UNIVERSITY OF GHANA, LEGON CHARACTERIZATION AND SOURCE APPORTIONMENT OF AIRBORNE PARTICULATE MATTER IN SOME URBAN NEIGHBOURHOODS OF ACCRA, GHANA By ALLISON FELIX HUGHES ID#: 10016433 This thesis is submitted to the University of Ghana, Legon, in partial fulfillment of the requirement for the award of Doctor of Philosophy, Physics Degree July 2014 University of Ghana http://ugspace.ug.edu.gh i DECLARATION I, Allison Felix Hughes, declare that except for the references to other peoples work, which have been duly cited, this thesis is the result of my own research and that it has neither in part nor whole been presented for the award of any degree elsewhere. ……………………………………………… CANDIDATE: ALLISON FELIX HUGHES ID#:10016443 DATE:………………………………………... SUPERVISORS: .................................................... ………………………………… Dr. V. C. K. Kakane Prof. J. K. A. Amuzu DATE:………………………... DATE:………………………….. ………………………………………. Prof. E. K. Agyei DATE:………………………………… University of Ghana http://ugspace.ug.edu.gh ii DEDICATION This thesis is dedicated to the Almighty God, my wife – Mrs. Nejay Fraddee Bull Hughes and my children – Margaretta, Sarah, Allison Neal, Alice Nealla and Samuel Nealix. University of Ghana http://ugspace.ug.edu.gh iii ACKNOWLEDGEMENTS First and foremost, I am most grateful to the Almighty God for bringing me this far by ordering my footsteps and for His abundant grace through these years. May His name be glorified forever. In a special way, I wish to express my sincere gratitude to all my supervisors, Dr. V.C.K. Kakane, Prof. J.K.A. Amuzu and Prof. E.K. Agyei, all of the Department of Physics, University of Ghana for their support, guidance, encouragement and for believing in me. May the Lord bless you. My special thanks go to Prof. Majid Ezzati formerly of the Harvard School of Public Health and currently at Imperial College, University of London, UK. Thank you, Prof Ezzati for the opportunity granted me to be part of the project on “Energy, Air Pollution and Health Inequality in the Neighbourhoods of Accra, Ghana” funded by the National Science Foundation (NSF), USA. My thanks go to other members of the team, Kathie Dionisio, Zheng Zhou and Jose Vallarino of the Harvard School of Public Health), Prof. Samuel Agyei-Mensah of the Department of Geography and Resource Development, University of Ghana, Raphael Arku of Harvard School of Public Health (formerly of Department of Geography and Resource Development, University of Ghana) and Audrey Quaye (formerly of the Environmental Science Programme, University of Ghana). I am also grateful to our research assistants, Nana Agyeman-Prempeh and Adam Fatah for field assistance. I would also like to acknowledge the support of University of Ghana for granting me study leave to pursue this doctoral study. My sincere thanks go to the School of Research and Graduate Studies, University of Ghana for granting me the “sandwich” University of Ghana http://ugspace.ug.edu.gh iv programme under the University of Ghana/Harvard University Split-Site PhD programme and to the United States Agency for International Development Mission in Ghana (USAID-Ghana) for sponsorship of a round-trip air ticket to USA and living expenses during my stay at Harvard. I cannot forget the love and care shown by Kathie Dionisio, Mama “D” and the entire Dionisio family. My profound gratitude to the current and former Heads of Department, Dr. A. Kuditcher and Dr. G. Nkrumah-Buandoh, all senior members and staff of the Department of Physics for their love and encouragement. Special thanks to Dr. Hubert A. Koffi for his support and assistance. I am sincerely grateful to the Senior Pastor Rev. Samuel Otu-Pimpong and members of Legon Baptist Church, for their prayer support and encouragement. I would like to thank my family especially my wife, Nejay, for their support and prayers. I am also thankful to my brothers and sister for their support and encouragement. There are many who behind the scenes have encouraged and supported my work, but whose names have not been mentioned. I find it appropriate to acknowledge all of you who made my life easier every day. May the Lord bless you all. Finally, I would like to extend my thanks to the Ghana Meteorological Authority (GMA) for providing the meteorological data. I wish to sincerely acknowledge the NOAA Air Resources Laboratory (ARL) for making available the HYSPLIT transport and dispersion model website used to calculate the backward trajectories in this thesis. Acknowledgement also goes to the US Environmental Protection Agency for provision of the PMF 3.0 receptor modelling software free of charge. University of Ghana http://ugspace.ug.edu.gh v THESIS RELATED PUBLICATIONS Part of the work presented in this thesis appears in the following publications: 1. Zheng Zhou , Kathie L Dionisio, Thiago G Verissimo, Americo S Kerr, Berent Coull, Raphael E Arku, Petros Koutrakis, John D Spengler, Allison F Hughes, Jose Vallarino, Samuel Agyei-Mensah and Majid Ezzati. (2013). Chemical composition and Sources of Particle Pollution in Affluent and Poor Neighbourhoods of Accra, Ghana. Environmental Research Letters, (8) doi:10.1088/1748-9326/8/4/044025 2. Kathie L. Dionisio, Michael S. Rooney, Raphael E. Arku, Ari B. Friedman, Allison F. Hughes, Jose Vallarino, Samuel Agyei-Mensah, John D. Spengler, Majid Ezzati. (2010). Within-Neighborhood Patterns and Sources of Particle Pollution: 
Mobile Monitoring and Geographic Information System Analysis in Four Communities in Accra, Ghana. Environmental Health Perspective 118(5): doi:10.1289/ehp.0901365 3. Kathie L. Dionisio, Raphael E. Arku, Allison F. Hughes, Jose Vallarino, Heather Carmichael, John D. Spengler, Samuel Agyei-Mensah, Majid Ezzati. (2010) Air Pollution in Accra Neighborhoods: Spatial, Socioeconomic, and Temporal Patterns. Environmental Science & Technology 44 (7), 2270-2276 University of Ghana http://ugspace.ug.edu.gh vi ABSTRACT A year-long campaign which accounted for seasonal differences have been conducted to examine the levels, chemical composition and sources of ambient particulate matter in multiple neighbourhoods of varying socio-economic status in Accra, Ghana. Between September 2007 and August 2008, simultaneous measurements of PM2.5 and PM10 aerosols at five monitoring sites in four neighbourhoods (Asylum Down, East Legon, Nima and James Town/Ussher Town) were done for one 48-hour period every six days. Harvard Impactor with polytetrafluorethylene (PTFE) Teflon filter of 37 mm supported by a Whatman drain disc were used to sample air particles in the Accra neighbourhoods. Gravimetric analysis and energy dispersive X-ray fluorescence (EDXRF) were used to determine the chemical composition and concentration of the aerosol particles. The mean mass concentration values for PM2.5 obtained at the five sites during the study period varied from 45.9 µg m-3 to 74.8 µg m-3. The mean mass concentration for PM10 also varied from 93.9 µg m-3 to 134.8 µg m-3. These levels were all substantially higher than the EPA (Ghana) guidelines values and other international air quality standards from WHO, USEPA and EU. Weak relationships were obtained between PM and weather parameters. Crustal elements were most abundant during the seasonal Harmattan period between late December and early February when Saharan dust is transported across West Africa. Enrichment factor analysis was used to provide an initial indication of the species of anthropogenic origin in the measured elemental composition. Source contributions were analysed using positive matrix factorization (PMF) model separately for Harmattan and non-Harmattan periods because large University of Ghana http://ugspace.ug.edu.gh vii changes to source profiles is expected during the Harmattan period. Anthropogenic sources resolved by PMF model (biomass burning, solid waste burning, resuspended dust and traffic/industry emissions) during both Harmattan and non-Harmattan periods had significant influence on the four neighbourhoods in Accra. University of Ghana http://ugspace.ug.edu.gh viii TABLE OF CONTENTS DECLARATION i DEDICATION ii ACKNOWLEDGEMENTS iii THESIS RELATED PUBLICATIONS v ABSTRACT vi TABLE OF CONTENTS viii ABBREVIATIONS xii LIST OF TABLES xv LIST OF FIGURES xvii CHAPTER 1: INTRODUCTION 1 1.1 Urban ambient air pollution 1 1.2 Airborne particulate matter 4 1.3 Sources of particulate matter 6 1.3.1 Natural sources 7 1.3.2 Anthropogenic sources 8 1.4 Size distribution, formation mechanisms, and chemical composition of particulate matter 9 1.4.1 Size distributions of particulate matter 9 1.4.2 Different modes of formation 11 1.4.2.1 Nucleation mode 11 University of Ghana http://ugspace.ug.edu.gh ix 1.4.2.2 Accumulation mode 13 1.4.2.3 Coarse mode 13 1.4.3 Chemical composition 14 1.4.4 Particulate matter in vehicular traffic 17 1.5 Literature review of airborne particulate matter studies in Africa 19 1.6 Overview of atmospheric pollution in Ghana 21 1.7 Source apportionment of particulate matter 25 1.7.1 Receptor modelling methods 26 1.7.2 Positive matrix factorization 28 1.7.3 Enrichment factor 30 1.8 Motivation and objectives of study 32 1.8.1 Objectives of study 34 1.8.2 Specific objectives of the study 34 CHAPTER 2: CLIMATE AND METEOROLOGY OF GHANA 35 2.1 Overview of the climate of Ghana 35 2.2 Local meteorology 38 2.3 Particulate matter and meteorological parameters 41 2.4 Air mass climatology and backward trajectory analysis 44 CHAPTER 3: METHODOLOGY 46 3.1 Description of sampling sites 46 3.1.1 James Town/ Ussher Town (JT) neighbourhood 48 University of Ghana http://ugspace.ug.edu.gh x 3.1.2 Nima (NM) neighbourhood 48 3.1.3 Asylum Down (AD) neighbourhood 49 3.1.4 East Legon (EL) neighbourhood 49 3.1.5 Study design 50 3.2 Description of sampler and filter media used 53 3.2.1 Aerosol sampler 53 3.2.2 Filter 57 3.3 Sampling period 59 3.4 Analytical methods 61 3.4.1 Gravimetric analysis 62 3.4.2 Energy dispersive X-ray fluorescence analysis 65 3.4.2.1 Analysis of elemental composition 66 3.4.2.2 Method detection limit of EDXRF 67 3.5 Traffic count methodology and measurements 67 3.6 Enrichment factor 71 CHAPTER 4: RESULTS AND DISCUSSIONS 73 4.1 PM2.5 and PM10 mass contributions 73 4.1.1 Non-Harmattan conditions, PM2.5 and PM10 fractions 77 4.1.2 Harmattan conditions, PM2.5 and PM10 fractions 80 4.2 PM2.5 and PM10 ratios 83 4.3 Meteorological influences on the concentration of particulate matter 86 4.3.1 Influence of meteorology on PM mass 86 4.3.1.1 Relationship with precipitation 87 University of Ghana http://ugspace.ug.edu.gh xi 4.3.1.2 Relationship with temperature 88 4.3.1.3 Relationship with wind speed 89 4.3.1.4 Relationship with relative humidity 90 4.4 Enrichment factor analysis 91 4.5 Source identification and apportionment using positive matrix factorization 98 4.5.1 Non-Harmattan season 101 4.5.1.1 Fine particulate matter 101 4.5.1.2 Coarse particulate matter 123 4.5.2 Harmattan season 143 CHAPTER 5: CONCLUSIONS AND RECOMMENDATIONS 149 5.1 Conclusions 149 5.2 Recommendations 152 REFERENCES 154 APPENDIX A 180 University of Ghana http://ugspace.ug.edu.gh xii ABBREVIATIONS AD Asylum Down ADT Average daily traffic AGL Above ground level AMA Accra Metropolitan Area AQG Air Quality Guidelines BC Black carbon CA Cluster Analysis CBD Central Business District CMB Chemical mass balance DMS Dimethylsulphide DVLA Driver and Vehicle Licensing Authority EC Elemental carbon ECG Electricity Company of Ghana EDAS Global Data Assimilation System ED-XRF Energy dispersive X-ray fluorescence EF Enrichment factor EL East Legon EPA-Ghana Environmental Protection Authority of Ghana EU European Union FRM Federal Reference and Equivalent Method HI Harvard Impactor HSPH Harvard School of Public Health HYSPLIT Hybrid Single-Particle Integrated Trajectory University of Ghana http://ugspace.ug.edu.gh xiii IMPROVE Interagency Monitoring of Protected Visual Environments INAA Instrumental neutron activation analysis ITCZ Inter-Tropical Convergence Zone JT James Town/Ussher Town NM Nima NOAA National Oceanic and Atmospheric Administration NOX Nitrogen oxides O3 Ozone OC Organic carbon OECD Organization for Economic Co-operation and Development PCA Principal component analysis PIXE Particle-induced X-ray emission PM Particulate matter PM10 Particulate matter with aerodynamic diameter smaller than 10 µm PM2.5 Particulate matter with aerodynamic diameter smaller than 2.5 µm PMF Positive matrix factorization PTFE Polytetrafluoroethylene PUF Polyurethane foam PVC Polyvinyl chloride SD Standard deviation SES Socioeconomic status SO2 Sulphur dioxide SSA Sub-Saharan Africa SSI Size selective inlet University of Ghana http://ugspace.ug.edu.gh xiv UNEP United Nations Environment Programme USAID United States Agency for International Development USEPA United States Environmental Protection Agency VOCs Volatile organic compounds VRA Volta River Authority WHO World Health Organization XRF X-ray fluorescence University of Ghana http://ugspace.ug.edu.gh xv LIST OF TABLES Table 1.1 Major sources of air pollutants 7 Table 1.2 Summary of vehicles registered by DVLA for Accra. 24 Table 2.1 Typical rainfall characteristics of agro-ecological zones in Ghana 41 Table 2.2 Average (or total amount) of meteorological parameters and ranges during the sampling periods in Accra, Ghana. 