Zone and Its Classification for Geospatial Analysis. In Maanan, M. and Robin, M. (Eds.). Environmental Science, Engineering and Technology, (1-32). New York: Nova Science Publishers. Yuhendra, J.T.S. and Hiroaki, K. (2011). Performance Analyzing o f High Resolution Pan-Sharpening Techniques: Increasing Image Quality for Classification using 106 University of Ghana http://ugspace.ug.edu.gh Supervised Kernel Support Vector machine. Research Journal o f Information Technology, 3(1), 12-23. Zhang, K., Douglas, B.C. and Leatherman, S.P. (2004). Global Warming and Coastal Erosion. Climate Change, 64, 41-58. Zhang, Y. (2010) Coastal Environmental Monitoring using Remotely Sensed Data and GIS Techniques in the Modem Yellow River Delta, China. Environmental Monitoring Assessment, DOI: 10.1007/s 10661 -010-1716-9 107 University of Ghana http://ugspace.ug.edu.gh APPENDICES Appendix 1: W LR Change Rate Calculation Output for the entire Period (1986 to 2011) T ran sec tld S tartY EndX EndY A zim uth W LR W R2 W SE W CI95 2 664183.93 284167.72 663647.24 126.61 -1.25 0.66 2.11 1.63 3 664023.38 284048.46 663486.69 126.61 -1.37 0.77 1.78 1.38 6 663541.73 283690.66 663005.04 126.61 -1.85 0.94 1.10 0.85 7 663381.19 283571.40 662844.49 126.61 -2.08 0.85 2.04 1.58 8 663220.64 283452.14 662683.95 126.61 -1.58 0.84 1.64 1.27 9 663060.09 283332.87 662523.40 126.61 -1.65 0.74 2.33 1.80 11 662738.99 283094.34 662202.30 126.61 -2.06 1.00 0.32 0.25 12 662578.44 282975.08 662041.75 126.61 -2.54 0.95 1.42 1.10 13 662417.89 282855.81 661881.20 126.61 -2.91 0.85 2.91 2.25 14 662257.34 282736.55 661720.65 126.61 -2.77 0.93 1.81 1.40 17 661775.70 282378.75 661239.01 126.61 -3.72 0.90 3.01 2.33 18 661615.15 282259.49 661078.46 126.61 -3.97 0.90 3.08 2.38 19 661454.60 282140.22 660917.91 126.61 -3.84 0.95 2.04 1.58 20 661294.05 282020.96 660757.36 126.61 -4.75 0.93 3.07 2.37 22 660972.95 281782.43 660436.26 126.61 -5.55 0.87 5.04 3.89 23 660812.41 281663.16 660275.71 126.61 -5.45 0.81 6.20 4.79 24 660651.86 281543.90 660115.16 126.61 -6.83 0.89 5.79 4.47 25 660491.31 281424.63 659954.62 126.61 -8.08 0.91 6.08 4.70 28 660009.66 281066.84 659472.97 126.61 -8.21 0.90 6.38 4.94 29 659849.11 280947.57 659312.42 126.61 -8.02 0.88 6.96 5.38 30 659683.73 280863.89 659187.97 123.42 -9.21 0.82 10.24 7.92 31 659516.81 280753.72 659021.05 123.42 -8.08 0.86 7.74 5.98 33 659182.97 280533.38 658687.21 123.42 -7.33 0.88 6.33 4.89 34 659016.04 280423.21 658520.29 123.42 -1.34 0.10 9.42 7.28 35 658849.12 280313.04 658353.36 123.42 -1.03 0.07 8.92 6.89 36 658682.20 280202.88 658186.44 123.42 -1.86 0.33 6.24 4.82 38 658348.36 279982.54 657852.60 123.42 1.16 0.03 16.35 12.64 39 658181.43 279872.37 657685.68 123.42 -0.91 0.02 16.35 12.64 40 658014.51 279762.20 657518.75 123.42 -1.21 0.04 14.01 10.83 41 657847.59 279652.03 657351.83 123.42 -0.62 0.01 14.27 11.03 44 657337.89 279357.81 656877.27 120.78 2.24 0.09 16.67 12.88 45 657166.07 279255.44 656705.45 120.78 1.91 0.06 17.52 13.54 46 656994.25 279153.08 656533.63 120.78 4.42 0.19 21.65 16.73 47 