43 Table 3.1 Characteristics of measurement sites 52 Table 3.2 Example of summary of useful filter properties 58 Table 3.3 Averages of crustal elements used to calculate enrichment factors of PM at all sampling sites in Accra 72 Table 4.1 Statistical summary of average PM10 and PM2.5 in Accra during entire sampling period. 74 Table 4.2 Average concentration of total PM2.5 mass (µg m -3) and its elemental components (ng m-3) at five monitoring sites during the non- Harmattan months. 78 Table 4.3 Average concentration of total PM10 mass (µg m -3) and its elemental components (ng m-3) at five monitoring sites during the non- Harmattan months. 79 Table 4.4 Average concentration of total PM2.5 mass (µg m -3) and its elemental components (ng m-3) at five monitoring sites during the Harmattan months. 81 Table 4.5 Average concentration of total PM10 mass (µg m -3) and its elemental components (ng m-3) at five monitoring sites during the Harmattan months. 82 Table 4.6 PM2.5 and PM10 annual average mass concentration [± SD (standard deviation)] for the five monitoring sites during non-Harmattan conditions. 83 University of Ghana http://ugspace.ug.edu.gh xvi Table 4.7 PM2.5 and PM10 annual mean mass concentration [± SD (standard deviation)] for the five monitoring sites during Harmattan conditions. 84 Table 4.8 Average PM2.5/PM10 ratio of PM elemental components at five monitoring sites during the non-Harmattan period. 85 Table 4.9 Calculated enrichment factor (EFs) referring to Al for all analysed elements in PM2.5. EFs were calculated using the average concentration of all samples from each site (see equation 3.5). Crustal concentrations were taken from Taylor and McLennan (1985). 92 Table 4.10 Calculated enrichment factor (EFs) referring to Fe for all analysed elements in PM10. EFs were calculated using the average concentration of all samples from each site (see equation 3.5). Crustal concentrations were taken from Taylor and McLennan (1985). 96 Table 4.11 S/N values for the PM2.5 data at AD-R, EL-R, JT-R, NM-R and NM-T 99 Table 4.12 S/N values for the PM10 data at AD-R, EL-R, JT-R, NM-R and NM-T 100 Table 4.13 Dataset of number of valid samples and species used in the PMF model 100 Table 4.14 Average contributions of identified sources to PM2.5 concentrations. 105 Table 4.15 Average contributions of identified sources to PM10 concentrations. 128 University of Ghana http://ugspace.ug.edu.gh xvii LIST OF FIGURES Figure 1.1 Idealized schematic of size distribution, deposition and coagulation of particles in ambient air. (Adapted from Whitby and Sverdrup, 1980) 12 Figure 1.2 (a) Cooking along the side of street for small-scale commercial activity (b) vehicular traffic congestion (c) vehicular exhaust emissions (d) wind-blown/harmattan dust in Accra. 22 Figure 2.1 Map of Ghana showing administrative regions and borders. 36 Figure 2.2 Movements of the Inter-Tropical Convergence Zones 37 Figure 2.3 Map of Ghana showing the various agro-ecological zones 40 Figure 2.4 Monthly average of atmospheric temperature (°C), relative humidity (RH, in %) and rainfall (mm) for the period September 2007 – August 2008 in Accra. 43 Figure 3.1 Four study neighbourhoods lying on a line from the coast to the northern boundaries of the Accra Metropolitan Area (AMA). AD, EL, JT and NM represent Asylum Down, East Legon, James Town/Ussher Town and Nima neighbours respectively. 47 Figure 3.2 Schematic diagram of Harvard Impactor particle sampler 55 Figure 3.3 A typical set-up of the ambient monitoring system. Residential site (left) and traffic site (right) with integrated ambient 48-hour Harvard Impactor (below) mounted on roofs of buildings to measure PM10 and PM2.5 concentrations. 56 Figure 3.4 Diagram of 48-hr averaging sampling schedule at four Accra neighbourhoods during study period. 60 Figure 3.5 Classification of vehicles counted 69 Figure 4.1 Time series plots of (a) PM2.5 and (b) PM10 mass concentrations during study period. 76 Figure 4.2 Relationship between PM2.5 and total daily cumulative precipitation. 87 University of Ghana http://ugspace.ug.edu.gh xviii Figure 4.3 Relationship between PM2.5 and total daily average temperature. 88 Figure 4.4 Relationship between PM2.5 mass concentrations and daily wind speed. 89 Figure 4.5 Relationship between PM2.5 mass concentrations and relative humidity. 90 Figure 4.6 Enrichment factors for PM2.5 chemical components for all sites in Accra. 94 Figure 4.7 Enrichment factors for PM10 chemical components for all sites in Accra. 97 Figure 4.8 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at AD-R. 102 Figure 4.9 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at EL-R. 103 Figure 4.10 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at JT-R. 103 Figure 4.11 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at NM-R. 104 Figure 4.12 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at NM-T. 104 Figure 4.13 PM2.5 source profiles for sea salt factors. 106 Figure 4.14 Time series of PM2.5 source contributions for sea salt factors. 107 Figure 4.15 Five-day HYSPLIT backward trajectories of air masses reaching Accra on 17 August 2008. 109 Figure 4.16 PM2.5 source profiles for biomass factors. 111 Figure 4.17 Time series of PM2.5 source contributions for biomass burning factors. 112 Figure 4.18 PM2.5 source profiles for solid waste burning factors. 114 Figure 4.19 Time series of PM2.5 source contributions for solid waste burning factors. 115 Figure 4.20 PM2.5 source profiles for soil dust factors. 117 University of Ghana http://ugspace.ug.edu.gh xix Figure 4.21 Time series of PM2.5 source contributions for soil dust factors. 118 Figure 4.22 PM2.5 source profiles for resuspended dust factors. 119 Figure 4.23 Time series of PM2.5 source contributions for resuspended dust factors. 120 Figure 4.24 PM2.5 source profiles for traffic/industry factors. 122 Figure 4.25 Time series of PM2.5 source contributions for traffic/industry dust factors. 123 Figure 4.26 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at AD-R. 125 Figure 4.27 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at EL-R. 126 Figure 4.28 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at JT-R. 126 Figure 4.29 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at NM-R. 127 Figure 4.30 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at NM-T. 128 Figure 4.31 PM10 source profiles for sea salt factors. 129 Figure 4.32 Time series of PM10 source contributions for sea salt factors. 130 Figure 4.33 PM10 source profiles for biomass burning factors. 132 Figure 4.34 Time series of PM10 source contributions for biomass burning factors. 133 Figure 4.35 PM10 source profiles for solid waste burning factors. 134 Figure 4.36 Time series of PM10 source contributions for solid waste burning factors. 135 Figure 4.37 PM10 source profiles for soil dust factors. 137 Figure 4.38 Time series of PM10 source contributions for soil dust factors. 138 University of Ghana http://ugspace.ug.edu.gh xx Figure 4.39 NOAA HYSPLIT model of air mass trajectories over Accra (location shown in star★) on (a) 9th December 2007, (b) 20th March 2008 and (c) 13th April 2008. 139 Figure 4.40 PM10 source profiles for resuspended dust factors. 141 Figure 4.41 Time series of PM10 source contributions for resuspended dust factors. 141 Figure 4.42 PM10 source profiles for traffic/industry factors. 142 Figure 4.43 Time series of PM10 source contributions for traffic/industry factors. 142 Figure 4.44 PM2.5 source profiles for peak Harmattan months (25th December 2007 to 7th February 2008). 145 Figure 4.45 Time series of PM2.5 source contributions for peak Harmattan months (25th December 2007 to 7th February 2008). 146 Figure 4.46 PM10 source profiles for peak Harmattan months (25th December 2007 to 7th February 2008). 147 Figure 4.47 Time series of PM10 source contributions for peak Harmattan months (25th December 2007 to 7th February 2008). 148 University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER 1: INTRODUCTION 1.1 URBAN AMBIENT AIR POLLUTION Air pollution has become one of the most serious environmental concerns in urban cities throughout the world – especially in developing countries (Mayer, 1999; Faiz and Sturm, 2000). The United Nations (UN) estimate that more than half of the world’s 7 billion human population lives in urban areas and occupies just about 3% of the earth’s land mass (UNEP, 2012). By the year 2030, the world’s urban population is expected to increase to almost 65% and a large amount of the new urban dwellers will be in slums and squatter settlements with poor air quality in low-income countries. Sub- Saharan Africa (SSA) is one of the poorest regions in the world. With an urban population growth rate of more than 2% annually, SSA has the fastest urbanization rate in the world (UNEP, 2012). According to an extensive report by UN Habitat (2010), the urban population of African cities is projected to triple by 2050 to about 1.23 billion people. Many of the large SSA cities have experienced rapid urbanization, industrialization and motorization over the last decade. The shape of cities in these developing countries, and how it will affect the well-being of future urban residents and the global environment, will depend greatly on decisions made now in preparation for this growth (UNFPA, 2007). The low income levels and lack of economic resources to deal with these problems have led to these SSA cities being included among the most polluted cities in the world. University of Ghana http://ugspace.ug.edu.gh 2 Air pollution can be defined as the introduction of particulate matter, chemicals, or biological materials into the atmosphere in such concentration and duration which may adversely affect human health, the environment, materials, and have economic impact. Urban ambient air pollution is a more specific term which refers to the outdoor air pollution experienced by populations living in urban areas, typically in or around cities. Air pollution at the local urban level has negative health effects on the exposed population. The World Health Organization (WHO) reports that urban air pollution is responsible for 2.0% of premature deaths (about 1.2 million people) every year from lung cancer, cardiovascular and respiratory diseases caused by outdoor air pollution (WHO, 2009). The Organization for Economic Co-operation and Development (OECD) in a recent study on the global environmental outlook warned that air pollution from particulate matter could become the biggest environmental cause of premature deaths by 2050 if no action is taken to improve air quality. In the OECD report, the number of premature deaths linked to particulate matter is projected to reach 3.6 million people per year globally, with most of the deaths occurring in developing countries (OECD, 2012). In Africa, urban ambient air pollution is currently responsible for an estimated 49,000 premature deaths annually with indoor use of solid fuels being responsible for eight times this value, the main burden being borne by sub-Saharan African countries (World Bank, 2012). Direct causes associated with air pollution-related deaths include aggravated asthma, bronchitis, emphysema, lung and heart diseases, and respiratory allergies (Breslow and Goldsmith, 1958; Krzyzanowski et al, 2005; HEI, 2010). University of Ghana http://ugspace.ug.edu.gh 3 Sources of air pollution are numerous and highly variable. The sources of air pollution in African urban cities differ from those in many other regions of the world (Cohen et al., 2002; Wang et al., 2003; Arku et al., 2008). However, large gaps exist in understanding particulate matter (PM) sources and their relative contributions in SSA cities. Routine monitoring of PM10 or PM2.5 anywhere in SSA other than South Africa is scarce. Aerosol monitoring has become a special concern, especially in SSA urban cities. The lack of ambient monitoring data for particulate matter in SSA cities has seriously hampered the ability to characterize and understand the patterns of PM concentrations, and to promote policy initiatives to address air quality. Some SSA countries have begun measurements on size specific fractions of particulate matter (i.e., PM10 and PM2.5) and their chemical characteristics (Nerquaye-Tetteh, 2006; van Vliet and Kinney, 2007; Mkoma et al., 2009). Understanding the composition of atmospheric aerosol particles is necessary for identifying their sources and predicting their effect on various atmospheric processes and health effects. For this reason, the process of identification and apportionment of particulate matter to their sources using the powerful multivariate receptor model – positive matrix factorization (PMF) is a crucial step in air quality management. In some African cities where monitoring has been carried out, levels of air pollution have been found to exceed the WHO recommended air quality guidelines standards (WHO, 2012). Urban air pollution in SSA results mainly from biomass combustion for cooking and heating (Akbar and Kojima, 2002; Smith et al., 2004; Bailis et al., 2005; University of Ghana http://ugspace.ug.edu.gh 4 Barnes et al., 2005) and the use of fossil fuels in transport, power generation, industry and domestic sectors. Other sources include open burning of refuse, re-suspended dust from unpaved roads, construction particles and natural windblown particles (Akbar and Kojima, 2002). 1.2 AIRBORNE PARTICULATE MATTER Aerosol or particulate matter (PM) is defined as suspension of liquid and solid particles in a gaseous phase – in this case, the atmosphere. Technically, the term aerosol refers to both the particles and the gas in which it is suspended. However, it is common practice in literature to use aerosol to refer only to the particles or particulate component inside the atmosphere (Vallero, 2008; Place, 2009, Kulkarni et al., 2011; Friend, 2012). It is the particulate component which is of the greatest interest in this study due to its impact on human health and the environment. The behaviour of particulate matter in the urban environment and its potential to affect human health and atmospheric visibility depends on the following characteristics:  physical (size and shape),  chemical (composition and concentration),  geographic location, season, day and time of day of emissions. Moreover, these physical and chemical characteristics also depend on the source type and on the formation processes that particles undergo within the atmosphere or at the emission source (CEPA, 1998). Particulate matter is not a single pollutant but is a complex mixture of different chemical species including elemental and organic carbon University of Ghana http://ugspace.ug.edu.gh 5 compounds; oxides of silicon, aluminium and iron; trace metals; sulphates; nitrates and ammonium. Particulate matter may be produced naturally or as a direct or indirect result of human activities (Godish, 1997) and they include a wide range of particles of different sources, sizes and compositions. They can be classified as primary or secondary depending on their formation mechanism. Particles that are directly emitted into the atmosphere from sources are referred to as primary particles while particles that are formed from atmospheric processes (i.e., photochemistry) or those which undergo changes in chemical composition following emission are called secondary particles (Wilson and Spengler, 1996). Primary particles can also be grouped as anthropogenic or natural depending on their origin. The origin of secondary particles is more difficult to link to their original sources because particulate matter is not a single pollutant but is a complex mixture of different chemical species including elemental and organic carbon compounds; oxides of silicon, aluminium and iron; trace metals; sulphates; nitrates and ammonium. However, secondary particles for example can result from the chemical transformations of primary emissions such as nitrogen oxides (NOX), sulphur dioxide (SO2), ozone (O3), and organic gases from natural and anthropogenic sources. Furthermore, both natural and anthropogenic particles can occur from either primary or secondary processes (Mathys et. al., 2002). The concentration of primary particles is a function of their emission rate, transport and dispersion and removal rate from the atmosphere. University of Ghana http://ugspace.ug.edu.gh 6 Primary particles are mostly present in the coarse size fraction while secondary particles are mostly in the fine fraction (Raes et al., 2000). The coarse fraction includes particles greater than 2.5 μm which are usually released directly into the atmosphere by mechanical processes or from natural sources. The fine fraction contains particles with a diameter less than 2.5 μm, which mainly arises from gas-to-particle conversion – a process whereby gases such as sulphur dioxide (SO2), nitric oxide and nitrogen dioxide (NOX), and volatile organic compounds (VOCs) are oxidized and condensed. 1.3 SOURCES OF PARTICULATE MATTER IN URBAN AIR Urban ambient particulate matter originates from a large variety of sources. These sources can be classified as either natural (produced in the absence of human activity) or anthropogenic (caused by human activity) (Miranda and Tomaz, 2008). Natural aerosols sources are uncontrollable but anthropogenic aerosols sources are controllable although they poses a challenge. Human activities and industrial development influence both anthropogenic and natural emissions. Hence when the effects of air pollutants on health is under consideration, especially in urban areas with high population densities, anthropogenic sources are very important and are those to which attention is usually directed with a view to control. Major sources of air pollution are summarized in Table 1.1. University of Ghana http://ugspace.ug.edu.gh 7 Table 1.1 Major sources of air pollutants Source: Adapted from (Schwela, 2010) 1.3.1 NATURAL SOURCES Natural sources of particulate matter include windblown dust, harmattan dust, natural vegetation (plant fragments, microorganism, pollen, etc.), wildfires, and sea spray. Crustal material, biogenic matter and sea-salt comprise the majority of natural aerosols. The major source of crustal elements (e.g. Si, Ca, Fe, Al, Mg, K, Na, Ti and O) is the wind-blown dust suspended from construction sites, roads and natural surfaces. Primary biogenic aerosol particles consist of several different types of particles, including pollen, spores, and micro-organisms such as virus, bacteria, fungus, protozoa, algae and Source type Source Causal action Natural Volcano eruption Release of solid particles, gases and heat waves. Sand storm Dust particles spread through wind circulation around the earth. Vegetation fire Plant Pollen Smoke from wildfire or forest management. Spread of plant pollen through wind motion. Sea spray Liquid droplets spread through wind near coastlines. Anthropogenic (Man-made) Transport Combustion of petrol/diesel and generation of particles and gases. Power generation Industry Release of particles and gases. Manufacturing and processing of steel, non-iron metals, textiles, refining of petroleum, handling of materials. Construction Particle pollution due to material handling and other associated activities. Agriculture Domestic Emissions from ploughing and use of fertilizers, pesticides and insecticides. Burning of solid organic materials indoors such as charcoal, firewood, dung, agricultural residues. University of Ghana http://ugspace.ug.edu.gh 8 fragments of insects. Virus and bacteria present sizes greater than whereas the remains of plants and fungal spores exhibit sizes in the coarse mode of particulate matter both in urban and suburban environments (Posfai and Molnar, 2000; Bauer et al., 2008). 1.3.2 ANTHROPOGENIC SOURCES Anthropogenic particulate sources are primarily found in urban and industrial locations. In these locations, anthropogenic emissions can become so large that the concentrations of various undesirable chemical species (pollutants) cause significant deterioration in air quality and visibility (Watson, 2002; Cohen et al., 2004), and can pose threats to human health (Dockery and Pope, 1996; Pope et al., 2002; Villeneuve et al., 2003). Anthropogenic aerosols are composed of primary emitted soot (elemental carbon), heavy metals, secondary formed carbonaceous material (organic carbon) and inorganic matter (nitrates, sulphates, ammonium and water). Vehicular emission, industrial emission as well as combustion of fuel for domestic heating and cooking purposes are considered as primary anthropogenic sources for urban air pollution (Kothai et al., 2008). Anthropogenic sources of particulate matter and its precursor gases include vehicular emissions, industrial emissions, cooking, heating, agriculture, wood burning, power plants, heavy duty diesel engine, construction and demolition, and dust from disturbed land. They arise mostly from the combustion of fossil fuels, tire and brake wear, and re-suspended dust (Ricci et al., 1996; US EPA, 1996; Buzorious et al., 1999). University of Ghana http://ugspace.ug.edu.gh 9 Heavy metals (e.g. V, Cr, Mn, Co, Ni, Cu, Zn, As, Cd, Ba and Pb) originate from a variety of industrial processes such as incineration, manufacturing, and smelting. 1.4 SIZE DISTRIBUTIONS, FORMATION MECHANISMS, AND CHEMICAL COMPOSITION OF PARTICULATE MATTER 1.4.1 SIZE DISTRIBUTIONS OF PARTICULATE MATTER The size distribution of particulate matter is crucial in understanding the transport and removal of particles in the atmosphere as well as their deposition in the human respiratory system. Once the particles are emitted into the atmosphere, they are subjected to changes in their size and chemical composition by processes of condensation of vapour species, evaporation, deposition (both wet and dry) and coagulation (Finlayson-Pitts and Pitts, 1986; Seinfeld and Pandis, 1998). The chemical compositions of these particles are determinants for the types of effects caused by particulate matter on humans, vegetation, and materials. Particle size is one of the most important parameter in characterizing the physical behaviour of particulate matter. The size of a particle affects many of its properties such as volume, mass and settling velocity (CCPA, 2001). Size also influences a particle’s transport, deposition and migration through the environment (Aboh, 2009). According to Mkoma (2008) ambient aerosol particles sizes can vary from 1 nm to 100 μm in diameter. Particulate matter in the size range from 0.1 to 10 μm is noted to have serious health, climatic and environmental effects (Finlayson-Pitts and Pitts, 2000). University of Ghana http://ugspace.ug.edu.gh 10 Particle size is normally defined in terms of its diameter. Due to variations in its shape and density, the particle size is expressed in terms of an “equivalent” diameter, which depends on a physical rather than a geometric property. The aerodynamic diameter is the most commonly used equivalent diameter of a particle (Godish, 2004). It is defined as the diameter of a sphere of unit density (1 g cm-3) that has the same settling velocity in air as the particle of interest (Hinds, 1982). The particle aerodynamic diameter, Dp, is given by Equation 1.1 as √ where Dg is the particle geometric diameter, ρp is the density of the particle, ρ0 is the reference density (1 g cm-3), and k is a shape factor, which is 1.0 in the case of a sphere (Finlayson-Pitts and Pitts, 2000). In this work, particle diameter unless otherwise indicated, refers to the aerodynamic diameter. For regulatory control and human health effects, atmospheric particulate matter is commonly classified as ultrafine (PM0.1: ), fine or respirable (PM2.5: ) and coarse or inhalable (PM10: ) particle size fractions. Respirable particles are mostly deposited in the upper part of the respiratory tract or nasal system. Inhalable particles (also referred to as thoracic particles) are mostly deposited in the pulmonary or lung system and persist in the body. The distinction between the coarse and fine particles is vital because they have different sources, formation mechanisms, composition, atmospheric lifetime, spatial distribution, temporal University of Ghana http://ugspace.ug.edu.gh 11 variability, and health impacts (Johnson et al., 2011). The objective for regulatory control of particulate matter is to reduce the fine and coarse particles mass concentrations rather than number concentrations. 1.4.2 DIFFERENT MODES OF FORMATION In the atmosphere, the particles size distributions are not uniform but particles typically show a trimodal size log-normal distribution when plotted as mass or volume versus aerodynamic diameter (Figure 1.1). These modes reflect the different source types and the different transformation processes at work in the atmosphere (Seinfeld and Pandis, 1998). The three distinct modes or fractions are generally categorized as follows: a) Nucleation mode: Dp < 0.1 μm, b) Accumulation mode: 0.1 μm < Dp < 2 μm, and c) Coarse or sedimentation mode: Dp > 2 μm. 1.4.2.1 NUCLEATION MODE Nucleation mode (also known as “nuclei mode” or “Aitken mode”) particles are formed from condensation of hot vapours during combustion processes and from gas-to-particle conversion in the atmosphere. Nuclei mode particles consist of freshly formed particles with diameters below about 10 nm resulting from active nucleation events. The Aitken mode particles are the larger aged particles with diameters between about 10 and 100 nm which may result from growth of smaller particles undergoing coagulation or nucleation from higher concentrations of precursors (USEPA, 2004). University of Ghana http://ugspace.ug.edu.gh 12 Figure 1.1 Idealized schematic of size distribution, deposition and coagulation of particles in ambient air. (Adapted from Whitby and Sverdrup, 1980) The nucleation or Aitken mode contains the largest number of particles but due to their small sizes, they account for not much of the total mass. Nucleation mode particles in the atmosphere have very short life of the order of a few minutes to several hours; this is due to their rapid coagulation with larger particles or growth into larger sizes by condensation. They are removed through diffusion of falling rain drops. Ultrafine particles include the nucleation mode and much of the Aitken mode. University of Ghana http://ugspace.ug.edu.gh 13 1.4.2.2 ACCUMULATION MODE Accumulation mode particles are generally formed by coagulation of nuclei mode particles and condensation of gas-phase vapours onto existing particles, causing them to grow into this size range (Seinfeld and Pandis, 1998). Particles in accumulation size range can also be emitted directly into the atmosphere as a result of incomplete combustion of wood, oil, coal, gasoline and other fuels. These particles typically contain significant amounts of organic material as well as soluble inorganic such as ammonium, nitrate and sulphate (Alfarra, 2004). They have the largest surface area and therefore account for a significant fraction of the total particle mass. As the mode name suggests, the particle removal mechanisms are least efficient in this size range. Hence, they can be transported over thousands of kilometre and remain airborne for days to weeks. They also account for most of the visibility impairment in the dry atmosphere. Accumulation mode particles can be removed from the atmosphere by rainout or washout (Sillanpää, 2006). Collectively, the accumulation and nucleation mode particles are often referred to as fine particles. 