656822.43 279050.72 656361.81 120.78 3.99 0.18 20.55 15.88 49 656478.79 278846.00 656018.17 120.78 4.03 0.23 17.46 13.50 50 656306.97 278743.64 655846.35 120.78 4.20 0.25 17.25 13.33 108 University of Ghana http://ugspace.ug.edu.gh 51 656135.15 278641.28 655674.53 120.78 4.73 0.26 19.08 14.75 52 655963.33 278538.92 655502.71 120.78 5.54 0.31 19.73 15.25 55 655423.57 278402.50 655241.23 101.69 3.96 0.21 18.14 14.03 56 655227.72 278361.98 655045.38 101.69 2.32 0.19 11.25 8.70 57 655031.87 278321.45 654849.53 101.69 1.57 0.13 9.61 7.43 58 654836.02 278280.93 654653.67 101.69 1.85 0.22 8.24 6.37 60 654444.31 278199.89 654261.97 101.69 6.32 0.86 5.95 4.60 61 654246.14 278208.98 654302.28 86.42 6.58 0.84 6.79 5.25 62 654046.53 278221.45 654102.67 86.42 7.83 0.69 12.55 9.70 63 653846.92 278233.93 653903.06 86.42 9.96 0.77 12.92 9.99 66 653248.09 278271.35 653304.23 86.42 12.54 0.89 10.49 8.11 67 653048.48 278283.83 653104.62 86.42 9.94 0.88 8.90 6.88 68 652852.16 278297.57 653054.53 77.01 7.86 0.86 7.67 5.93 69 652657.28 278342.54 652859.65 77.01 4.53 0.54 9.93 7.68 72 652072.64 278477.46 652275.02 77.01 1.63 0.18 8.23 6.36 73 651877.76 278522.43 652080.14 77.01 1.24 0.10 9.08 7.02 76 651281.40 278481.39 651186.19 96.07 2.82 0.38 8.52 6.58 77 651082.52 278460.23 650987.31 96.07 3.28 0.52 7.51 5.80 78 650883.64 278439.08 650788.44 96.07 3.75 0.62 6.98 5.40 81 650287.01 278375.61 650191.80 96.07 3.19 0.47 8.00 6.19 82 650088.13 278354.45 649992.93 96.07 2.69 0.29 9.95 7.69 83 649889.25 278333.29 649794.05 96.07 3.29 0.31 11.60 8.97 85 649503.14 278201.32 649253.35 106.11 2.33 0.13 14.32 11.07 86 649310.99 278145.81 649061.21 106.11 1.68 0.07 14.96 11.56 87 649118.85 278090.31 648869.06 106.11 1.26 0.04 14.03 10.85 90 648542.42 277923.78 648292.64 106.11 0.79 0.01 15.24 11.78 91 648350.28 277868.27 648100.49 106.11 -0.46 0.01 13.57 10.49 92 648158.14 277812.77 647908.35 106.11 -0.36 0.00 14.24 11.01 94 647783.35 277608.63 647361.51 117.95 -0.62 0.02 11.97 9.25 95 647606.68 277514.89 647184.84 117.95 -0.68 0.02 12.02 9.29 96 647430.01 277421.15 647008.17 117.95 -1.00 0.03 12.56 9.71 97 647253.35 277327.40 646831.50 117.95 -0.17 0.00 10.99 8.49 100 646723.34 277046.18 646301.49 117.95 -0.84 0.03 10.87 8.41 101 646546.67 276952.43 646124.82 117.95 -1.00 0.06 9.50 7.35 102 646370.00 276858.69 645948.15 117.95 -1.20 0.07 10.60 8.20 105 645839.99 276577.46 645418.14 117.95 -0.03 0.00 10.16 7.86 106 645668.28 276440.12 645186.14 122.39 0.69 0.02 12.08 9.34 107 645499.40 276332.97 645017.26 122.39 0.84 0.05 8.56 6.62 110 644992.77 276011.54 644510.62 122.39 1.18 0.11 8.16 6.31 111 644823.89 275904.40 644341.74 122.39 1.06 0.07 9.12 7.05 112 644655.01 275797.25 644172.86 122.39 1.20 0.13 7.23 5.59 109 University of Ghana http://ugspace.ug.edu.gh 113 644486.13 275690.11 644003.98 122.39 1.36 0.11 9.30 7.19 115 644177.31 275327.98 643553.77 133.85 1.03 0.16 5.61 4.34 116 644033.09 275189.41 643409.55 133.85 0.79 0.09 5.88 4.54 117 643888.87 275050.85 643265.33 133.85 1.66 0.22 7.39 5.71 118 643744.65 274912.28 