1.4.2.3 COARSE MODE Coarse or sedimentation mode particles originate mainly from mechanical processes such as grinding, wind or erosion. Particles in the coarse range are relatively large and usually consist of mostly anthropogenic and natural dust particles that normally settle out of the atmosphere relatively quickly by sedimentation and impaction on surfaces (Seinfeld and Pandis, 1998). They can also be removed by falling rain drops. University of Ghana http://ugspace.ug.edu.gh 14 Moreover, the chemical composition of coarse particles is usually dominated by inorganic material such as sand or large salt particles from sea spray. Because coagulation is very weak for particles in the accumulation mode, little of the mass in the mode is transferred into coarse mode. Together with the accumulation mode, they make up most of the total mass of suspended particles. Their residence time in the atmosphere is only a few hours or days depending on their size, prevailing meteorological conditions, and altitude (Aboh, 2009). 1.4.3 CHEMICAL COMPOSITION The chemical composition of fine and coarse particulate matter differ substantially from each other, indicating that these particles originate from different sources (Harrison and Yin, 2000). The location of the polluting source, chemical reactions in the atmosphere, time of the day, and meteorological conditions results in significant variations in the chemical composition of these particles (Finlayson-Pitts and Pitts, 1986). The composition of particulate matter is also influenced by other anthropogenic factors such as the degree of urbanization, industrial and agricultural activities, and the type and volume of vehicular traffic. Chemical composition of particles is generally composed of a mixture of species from a number of sources. These particles may be made up of variable amounts of sulphate, ammonium, nitrate, sea salt (sodium, chloride), trace metals, crustal elements and carbonaceous materials (Chow, 1995; Vallius, 2005). Information on the chemical composition aerosol of particles is essential for their health and environmental assessment. Characterization of the chemical compositions of PM is University of Ghana http://ugspace.ug.edu.gh 15 vital for understanding aerosol physical and chemical processes as well as identifying possible sources of ambient particles and their contributions. Much of the sulphate component in the atmosphere is derived mainly from the oxidation of anthropogenic and natural sulphur-containing compounds such as sulphur dioxide (SO2) and dimethylsulphide (DMS), respectively (Alfarra, 2004). Most of the sulphate is found in the fine fraction of PM as ammonium sulphate although some are present as sodium sulphate of marine origin. SO2 emissions originate from anthropogenic as well as natural sources. Anthropogenic SO2 emissions emanate from the combustion of fossil fuels, biomass combustion, agricultural waste burning, metal smelting, and other industrial processes (Benkowitz et al., 1996; Smith et al., 2011). Natural sources of SO2 emissions include oxidation of dimethylsulphide (DMS) mainly from the sea in the form of sea salt and from wild fires (Bates et al., 1992). In 2005 it was estimated that typical annual average concentrations of sulphur dioxide in urban areas in developing countries were 40-80 μg m-3, those in North America and Europe were 10-30 μg m-3, and in cities in the European Union (EU) 6-35 μg m-3 (WHO, 2005). Nitrate (NO3) is a major constituent in atmospheric aerosols formed primarily by oxidation of atmospheric nitrogen oxides (NOX). Nitrate can be found in both the fine and coarse particulate matter. Nitrate in fine particles comes mainly from the reaction of gas-phase nitric acid with gas-phase ammonia to for ammonium nitrate. Nitrate in coarse particles comes primarily from the reaction of gas-phase nitric acid with pre- existing coarse particles. The relative humidity and temperature of the atmosphere also University of Ghana http://ugspace.ug.edu.gh 16 have strong influence on the amount of nitrate (Stelson and Seinfeld, 1982). Chloride is mainly emitted as sodium chloride (NaCl) in sea salt particles. Chlorides also enter aerosol particles due to neutralization of hydrochloric acid (HCl) vapour by ammonia. Carbonaceous compounds in PM consist of two components which are elemental carbon and organic carbon. Elemental carbon (EC) is frequently referred to as black carbon (BC) or soot. It is chemically similar to impure graphite and mainly emitted into the atmosphere from combustion processes such as fossil fuel and biomass burning. EC is commonly used as a tracer for the exhaust of diesel-powered vehicles. Organic carbon (OC) is the dominant fraction of the carbonaceous aerosol. It originates from the combustion of fossil fuels, residential wood burning, meat cooking and biogenic sources (Pandis et al., 1992; Jeong et al., 2004). Crustal material such as aluminium (Al), silicon (Si), calcium (Ca), titanium (Ti), magnesium (Mg) and iron (Fe) are found predominately in coarse particles and their composition vary according to local geology in addition to surface conditions. Potassium (K) may be found in both fine and coarse particles. Potassium in coarse PM comes from soil and those in fine PM comes from biomass burning. Trace elements such as sodium (Na), potassium (K), iron (Fe), chromium (Cr), cobalt (Co), nickel (Ni), manganese (Mn), cooper (Cu), selenium (Se), barium (Ba), chlorine (Cl), gallium (Ga), caesium (Cs), europium (Eu), tungsten (W), and gold (Au) are found in both fine and coarse particles. Biomass burning which includes residential wood combustion and forest fires, releases trace elements into the atmosphere. Trace metals in fine PM are University of Ghana http://ugspace.ug.edu.gh 17 primarily anthropogenic originating from combustion of fossil fuels, waste incineration and metal smelting activities. Many of these components can be used as tracers for specific sources. For example, sodium (Na) is a tracer that is almost exclusively associated with sea salt. Likewise, silicon and aluminium are tracers for mineral dust, whereas elemental carbon particles are mainly derived from combustion. Finally, the secondary components (sulphate, nitrate and ammonium) can be attributed to their gaseous precursors. 1.4.4 PARTICULATE MATTER IN VEHICULAR TRAFFIC Transportation is the single largest source of air pollution in urban areas (Agyemang- Bonsu et al., 2007). The United Nations Environment Programme (UNEP) estimates that about 90% of urban air pollution in rapidly growing cities in developing countries is as a result of vehicle emissions (UNEP, 2010). The rapidly growing numbers of motor vehicles in urban cities of developing countries are posing serious risk to the health of the population. Traffic-related particles from vehicles fall mainly into the submicron or fine mode range. They are able to penetrate deep into the respiratory tract, especially into the alveolar regions of the lung. In most sub-Saharan African urban cities, the vehicle emissions problem tend to be dominated by emissions from the high number of old and poorly maintained motor vehicles without catalytic converters, and the use of poor quality fuel (Kojima and Lovei, 2001). Moreover, the increasing numbers of second-hand cars and poor road networks have led to traffic congestion in most African cities. Almost 80% of the vehicles imported into Ghana are second-hand University of Ghana http://ugspace.ug.edu.gh 18 vehicles (Chalfin, 2008). Traffic congestion can cause vehicular emissions to increase appreciably, which can lead to high human exposures to traffic-related pollutants (Vardoulakis et al., 2003; Nesamani et al., 2007). Traffic congestion in urban areas hampers economic productivity, damages people’s health, and degrades the quality of their lives. Exhaust emissions from both diesel and gasoline fuelled vehicles have been identified as a major source of primary and secondary anthropogenic aerosols in urban cities. Road-traffic emissions in urban areas come from a number of sources, which include vehicle exhaust or tailpipe emissions (both fuel and lubricating oil combustion) and contributions from non-exhaust vehicle-related particles emissions (Rogge et al., 1993; Kupiainen, 2007). A study by Kam et al. (2012) in developed countries shows that exhaust emissions contribute predominantly to fine PM and non-exhaust emissions contribute mainly to the coarse PM. Vehicle exhaust emissions consist of particles formed in the internal combustion engines as product of incomplete combustion. The particles derived from vehicle exhaust emission are primarily composed of elemental carbon (EC) and organic carbon (OC). Exhaust or tailpipe emissions depend on the type and age of the engine, the type of fuel and lubricant used, and the configuration of the exhaust system (Ondráček et al., 2011). Heavy metals that have often been associated with vehicular emissions include Cu, Zn, Pb, Br, Fe, Ca, Cr, and Ba (Wahlin et al., 2006; Thorpe and Harrison, 2008). University of Ghana http://ugspace.ug.edu.gh 19 Non-exhaust emissions from on-road vehicles consist of non-combustion particles from tires and brake wear, road surface abrasion and re-suspension of road dust induced by the vehicle-generated turbulence (Thorpe and Harrison, 2008). Abrasion of brake linings and tires and re-suspended road dust, for example, release to the atmosphere particles with traces of elements such as strontium (Sr), copper (Cu), molybdenum (Mo), barium (Ba), cadmium (Cd), chromium (Cr), manganese (Mn) and iron (Fe) (EC, 2004). Re-suspension of road dust depends on a number of factors – road surface, humidity, intensity of traffic and wind speed (Kupiainen, 2007). Significant contributions of Al, Si, K, Ca, Ti, Mn, Fe, Zn and Sr, mostly in the coarse particle fraction, are apportioned to the road/tyres source. Non-exhaust particles can contribute significantly, to overall particulate emissions from the transport sector. However, the scarcity of data in developing countries on non-exhaust emissions makes it difficult to quantify its impact on the overall ambient concentrations. 1.5 LITERATURE REVIEW OF AIRBORNE PARTICULATE MATTER STUDIES IN AFRICA Africa is the second fastest growing economy next to Asia. This rapid economic development has led to urban air pollution in many urban cities in Africa reaching levels that rival the levels that existed in the old days in cities of developed countries. However, the causes of air pollution are multiple. Air pollution can be transported over long distances and have impacts far from their point of emission. Most of sub-Saharan Africa (SSA) urban cities are now experiencing similar air pollution problems, University of Ghana http://ugspace.ug.edu.gh 20 especially anthropogenic particulate matter arising from the rapid growth of vehicle traffic, open waste burning, and the presence of major industries close to residential areas. Air pollution research in African countries is still in its infancy. In contrast to the other parts of the world, very few data are available on air pollution derived from particulate matter in most SSA countries and Ghana in particular – despite their adverse implications on human health, visibility and climate. With the exception of South Africa, data on airborne particulate matter in SSA countries are extremely scarce since they are not routinely collected. Only a few source apportionment studies of particulate matter have been carried out in SSA countries. Chemical characteristics and source apportionment of particulate matter using receptor modelling have been studied in South Africa (Engelbrecht et al., 2001; Kgabi, 2010), Ghana and Gambia (Zhou et al., 2013, Zhou et al., 2014), Nigeria (Oluyemi & Asubiojo, 2001), Tanzania (Bennet et al., 2005; Mkoma et al., 2010), Ethiopia (Etyemezian et al., 2005) and Benin (Fourn and Fayomi, 2006). Some studies to investigate particulate matter characterization and sources in Greater Accra in the past few years have identified three to eight potential sources (Aboh et al., 2009; Ofosu et al., 2012). Much is now known about the levels and sources of air pollution in urban environments, but most information is from more-developed countries. Hence there is an urgent need for measurements of air pollution in African cities. There is also an urgent need to understand the sources and contributions of University of Ghana http://ugspace.ug.edu.gh 21 particulate matter in Accra, Ghana which would enable the development of better emissions control policies and standards. 1.6 OVERVIEW OF ATMOSPHERIC POLLUTION IN GHANA Ghana is one of the most densely populated countries in the sub-region with a population of about 24,658,823 (2010 estimate), growth rate of 2.5%, birth rate of 31.7 birth/1000 population, and death rate of 7.53 death/1000 population per year. More than 40% of the population lives in the urban areas of the country. In 2009, Ghana attained lower middle-income status and in 2010, became an oil producing country (World Bank, 2012). In Ghana, and most of the other sub-Saharan Africa (SSA) countries, there is absence of regular and systematic air quality-monitoring for air quality assessment. As a result, there is a lack of data on the concentrations as well as the characteristics of atmospheric particulate matter. There have been routine monitoring programs in Accra by the Environmental Protection Authority of Ghana (EPA-Ghana) to measure only PM10. There is no systematic record of PM2.5 concentrations in Accra, partly because air quality standard for PM2.5 has not been established in Ghana. The Environmental Protection Agency of Ghana between March 2005 and December 2008 collaborated with the United States Agency for International Development (USAID), the United States Environmental Protection Agency (USEPA) and the United Nations Environment Programme (UNEP) to set up an urban air quality monitoring network in Accra. The objective of the project was to accurately characterize the seriousness and nature of air University of Ghana http://ugspace.ug.edu.gh 22 pollution problems in Accra and to make recommendations for the development of air quality management strategy for Ghana. Results from the study revealed that vehicular exhaust emissions, open burning of waste and other materials, emissions from industrial sources, residential cooking, commercial activities and wind-blown dust are the major contributors to the air quality measured at the monitoring sites (Figure 1.2). Figure 1.2 (a) Cooking along the side of street for small-scale commercial activity (b) vehicular traffic congestion (c) vehicular exhaust emissions (d) wind- blown/harmattan dust in Accra. The results from the selected monitoring sites showed that PM10 concentrations at the roadside locations (178 µg m-3) were higher than those at commercial (81 µg m-3), industrial (72 µg m-3) and residential (61 µg m-3) during the wet season. The roadside (445 µg m-3) site was also found to be higher than those at the commercial (198 µg m-3), (c) (d) (b) (a) University of Ghana http://ugspace.ug.edu.gh 23 industrial (198 µg m-3) and residential (150 µg m-3) during the dry season (Nerquaye- Tetteh, 2009). High particulate matter concentrations at the roadside locations and commercial areas have the tendency to adversely affect the health of the populace. Road traffic congestion is becoming the fastest source of pollution in many major urban centres in sub-Saharan Africa. Motorization in the Accra Metropolitan Area (AMA) is high by African standards, at 90 vehicles per 1,000 population, as compared to 20-30 for Nairobi, Dar es Salaam, and Addis Ababa (GEF, 2007). This is partly because of the high number of taxis, trotros and okada (motorcycles used for commercial services).This trend is increasing with 90% of the vehicles imported into the country remaining in the cities. Table 1.2 shows a summary of the number of vehicles registered by the Driver and Vehicle Licensing Authority (DVLA) for Accra alone from 2000 to 2009. Traffic in Accra is characterized by heavy congestion (particularly during the peak periods), low vehicle utilization, heavy dependence on informal private bus services, weak implementation of traffic management measures, inadequate facilities for pedestrians and bicyclists, poor road safety arrangements and high accident rates. Vehicular exhaust emissions have been a significant cause of poor urban air quality over the years in Ghana (EPA, 2002). This together with heavy traffic in certain urban locations contributes to poor urban air quality. University of Ghana http://ugspace.ug.edu.gh 24 Table 1.2 Summary of vehicles registered by DVLA for Accra. Year Registered Vehicles (cumulative) Year Registered Vehicles (cumulative) 2000 511,083 2005 767,067 2001 567,780 2006 841,314 2002 613,153 2007 932,540 2003 643,824 2008 1,033,140 2004 703,372 2009 1,128,138 Source: (DVLA, 2010) Another major source of airborne particulate matter pollution in Ghana is biomass burning. Wood and charcoal are the most common solid fuels used in Ghana (Benneh et al., 1993). Songsore and McGranahan (1998), based on data from 2000 Population and Housing Census and the 2003 Demographic and Health Survey, report that more than 60% of Accra's population use biomass fuels (charcoal or firewood) as their primary fuel for cooking. Boadi and Kuitunen (2005) in a study undertaken in the Accra Metropolitan Area revealed that biomass use was higher in households with lower income and socioeconomic status. The EPA-Ghana in the Ghana State of Environment Report 2004, listed the following as the main issues related to air quality in the country:  Inefficient utilization of fuels;  Poorly planned modes of transport;  Poorly serviced motor vehicles;  Inefficient cook-stoves and fireplaces; University of Ghana http://ugspace.ug.edu.gh 25  Rudimentary kilns and stoves in industries;  Charcoal production;  Widespread bush burning. 1.7 SOURCE APPORTIONMENT OF PARTICULATE MATTER In order to design effective programmes and strategies for reduction of particulate matter concentration in ambient air, it is important and necessary to identify the original sources of the particulate matter, and to quantitatively apportion the observed aerosol particle concentrations among the various source types. Source apportionment methods provide the tools to formulate efficient and effective particulate matter control strategies and to develop policy to prevent human exposure. Source apportionment of particulate matter refers to the methods used to quantify the contributions of different source categories to the particulate matter concentrations measured in the atmosphere (Pant and Harrison, 2012). Particulate matter pollution is comprised of myriad of chemical species emitted from multiple sources. Source apportionment models utilize the mass and chemical compositions of the particulate matter measured at a sampling site (receptor site) to resolve the main source types and estimate their contributions to the measured ambient particulate matter at that site (Belis et al., 2013). Particulate matter emissions from specific sources often have unique elemental profile by which the contribution of these sources to the total aerosol particles at the receptor can be recognized (Watson, 1984). University of Ghana http://ugspace.ug.edu.gh 26 Air pollution models can be divided into two primary categories: receptor models and dispersion models. Receptor models are formulated to begin with pollutant information monitored at a receptor and to look backward, using data on several species and information about relative concentrations of those species from possible sources, to apportion the pollutant to the sources (Miller et al., 2002). In contrast, chemical transport, or dispersion models start with the source characteristics and use physics, mathematical, and chemistry calculations to predict pollutant concentration at some distance from the source. Important input for those dispersion models includes information about the emissions from the source, the local atmospheric conditions, and some geographical characterization. Both types of models have been highly developed and forms of them are widely used for prediction and diagnosis of events (USEPA, 2004). 1.7.1 RECEPTOR MODELLING METHODS Receptor models are mathematical or statistical techniques used to identify and quantify the sources of airborne pollutants at a receptor or sampling site. The fundamental principle of receptor modelling is that mass conservation is assumed and on this basis a mass balance analysis can be used to determine and apportion ambient particulate matter concentrations to individual emitting sources (Hopke et al., 2006). Most receptor models used for source apportionment are in principle based on the assumption that the concentration of the pollutants measured at the receptor site is equal to the sum of the concentrations induced by the surrounding emission sources emitting the pollutants University of Ghana http://ugspace.ug.edu.gh 27 (Fabretti et al., 2009; Belis et al., 2013). The main objective of receptor models is, therefore, to identify the possible sources of particulate contaminant and to obtain data on their contributions to the bulk particle mass. Receptor models are commonly divided into two primary categories: chemical mass balance and multivariate methods (Henry et al., 1984; Pollice, 2009). The selection of the appropriate method depends on prior knowledge on the sources and source profiles. If the sources are known and detailed information on source profiles is available, Chemical Mass Balance (CMB) models can be applied. Alternatively, in cases where the sources are unknown and only the concentration of ambient pollution is known, multivariate methods such as Positive Matrix Factorization (PMF) methods are preferred. Examples of the commonly used receptor model software in physical and chemical sciences applications include the USEPA’s Positive Matrix Factorisation ((PMF 3.0) USEPA, 2008), Unmix 6.0 (USEPA, 2007) and Chemical Mass Balance ((CMB 8.2) USEPA, 2004). Enrichment factor analysis and multivariate techniques such as correlations and Positive Matrix Factorization (PMF) were used to define a relationship between the sources and the receptor. These analytical methods were combined to assist in the identification of sources and the apportionment of the observed pollutant concentrations to those sources in the urban area of Accra, Ghana. Receptor models of positive matrix factorization (PMF) and enrichment factor (EF) were used in this study. University of Ghana http://ugspace.ug.edu.gh 28 1.7.2 POSITIVE MATRIX FACTORIZATION Positive matrix factorization (PMF) is a powerful and widely used multivariate method used extensively for source apportionment of ambient particulate matter (Paatero and Tapper, 1994; Paatero, 1997). PMF model resolves the dominant positive factors that contribute to PM samples without prior knowledge of the sources. This is achieved by using measured concentrations of particulate matter to estimate the number of sources, the source composition, and the source contribution to each sample. The goal of PMF modelling is to  determine the number of factors (sources or chemical/physical processes) that adequately explain the input data set variability and  find correlation among the measured variables. The advantages of the PMF relative to the traditional factor analysis methods such as Cluster Analysis (CA) and Principal Component Analysis (PCA) are:  Unreliable data, such as observations below detection limit or missing values, can be included in the PMF analysis by giving them low weights to decrease their influence in the modelling (Paatero and Tapper, 1994).  