643121.11 133.85 2.47 0.39 7.41 5.73 121 643311.99 274496.59 642688.44 133.85 1.85 0.24 7.86 6.08 122 643167.76 274358.02 642544.22 133.85 2.33 0.25 9.63 7.44 123 643023.54 274219.46 642400.00 133.85 1.78 0.32 6.17 4.77 124 642879.32 274080.89 642255.78 133.85 0.66 0.05 7.16 5.54 127 642446.66 273665.20 641823.12 133.85 1.52 0.20 7.31 5.65 128 642302.44 273526.63 641678.90 133.85 1.69 0.16 9.33 7.21 129 642158.22 273388.07 641534.67 133.85 1.50 0.14 8.71 6.73 130 642037.68 273110.83 641311.03 143.84 1.32 0.14 7.64 5.91 133 641683.67 272626.39 640957.02 143.84 0.50 0.02 8.03 6.21 134 641565.67 272464.91 640839.01 143.84 0.46 0.03 6.50 5.02 135 641447.66 272303.43 640721.01 143.84 0.10 0.00 6.41 4.95 136 641329.66 272141.96 640603.01 143.84 -0.42 0.02 6.96 5.38 140 640864.58 271483.13 640132.22 144.46 -0.12 0.00 5.40 4.17 141 640748.33 271320.39 640015.97 144.46 0.21 0.01 4.74 3.66 142 640659.49 270819.40 639781.34 167.35 0.22 0.03 2.99 2.31 143 640615.68 270624.26 639737.53 167.35 0.81 0.60 1.57 1.22 148 640396.64 269648.55 639518.50 167.35 1.85 0.39 5.52 4.27 149 640352.83 269453.40 639474.69 167.35 1.28 0.24 5.44 4.21 150 640309.03 269258.26 639430.88 167.35 0.68 0.11 4.51 3.49 151 640265.22 269063.12 639387.07 167.35 1.12 0.21 5.24 4.05 152 640221.41 268867.97 639343.27 167.35 1.15 0.18 5.82 4.50 153 640185.17 268607.79 639295.13 171.47 0.52 0.05 5.17 4.00 154 640155.50 268410.00 639265.46 171.47 0.20 0.01 4.95 3.83 155 640125.83 268212.21 639235.79 171.47 0.24 0.01 5.79 4.48 156 640096.17 268014.42 639206.12 171.47 0.25 0.01 5.63 4.35 157 640066.50 267816.64 639176.45 171.47 0.55 0.07 4.65 3.59 158 640036.83 267618.85 639146.79 171.47 0.18 0.02 2.95 2.28 167 639769.82 265838.76 638879.77 171.47 -0.49 0.13 3.03 2.35 168 639740.15 265640.98 638850.11 171.47 -0.68 0.17 3.58 2.77 169 639710.48 265443.19 638820.44 171.47 -0.14 0.01 3.20 2.47 170 639681.60 265194.02 638785.36 174.76 0.06 0.00 4.10 3.17 171 639663.33 264994.86 638767.09 174.76 0.48 0.18 2.40 1.85 172 639645.05 264795.70 638748.82 174.76 0.10 0.01 2.40 1.85 173 639626.78 264596.53 638730.55 174.76 0.17 0.02 3.19 2.47 174 639608.51 264397.37 638712.27 174.76 -0.32 0.07 2.68 2.08 175 639590.24 264198.21 638694.00 174.76 -0.68 0.23 2.94 2.27 1 1 0 University of Ghana http://ugspace.ug.edu.gh 176 639571.97 263999.04 638675.73 174.76 -0.30 0.06 2.94 2.27 177 639553.69 263799.88 638657.46 174.76 -0.93 0.35 3.01 2.33 178 639535.42 263600.72 638639.19 174.76 -1.10 0.53 2.46 1.90 179 639517.15 263401.55 638620.91 174.76 -1.51 0.62 2.79 2.16 187 639370.98 261808.24 638474.74 174.76 -0.95 0.28 3.59 2.78 188 639352.70 261609.08 638456.47 174.76 -0.31 0.08 2.53 1.95 189 639334.43 261409.92 638438.20 174.76 0.11 0.02 1.96 1.51 190 639316.16 261210.75 638419.92 174.76 -0.45 0.04 5.04 3.89 191 639297.89 261011.59 638401.65 174.76 -0.48 0.07 4.31 3.33 192 639279.62 260812.43 638383.38 174.76 -0.47 0.07 4.04 3.12 193 639261.34 260613.26 638365.11 174.76 -1.10 0.28 4.17 3.22 194 639242.25 260422.63 638346.83 174.22 -0.78 0.14 4.60 3.55 195 639222.10 