Data characterized by heavy positive skewed distribution can also be handled by down-weighting those extreme points to reduce undue influence in the model (Huang and Conte, 2009). University of Ghana http://ugspace.ug.edu.gh 29  PMF model assumes the non-negativity of factors and does not rely on information from the correlation matrix but utilizes a point-by-point least- squares minimization scheme by taking into account the uncertainty of each data point (Pongkiatkul and Oanh, 2012). In PMF, a matrix X (n × m), where n is the number of samples and m is the number of chemical species is factored into two matrices, G (n × p) and F (p × m), where p is the number of independent source types or factors extracted, and a residual matrix E is used to account for the unexplained part of X. The factor analysis model can be written in matrix form as shown in equation 1.2: or in component form as ∑ where is the measured concentration at a receptor for the species in the air sample, is the particulate mass concentration from the source type contributing to the air sample, is the species mass fraction from the source, and is the residual between measured and modelled concentrations for the species in the sample. The objective of the multivariate receptor modelling is to determine the number of sources , the source contributions , and the chemical profiles of the University of Ghana http://ugspace.ug.edu.gh 30 identified sources that best reproduce . To obtain the best fit of the model, PMF minimizes the object function, or chi-squared based upon the uncertainties of each observation. The object function is given by the equation: ∑∑( ) ∑∑ ( ∑ ) where is the uncertainty in the species in the sample; PMF simultaneously adjust the elements of G and F in each iterative step until a minimum value of Q is obtained (Paatero, 1997; Polissar et al., 1998). The Q value can be used to determine the optimum number of the factors. The theoretical Q value should be approximately equal to a value of , the number of entries of data array or the number of degree of freedom of the datum in the data set. 1.7.3 ENRICHMENT FACTOR Enrichment factor model helps to differentiate between elements originating from anthropogenic activities and those from natural sources. It is used mainly to provide an initial assessment on the degree of contributions from man-made activities to that of the measured atmospheric elemental concentrations. Consequently, it is used to estimate the degree of anthropogenic contamination. Enrichment factor (EF) analyses have been University of Ghana http://ugspace.ug.edu.gh 31 extensively used in particle source apportionment studies to identify the major sources of air pollution and to quantify contributions of all sources of all measured pollutants (Winchester, 1981; Davidson et al., 1986; Chao and Wong, 2002; Cao et al., 2003; Zhang et al., 2008). University of Ghana http://ugspace.ug.edu.gh 32 1.8 MOTIVATION AND OBJECTIVES OF STUDY Urban air pollution from particulate matter is on the rise in sub-Saharan Africa. Rapid growths in population, urbanization, industrialization, and motorization have led to serious deterioration of the air quality in urban cities in developing countries like Ghana. Increasing epidemiological evidence has established the adverse health effects associated with ambient particulate matter pollution (Pope et al., 1992, Dockery et al., 1993, Dockery et al., 1994). Little or no data exists on particulate matter levels and composition in Ghana to gauge the severity or nature of the air pollution problem and its impact on the population. This presents a challenge for those actively producing the emissions and agencies involved with monitoring and regulating emissions. Thus, there is an urgent need to better monitor and manage urban air quality and identify the most effective measures to reduce pollution. In order to assess and to lessen the impacts of particulate matter pollution, any program aimed at regulating the levels of particles in the urban atmosphere requires knowledge on the size distribution, chemical composition and sources of aerosol particles. Monitoring of particulate matter pollution was performed so that the data can be analysed to obtain more information about the trends, nature and concentrations of the particles. University of Ghana http://ugspace.ug.edu.gh 33 For these reasons, PM2.5 and PM10 particles were measured at four neighbourhoods in Accra: (i) the indigenous older inner core low-income areas of James Town/Usher Town; (ii) the informal migrant, low-income settlements of Nima/Maamobi outside the Ring Road; (iii) Asylum Down, a middle-income neighbourhood located between the above two low-income areas; and (iv) the high-income area of East Legon. The above data collected were analysed for particle mass and composition. Data on particle composition allowed for the determination of within-neighbourhood and between-neighbourhood variations of particle levels and sources. The chemical characteristics (composition) of particulate matter, source apportionment, the influence of meteorological processes and their relationships at the four neighbourhoods are addressed in this study. The advanced data analysis modelling techniques called receptor models provided the tool to determine the information describing the origin of pollution emissions. Positive Matrix Factorisation (PMF) was the receptor model used in this thesis to identify and apportion the sources of pollution emissions. University of Ghana http://ugspace.ug.edu.gh 34 1.8.1 OBJECTIVES OF STUDY The overall objective of this study is to contribute to current knowledge on the characteristics and sources of urban ambient particulate matter in Accra, Ghana. It also seeks to resolve the number of relevant sources, their chemical composition, the amount that each source contributes to particulates matter and the influence of meteorological processes on particles in Accra, Ghana. This will improve understanding of the sources that should be included within an air quality control program within a sub-Saharan developing country. 1.8.2 SPECIFIC OBJECTIVES OF THE STUDY The specific objectives are:  To determine the mass concentrations of the PM2.5 and PM10 in Accra, Ghana.  To characterize chemical composition and concentrations of particulate matter using EDXRF.  To determine and compare major sources of ambient fine particulate matter in specific neighbourhoods of Accra.  To investigate the natural and anthropogenic contributions to particulate matter pollution in Accra, Ghana.  To investigate the influence of meteorological conditions in Accra on particulate matter emissions.  To investigate the influence of seasonal variations in particle mass and concentrations. University of Ghana http://ugspace.ug.edu.gh 35 CHAPTER 2: CLIMATE AND METEOROLOGY OF GHANA 2.1 OVERVIEW OF THE CLIMATE OF GHANA Ghana (Figure 2.1) lies on the west coast of Africa along the Gulf of Guinea between latitude 4.5°N and 11.5°N and longitude 3.5° W and 1.3°E with a total land area of 239,460 km2 and 8,520 km2 of water (EPA, 2002). It is bordered on the East by the Republic of Togo, the West by Cote d’Ivoire, the North by Burkina Faso and the South by the Gulf of Guinea or Atlantic Ocean. The country is divided into 10 administrative regions. The climate of Ghana is tropical and humid. Ghana’s climate is controlled by three air masses namely, the Tropical Maritime Air Mass (mT) or South-West Monsoon, the Tropical Continental Air Mass (cT) or North-East Trade Wind and the Equatorial Eastern (E). The cool and moisture saturated South-West Monsoon which originate from the Atlantic Ocean and the warm, dry and dusty Tropical Continental Air Mass (Harmattan) from the Saharan desert approach the tropics from opposite sides of the equator and flow towards each other into a low pressure belt known as the Inter- Tropical Convergence Zone (ITCZ) (Ojo, 1977). Most of the country experiences two rainy seasons which occur from March to July and from September to November. The only exception is in the northern savannah where only one rainy season occurs from May to September, with the most rain falling in August and September, and one dry season. University of Ghana http://ugspace.ug.edu.gh 36 Figure 2.1 Map of Ghana showing administrative regions and borders (Ghana Homepage, 2014). Generally, the forested zones experience heavier rainfall and higher temperatures than the coastal zones while the northern savannah area experiences hotter and drier conditions with daytime temperatures mainly about 90 °C and above for much of the year. The quantity of rainfall in Ghana is variable. This seasonal variation in the rainfall patterns in the country is due to the annual north-south movement of the low pressure belt known as the Inter-Tropical Convergence Zone (ITCZ). University of Ghana http://ugspace.ug.edu.gh 37 The ITCZ is where the hot, dry and dust Harmattan air mass from the Sahara in the North meets the cool, moist monsoon air mass from the South Atlantic (Afeti and Resch, 2000; Andah et al, 2003; Breuning-Madsen and Awadzi, 2005). Figure 2.2 depicts the movement of both the northern and southern ITCZ. From October to April, the cold and dusty northeasterly Harmattan winds blow south-west from the Sahara Desert, reducing visibility to as little as 1 km and slightly decreasing night temperatures. According to Salm and Falola (2002), the harmattan condition is widespread in the north between November and March, but reaches south to the coastal regions, usually for a few weeks to a month during December and January. Figure 2.2 Movements of the Inter-Tropical Convergence Zones (Source: http://winds.jpl.nasa.gov_images_equatormap1.gif.htm) Annual rainfall ranges from about 1,100 mm in the north to about 2,100 mm in the south-western part of the country. The average relative humidity in the northern and southern parts of the country is approximately 44% and 80%, respectively (Dickson and Benneh, 1988). Moreover, temperatures within the country vary with seasons and elevation. Hence, most parts of the country record the highest and lowest temperatures University of Ghana http://ugspace.ug.edu.gh 38 in March and August, respectively. The average temperature is usually between 25 °C and 30 °C (Breuning-Madsen and Awadzi, 2005). 2.2 LOCAL METEOROLOGY Ghana is geographically divided into six agro-ecological zones on the basis of their prevalent climatic conditions and vegetation (Figure 2.3). The agro-ecological zones are rainforest, coastal savannah, semi-deciduous forest, transitional zone, Guinea savannah and Sudan savannah zones. The city of Accra lies within the coastal-savannah zone with a relatively humid climate and low annual rainfall varying between 600 and 900 mm, with an annual average of about 800 mm distributed over less than 80 days (Obuobie et al., 2006; Oppong and Badu, 2013). Relative humidity is generally high, ranging from 65% in the mid- afternoon to 95% at night. Accra’s rainfall pattern is bimodal with the major rainy season occurring from March to July, and the minor rainy season from September to October (Table 2.1). The major dry season occurs between November to the middle of March or early April (Gordon et al., 2003). This dry period coincides with the Harmattan season which occurs from November to March. The Harmattan is dry and dusty winds from the Sahara Desert which blows from the northeast. It is characterized by dry, hot days and relatively cool nights. The mean monthly temperature in the city varies from 24 ºC in August to 28 ºC in March, with an annual average of 27 ºC. The city lies close to the equator and University of Ghana http://ugspace.ug.edu.gh 39 therefore the daylight hours are quite uniform throughout the year. The predominant wind direction in Accra is from the WSW to NNE sectors with wind speeds normally varying from 8 km h-1 to 16 km h-1. The maximum wind speed recorded in Accra is 107.4 km h-1. In this study, seasonal variations are divided into Harmattan and non-Harmattan seasons. The Harmattan period was from 27th December 2007 to 7th February 2008. The period from 1st September 2007 to 26th December 2007 and 8th February to 17th August 2008 will be described as non-Harmattan. Hence, the entire study spanned the period 27th December 2007 to 17th August 2008. University of Ghana http://ugspace.ug.edu.gh 40 Figure 2.3 Map of Ghana showing the various agro-ecological zones (Benneh andAgyapong, 1990). University of Ghana http://ugspace.ug.edu.gh 41 Table 2.1 Typical rainfall characteristics of agro-ecological zones in Ghana Agro-ecological zone Area (km2) Mean annual rainfall (mm) Range (mm) Major rainy season Minor rainy season Sudan savannah 2200 1000 No data May–Sept No data Guinea savannah 147900 1000 800–1200 May–Sept No data Transitional zone 8400 1300 1100–1400 Mar–July Sept–Oct Deciduous forest 66000 1500 1200–1600 Mar–July Sept–Nov Rain forest zone 9500 2200 800–2800 Mar–July Sept–Nov Costal savannah 4500 800 600–1200 Mar–July Sept–Oct Source: Data adapted from the Ghana Meteorological Authority, Accra, Ghana. 2.3 PARTICULATE MATTER AND METEOROLOGICAL PARAMETERS The inclusion of meteorological parameters in ambient particulate matter studies has become important since they can determine the transport and dilution of pollutants in the atmosphere, influence chemical transformations of species in the air, and affect the mechanisms and rates of removal of pollutants from the atmosphere. Many studies have shown that PM10 and PM2.5 concentrations in the ambient air are influenced by various meteorological parameters such as temperature, relative humidity (RH), wind speed, University of Ghana http://ugspace.ug.edu.gh 42 wind direction, and precipitation (Rajkumar and Chang, 2000; DeGaetano and Doherty, 2004; Tsai and Cheng, 2004; Karar et al., 2006; Tecer et al., 2008; Mkoma and Mjemah, 2011). The meteorology of an area can change many times in one day and affect the PM levels, hence the need to investigate the relationship between meteorological variables and PM levels. In the absence of site-specific meteorological data for the four neighborhoods, temperature, rainfall, relative humidity, wind speed and wind direction, were obtained from the Ghana Meteorology Authority, Accra. The meteorological parameters obtained were thought as being representative of the monitoring sites. In view of the fact that meteorological conditions influence the concentration of particulate matter, the Harmattan season (December – February) and the non-Harmattan season (September – December and February – August) were examined separately. During this study, the Harmattan season was rather dry with no rainfall in Accra. Moreover, there was little or no variation in the average local ambient temperature for both seasons. The relative humidity fluctuated between 18 – 92% and 57 – 91% for Harmattan and non-Harmattan seasons respectively (Table 2.2). The mean monthly values for temperature, relative humidity and rainfall are presented for the days with particulate sampling (Figure 2.4). The rainfall and relative humidity exhibited an inverse relationship with temperature. The mean wind speed varied over the entire sampling period with the highest (10.5 m s-1) in August 2008 and the lowest (6.3 m s-1) in January 2008. University of Ghana http://ugspace.ug.edu.gh 43 Table 2.2 Average (or total amount) of meteorological parameters and ranges during the sampling periods in Accra, Ghana. Parameters Min Max Total/Mean Harmattan season Precipitation (mm) 0 0 0 Relative humidity (%) 18 92 57 ± 23 Temperature (°C) 28 30 29 ± 0.7 Wind speed (m s-1) 4 10 6.5 ± 2.1 Non-Harmattan season Precipitation (mm) 0 62.8 179 Relative humidity (%) 57 91 74 ± 11 Temperature (°C) 23 30 28 ± 2 Wind speed (m s-1) 3 12 8 ± 2 Figure 2.4 Monthly average of atmospheric temperature (°C), relative humidity (RH, in %) and rainfall (mm) for the period September 2007 – August 2008 in Accra. (Source: Drawn using data from Ghana Meteorological Authority, Accra, Ghana.) University of Ghana http://ugspace.ug.edu.gh 44 2.4 AIR MASS CLIMATOLOGY AND BACKWARD TRAJECTORY ANALYSIS Backward trajectory analysis is very useful in air pollution studies and can provide important information on air mass origins. Backward trajectory analysis can be used to track the history of air masses in which atmospheric aerosols are transported to the measurement site. In this study, air mass backward trajectory analysis was performed using the Hybrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT version 4) developed by the National Oceanic and Atmospheric Administration (NOAA) Air Resource Laboratory (Draxler and Rolph, 2003). The model runs on the principle, which is based on the integration of the position of air mass with regard to time. Backward air trajectories are presented in graphical forms indicating the pollutant sources and vertical heights above ground level (AGL). Air mass backward trajectory analyses were calculated between September 2007 and August 2008 at 00:00 UTC or 24 h local time for the non-Harmattan and Harmattan seasons. Each four-day isentropic air mass backward trajectory was routed through three different heights (500, 1500 and 3000 m) AGL. This was to identify pollutants which are transported by air parcels near ground level (≤ 500 m) and those associated with long-range transport (between 500 m and 3000 m) which are likely to be free of the influence of local wind patterns caused by buildings, trees and other objects.. The model was run for days during which high PM source contributions were observed at the four sites. University of Ghana http://ugspace.ug.edu.gh 45 The three-dimensional motion was obtained using 3-hourly output from the Global Data Assimilation System (GDAS) for both horizontal and vertical air mass streams. Results of backward trajectory analysis can be combined with observations to better explain the chemical characteristics of PM2.5 at a receptor site (Aneja et al., 2006). University of Ghana http://ugspace.ug.edu.gh 46 CHAPTER 3: METHODOLOGY 3.1 DESCRIPTION OF SAMPLING SITES Accra (5o 33ꞌ N, 0o 13ꞌ W), the capital city of Ghana is one of the fastest-growing cities in sub-Sahara Africa (Møller-Jensen et al., 2005; CIA, 2011). With an estimated population growth rate of about 4% per annum, Accra has an estimated population of about 1.7 million (GSS, 2010). The Accra Metropolitan Area (AMA) with a population density for 112 per kilometre squared, has almost 42% of the total population of the Greater Accra Region (www.ghanadistricts.com). Ambient aerosol samples were collected in four neighbourhoods of varying socioeconomic status (SES) located in Accra between September 2007 and August 2008. The four study neighbourhoods of James Town/Ussher Town (JT), Asylum Down (AD), Nima (NM), and East Legon (EL) lie on a line from the coast of the Atlantic Ocean to the northern boundaries of the Accra Metropolitan Area (AMA) (Figure 3.1). University of Ghana http://ugspace.ug.edu.gh 47 Figure 3.1 Four study neighbourhoods lying on a line from the coast to the northern boundaries of the Accra Metropolitan Area (AMA). AD, EL, JT and NM represent Asylum Down, East Legon, James Town/Ussher Town and Nima neighbours respectively. ATLANTIC OCEAN University of Ghana http://ugspace.ug.edu.gh 48 3.1.1 JAMES TOWN/USSHER TOWN (JT) NEIGHBOURHOOD James Town/Ussher Town (JT), a low lying area (less than 5 m above sea level), is a low-income indigenous area of Accra. It is one of the oldest neighbourhoods in Accra and remains an active fishing centre lying between the coast and the Central Business District (CBD). JT with a population density of 329 people per 10,000 m2 is one of the densely populated areas within the Accra Metropolitan Area. The major road through this neighbourhood is paved but large blocks of residential areas have only pedestrian alleys paved with either cement or pavement blocks. The main local sources in JT are biomass burning mainly for domestic and commercial cooking, open burning of refuse and used tires, road and re-suspended soil dust, sea spray, emissions from vehicles and industrial processes. 3.1.2 NIMA (NM) NEIGHBOURHOOD Nima (NM) is a low-income non-indigenous area of Accra. The population of Nima has increased substantially from 29,797 in 1960 to 52,906 in 1984, to 69,044 in 2000. Like JT, NM is an overcrowded, unsanitary, and unplanned neighbourhood with few public amenities. Nima is located to the northeast of Accra city centre and about five miles away from it. It extends over an area of about 351.6 acres of sloping ground; bounded in the west by Maamobi, east by Kanda, north by Kokomlemle and south by Accra New Town. The major road in Nima is the Nima Highway which links this neighbourhood to all the surrounding neighbourhoods. For most of the daytime period, this road is extremely busy and congested with vehicular traffic. In this neighbourhood, traffic, road University of Ghana http://ugspace.ug.edu.gh 49 and re-suspended soil dust, open burning of refuse, biomass burning by domestic houses and commercial street food vendors, are the main local sources. 3.1.3 ASYLUM DOWN (AD) NEIGHBOURHOOD Asylum Down (AD) is a middle-income, mostly residential neighbourhood bordered by the Ring Road Central, one of the largest, busiest and congested roads in Accra. AD has a combination of residential and commercial buildings. It is located 3 km inland from the coast lying north of Castle Road and east of Barnes Road. Traffic through this neighbourhood is generally light, although the Ring Road Central is congested in the afternoon and evening most days of the week. The main local sources in AD are road and re-suspended soil dust, small businesses such as dry cleaners and auto body shops, outdoor burning and traffic. 3.1.4 EAST LEGON (EL) NEIGHBOURHOOD The high-income East Legon (EL) neighbourhood is a sharp contrast to the low-income areas of JT and NM. It is an upper-class, sparsely populated, planned residential neighbourhood with well-developed infrastructure. It has a population density of 5 people per 10,000 m2. EL is located about 10 km inland away from the coast and lies north of Accra International Airport, 13 km northeast of Central Business District. There is heavy vehicular traffic (from cars, trotros, buses and taxis) on the main road which connects EL to the city centre, primarily during the morning and evening commute periods. The main local sources in EL are road and soil dust, traffic and biomass burning. University of Ghana http://ugspace.ug.edu.gh 50 3.1.5 STUDY DESIGN In this study, measurements of particulate matter in samples collected from four selected neighbourhoods in Accra were carried out between September 2007 and August 2008. Monitoring sites within the neighbourhoods were selected as follows: 1. One site in the Nima neighbourhood was located on a road with medium to heavy traffic for the whole day or parts of the day. This site was designated as the neighbourhood traffic site (NM-T). 2. One site in each of the four neighbourhoods was located in areas selected to be typical residential parts of the neighbourhoods. The residential sites in Asylum Down, East Legon, James Town/Ussher Town, and Nima were designated as AD-R, EL-R, JT-R and NM-R respectively. They were situated on secondary roads or alleys with no or significantly less traffic than the Nima neighbourhood traffic site (NM-T), although within these sites there could be other polluting sources such as biomass burning or small-scale industries. 3. Other criteria for site locations were security of the equipment, access to electricity, free from public interference, and uninterrupted access to the equipment for operation and maintenance. The sites were also chosen to allow for free flow of air around the samplers’ inlet, thus preventing the sampling of stagnant air or highly sheltered microenvironment. The characteristics of measurement sites are provided in Table 3.1. University of Ghana http://ugspace.ug.edu.gh 51 This study was designed to characterize the particulate matter pollution over the city of Accra by examining its seasonal and daily patterns as well as its variation between neighborhoods. Moreover, the design will be used to identify and quantify the primary contributions of air pollution sources responsible for the airborne particulate matter pollution in the city using receptor-based modeling. University of Ghana http://ugspace.ug.edu.gh 52 Table 3.1 Characteristics of measurement sites Residential (R) Traffic (T) JT Height (m) 6.24 Distance to nearest primary road (m) 192 Road surface Paved Traffic pattern a Connecting secondary road with light traffic AD Height (m) 2.85 Distance to nearest primary road (m) 166 Road surface Paved Traffic pattern a Connecting secondary road with medium-heavy traffic during morning and evening rush hours and light traffic at other times NM Height (m) 6.61 3.33 Distance to nearest primary road (m) 188 25 Road surface Paved Paved Traffic pattern a Local secondary road with light traffic Primary road with heavy moving traffic all day and light traffic over night b EL Height (m) 2.88 Distance to nearest primary road (m) 330 Road surface Paved Traffic pattern a Local secondary road with light traffic a Traffic patterns are based on qualitative assessment by study investigators. Road type was categorized as alley, local secondary road, connecting secondary road and primary road. b NM-T was located next to a busy central bus station which frequently had idling vehicles. University of Ghana http://ugspace.ug.edu.gh 53 3.2 DESCRIPTION OF SAMPLER AND FILTER MEDIA USED 3.2.1 AEROSOL SAMPLER Several particulate matter (PM) sampling methods which allow for collection of particles on a filter that is weighed and subsequently subjected to analytical analysis have been deployed for PM studies (Wilson et al., 2002). These methods employ samplers such as dichotomous samplers, United States Environmental Protection Agency (USEPA) Federal Reference and Equivalent Method (FRM) samplers, Interagency Monitoring of Protected Visual Environments (IMPROVE) samplers and Harvard Impactor (HI) samplers which are some of the most commonly used particle samplers for the collection of PM2.5 and PM10 integrated samples. Manual sampling devices are needed to collect atmospheric particles for subsequent gravimetric mass measurement and elemental analysis by energy dispersive X-ray fluorescence (ED-XRF). The collection is usually done on a filter placed inside a sampler to obtain an integrated sample and a time averaged concentration measurement. A relatively low sampling flow rate of a few liters per minute is adequate if a highly sensitive technique, such as proton-induced X-ray emission (PIXE) fluorescence or energy dispersive X-ray fluorescence (ED-XRF), is used to analyze the collected particle sample (Pui and Liu, 1988). Harvard Impactors (Marple and Willeke, 1976; Marple et al., 1987) were used for integrated particles mass measurement in this study. HI is a relatively low flow particle University of Ghana http://ugspace.ug.edu.gh 54 sampler that uses an oiled impactor plate to minimize particle bounce and provide a sharp cut point, giving measurements similar to U.S. EPA reference methods (Chow, 1995). The Harvard Impactor samplers consist of a particle size selective inlet (SSI), an acceleration nozzle, either oiled impaction plates or polyurethane foam (PUF) pad, and Teflon membrane filter mounted in a plastic holder. The filter holder consists of a filter cassette top and a filter cassette bottom that press a plastic filter slide between two gaskets (Marple et al., 1987). Figure 3.2 adapted from a similar figure by Sarnat et al. (2003) shows a schematic diagram of the Harvard Impactor. In this study, PM10 concentration was measured using a HI with a D50 (50% collection efficiency) of 10 μm (aerodynamic diameter) at a flow rate of 4 l min-1 (± 10%), with two consecutive pre-oiled impactor plates serving as the impaction surface to reduce the effects of particle bounce. PM2.5 concentration was measured using a modified HI combined with a polyurethane foam (PUF) PM2.5 size-selective inlet, with a D50 of 2.5 μm at a flow rate of 5 l min-1 (±10%), with a PUF pad serving as the