260223.65 638326.68 174.22 -0.51 0.07 4.25 3.29 196 639201.95 260024.66 638306.53 174.22 -0.47 0.08 3.71 2.87 197 639181.80 259825.68 638286.38 174.22 -0.68 0.18 3.44 2.66 206 639000.45 258034.84 638105.03 174.22 -1.12 0.41 3.21 2.48 207 638980.30 257835.86 638084.88 174.22 -1.18 0.26 4.71 3.64 208 638960.15 257636.88 638064.73 174.22 -1.95 0.33 6.64 5.13 209 638940.00 257437.89 638044.58 174.22 -2.85 0.41 8.12 6.28 210 638919.85 257238.91 638024.43 174.22 -2.54 0.44 6.77 5.23 211 638905.00 256957.70 638005.05 179.44 -3.10 0.53 6.88 5.32 212 638903.06 256757.71 638003.10 179.44 -2.52 0.47 6.42 4.96 213 638901.12 256557.72 638001.16 179.44 -1.63 0.20 7.86 6.08 214 638899.18 256357.73 637999.22 179.44 -1.55 0.20 7.38 5.71 215 638897.24 256157.74 637997.28 179.44 -1.35 0.17 7.20 5.56 216 638895.29 255957.74 637995.34 179.44 -1.20 0.13 7.53 5.82 217 638893.35 255757.75 637993.40 179.44 -1.16 0.11 7.73 5.97 218 638891.41 255557.76 637991.45 179.44 -0.89 0.06 8.73 6.75 219 638889.47 255357.77 637989.51 179.44 -0.54 0.02 9.84 7.61 220 638887.53 255157.78 637987.57 179.44 -0.36 0.01 9.77 7.55 232 638864.23 252757.90 637964.27 179.44 0.10 0.00 9.77 7.56 233 638862.85 252508.17 637963.78 182.61 0.05 0.00 9.33 7.22 234 638871.96 252308.38 637972.90 182.61 0.08 0.00 9.99 7.72 235 638881.07 252108.59 637982.01 182.61 0.28 0.00 10.91 8.43 236 638890.19 251908.79 637991.12 182.61 0.01 0.00 13.23 10.23 237 638899.30 251709.00 638000.23 182.61 0.32 0.00 14.77 11.42 238 638908.41 251509.21 638009.35 182.61 0.61 0.01 16.21 12.53 239 638917.52 251309.42 638018.46 182.61 0.63 0.01 18.28 14.13 240 638926.64 251109.63 638027.57 182.61 0.64 0.01 20.22 15.63 241 638935.75 250909.83 638036.68 182.61 -0.08 0.00 23.33 18.03 242 638944.86 250710.04 638045.80 182.61 -0.18 0.00 24.46 18.91 1 1 1 University of Ghana http://ugspace.ug.edu.gh 243 638953.98 250510.25 638054.91 182.61 -0.61 0.00 24.64 19.05 244 638963.09 250310.46 638064.02 182.61 -0.08 0.00 25.98 20.08 245 638972.20 250110.66 638073.13 182.61 0.14 0.00 26.99 20.86 246 638981.31 249910.87 638082.25 182.61 0.06 0.00 28.09 21.72 247 638990.43 249711.08 638091.36 182.61 0.30 0.00 29.65 22.92 248 638999.54 249511.29 638100.47 182.61 -0.37 0.00 31.86 24.63 249 639008.65 249311.49 638109.58 182.61 -1.55 0.01 35.09 27.12 250 639017.76 249111.70 638118.70 182.61 -2.13 0.02 37.37 28.88 251 639026.88 248911.91 638127.81 182.61 -2.17 0.02 38.60 29.84 252 639034.19 248796.62 638135.24 177.23 -1.75 0.01 39.77 30.74 253 639024.52 248596.86 638125.58 177.23 -0.13 0.00 38.93 30.09 254 639014.86 248397.09 638115.91 177.23 0.30 0.00 40.68 31.44 1 1 2 University of Ghana http://ugspace.ug.edu.gh Appendix 2 Field Pictures Kedzi After the KSDP Destruction at Atorkor Discussion with James-Ocloo Remains o f Fort Prinzenstein 113 University of Ghana http://ugspace.ug.edu.gh Appendix 2 Field Pictures Recording Readings Asistant at Field Kedzi After the KSDP Destruction at Atorkor Discussion with James-Ocloo Remains o f Fort Prinzenstein 113 University of Ghana http://ugspace.ug.edu.gh