impaction surface. At each monitoring site, Harvard Impactor samplers were installed on rooftops of either homes or businesses, at heights ranging from 4 to 7 meters above ground level so that the air was relatively well mixed and less likely to be strongly affected by a source in the immediate vicinity (Dionisio et al., 2010). The inlet nozzles of the samplers were approximately 1.2 meters above rooftop level with about a meter spacing between the inlets to ensure that the sampler inlet was able to sample wind coming from all directions. A typical set-up of the ambient monitoring system is shown in Figure 3.3. University of Ghana http://ugspace.ug.edu.gh 55 Figure 3.2 Schematic diagram of Harvard Impactor particle sampler (Sarnat et al., 2003). University of Ghana http://ugspace.ug.edu.gh 56 Figure 3.3 A typical set-up of the ambient monitoring system. Residential site (left) and traffic site (right) with integrated ambient 48-hour Harvard Impactor (below) mounted on roofs of buildings to measure PM10 and PM2.5 concentrations. University of Ghana http://ugspace.ug.edu.gh 57 3.2.2 FILTERS The most effective and prevalent method of capturing aerosol particles is by filtration. Filters are porous media for collecting particulate matter. Filters used for aerosol sampling are made of a tightly woven fiber mat or plastic membrane penetrated by microscopic pores. The two general types of filters are fibrous filters and membrane filters. Fibrous filters are characterized by low pressure drop and high collection efficiencies for all particle sizes. Examples of the most common types of fibers in this filter are cellulose wood fibers, glass fibers, quartz fibers and plastic fibers. Glass and quartz fiber filters exhibit a high retention of particles with sizes above 0.3 µm (IAEA, 1992). Membrane filters retain particles on the surface of the filter, enabling non-depth analyses such as X-ray fluorescence (XRF). They generally have higher flow resistance and lower loading capacities than fiber filters, and are also more expensive. Membrane filters are more suitable for trace elemental analysis studies using analytical methods such as instrumental neutron activation analysis (INAA), PIXE, ED-XRF or microscopic analysis. They are available in a wide range of materials including Teflon, polyvinyl chloride (PVC), cellulose ester and polycarbonate. Filter media used in monitoring activities must display the physical and chemical properties appropriate for the sampling methodology and the laboratory analysis methods used. A summary for some useful properties of filters are shown in Table 3.2. University of Ghana http://ugspace.ug.edu.gh 58 Table 3.2 Example of summary of useful filter properties (USEPA, 1999). Filter and Filter Composition Density mg/cm2 pH Filter Efficiency % Teflon® (Membrane) (CF2)n (2μm Pore Size) 0.5 Neutral 99.85 Cellulose (Whatman 41) (C6H10O5)n 8.7 Neutral (Reacts with HNO3) 58% at 0.3 µm Glass Fiber (Whatman GF/C) 5.16 Basic pH - 9 99.0 "Quartz" Gelman Microquartz 6.51 pH - 7 98.5 Polycarbonate (Nuclepore) C15H14+CO3 (0.3μm Pore Size) 0.8 Neutral 93.9 Cellulose Acetate/Nitrate Millipore (C9H13O7)n (1.21 μm Pore Size) 5.0 Neutral (Reacts with HNO3) 99.6 All PM2.5 and PM10 samples in this work were collected on polytetrafluoroethylene (PTFE) Teflon filters with ring (Pall Life Science, Teflon, 0.2 µm pore size, 37 mm diameter), back-supported by a Whatman drain disc. These PTFE Teflon filters have been used in low volume sampling applications due to their advantage of minimizing particle bounce and preserving unstable compounds, low background levels of many analytes and cleanest IR absorbance spectrum (Harrison and Yin, 2004). Therefore, Teflon was chosen as a medium for particle collection. The collected aerosol particles on the Teflon filter were used to obtain the mass concentration by gravimetric analysis and the elemental characterization by nondestructive energy dispersive X-ray Fluorescence (ED-XRF) spectrometry. The net mass of particles is determined by the University of Ghana http://ugspace.ug.edu.gh 59 difference in the filter pre-weighed mass and after-sampling mass. Volumetric airflow rates through each filter were measured at the start and end of each 48-hour measurement using a pre-calibrated rotameter. The equipment and air flow rates of the pumps were checked at 24-hours intervals and adjusted as needed. The average flow rate for each sample was calculated from the mean of the initial and final sampler flows. Field blanks and duplicate filters were collected at multiple sites routinely on a rotating basis. The Teflon filters were changed manually every 48 hours during the sampling period. Durations of each sampling event were recorded by a timer on the pumps. All the filters used for this study were prepared at Harvard School of Public Health (HSPH) and shipped to Accra for deployment at the sampling sites. The exposed filters and field blanks were placed into polycarbonate Petri dishes and returned to the HSPH laboratory for gravimetric analysis. 3.3 SAMPLING PERIODS Ambient sampling of particulate matter was collected for one year in James Town/Ussher Town (JT), Asylum Down (AD), Nima (NM), and East Legon (EL) – four neighbourhoods located the Accra Metropolitan Area (AMA). The sites were designated as follows:  James Town/Ussher Town (JT-R),  Asylum Down (AD-R),  Nima (NM-R and NM-T), and  East Legon (EL-R). University of Ghana http://ugspace.ug.edu.gh 60 Data was collected from 1st September 2007 to 17th August 2008 (Figure 3.4). The samples were collected over 48-hour sampling periods once every six days simultaneously at five rooftop monitoring sites. The 48-hour measurements started at 6:00 am during each sampling period throughout the entire study period. Figure 3.4 Diagram of 48 h averaging sampling schedule at four Accra neighbourhoods during study period. As a result of unannounced or planned load shedding exercise embarked upon by the Volta River Authority (VRA) and the Electricity Company of Ghana (ECG) during study period, frequent and extended electric power outages necessitated using a power backup system at each monitoring site to prevent or limit the effects of data lost. The system consisted of 12-V 100-Ah batteries connected to chargers and inverters. Sept. ‘07 Oct. ‘07 Nov. ‘07 Dec. ‘07 Jan. ‘08 Feb. ‘08 Mar. ‘08 Apr. ‘08 May ‘08 Jun. ‘08 Jul. ‘08 Aug. 08 1 17 NM (NM-R, NM-T) JT (JT-R) AD (AD-R) EL (EL-R) Legend 1st Sept. 2007 – 17th Aug. 2008 1st Sept. 2007 – 17th Aug. 2008 1 st Sept. 2007 – 17 th Aug. 2008 1st Sept. 2007 – 17th Aug. 2008 University of Ghana http://ugspace.ug.edu.gh 61 3.4 ANALYTICAL METHODS Determination of the composition of ambient particulate matter involves the analysis of deposits collected on filter media. The chemical composition of particulate matter in ambient air provides essential information for assessment of contribution from different sources. For this reason, analytical techniques such as gravimetric analysis and energy dispersive X-ray fluorescence (EDXRF) are widely used to analyze the filter samples of air particulate matters. There are several other analytical methods available for the determination of multi- element on filter media and the choice of the most appropriate method depends on the analytes of interest, the specificity needed (i.e., screening vs. fully quantitative) and the sensitivity needed, commonly referred to as the detection limit. The sensitivity, which indicates the ability of an analytical method to detect specific elements at the lower end of the range of concentrations of concern, may change from instrument to instrument because of X-ray generator frequency, multichannel analyzer sensitivity, and sample interferences (US EPA, 1999a; US EPA, 1999b). Specificity is the ability of an analytical technique to differentiate between a certain substance and other similar chemicals. Other analytical techniques used for elemental analysis of particulate matter on filter materials are instrumental neutron activation analysis (INAA), inductively-coupled plasma with mass spectroscopy (ICP-MS), inductively-coupled plasma with atomic emission spectroscopy (ICP-AES), particle-induced X-ray emission (PIXE), proton (or University of Ghana http://ugspace.ug.edu.gh 62 particle) elastic scattering analysis (PESA), total reflection X-ray fluorescence (TRXRF), synchrotron induced X-ray fluorescence (S-XRF) and scanning electron microscopy with X-ray fluorescence (SEM/XRF). Brief descriptions of analytical techniques used in the studies on which this thesis is based will follow in subsequent subsections. 3.4.1 GRAVIMETRIC ANALYSIS The integrated mass concentration of particulate matter in air can be measured in several ways. The most common and traditional way is through filter-based gravimetric methods. Gravimetric measurement is the net mass on a filter found by weighing the filter before and after sampling with a microbalance in a temperature and relative humidity controlled environment (Feeney et al., 1984). The principle of the gravimetric method is that air is drawn into the size selective inlet (SSI) of a sampler and through a pre-weighed Teflon filter by means of a low-flow volume electrically powered pump, so that particulate matter is collected quantitatively on the filter surface (US-EPA, 1999). The Harvard Impactor (HI) operates on a measurement principle based on gravimetric analysis which is similar to the US EPA’s Federal Reference Method (FRM) used for regulatory compliance. The HI collected particles on 37 mm Teflon filter over a 48-hour period. At the end of this period, the sampled filters were removed from the HI manually and placed into Petri dishes. Fresh filters were loaded into the impactor to start the next sampling period. Gravimetric analysis of the filters was performed with a microbalance. University of Ghana http://ugspace.ug.edu.gh 63 All filters used in this study were weighed before and after sampling on a Mettler Toledo MT-5 microbalance with a sensitivity of 1 µg located in a clean weighing room at the Harvard School of Public Health Laboratory, Boston, Massachusetts. The samples were allowed to stabilize and conditioned in the weighing room under controlled temperature (20.5 ± 0.2) ºC and relative humidity (39 ± 2) % for at least 24 hours prior to weighing. Residual charge on a Teflon filter could produce an electrostatic discharge between the filter on the pan and the metal casing of the microbalance (Engelbrecht et al., 1980; Allen et al., 1999; Koistinen et al., 1999; Hänninen et al., 2002). Gravimetric analysis is strongly affected by such charges. The electrostatic charges of the sample were eliminated via a polonium radioactive source by moving the filters near the source (Allen et al. 1999; Lawless and Rodes, 1999). In both pre- and post-weighing, filters were weighed twice; if these two masses were not within a precision 5 μg of one another, they were weighed a third time. This method was employed to improve the precision of the gravimetric analysis and to eliminate outliers due to large analytical errors. The mean of the two masses within 5 μg of one another was used for calculating concentrations. After every batch of 10 filters, the zero, span, and linearity of the balance were checked via a set of class “S” weights. Final filter weights were adjusted using an air buoyancy correction (Koistinen et al., 1999). In cases where duplicate measurements were taken, the two measurements were averaged so as to use all available data. All sample weights were corrected by subtracting the mean field blank weights. University of Ghana http://ugspace.ug.edu.gh 64 The total mass of the particulate matter collected on a Teflon filter in μ is given by equation 3.1. [( ) ] where, = total mass of particulate matter collected during sampling period (μ ) = final mass of conditioned filter after sample collection = initial mass of the conditioned filter before sample collection Equation 3.2 gives the total volume of ambient air passing through the sampler (V) in cubic meters. [( ) ] where = total sample volume (m3) = average flow rate over the entire duration of the sampling period (l/min) = duration of sampling period (min) Consequently, the PM concentration in μ can be calculated using equation 3.3. where = mass concentration of PM (μ = total mass of PM collected during sampling period μg = total volume of air sampled (m3) University of Ghana http://ugspace.ug.edu.gh 65 3.4.2 ENERGY DISPERSIVE X-RAY FLUORESCENCE ANALYSIS Energy dispersive X-ray fluorescence (EDXRF) spectrometry is one of the most commonly used analytical technique to determine the elemental composition of airborne particulate matter collected onto filter (Bandhu et al., 2000). It is capable of providing the composition of a variety of materials in a non-destructive manner. This technique is able to provide qualitative and quantitative information on multi-elemental concentration data of particulate matter. This data is essential for source identification work and for application of receptor models for source apportionment. A large number of studies on the elemental concentration of particulate matter using EDXRF spectrometry technique have been reported (Watson et al., 1999; Markowicz et al., 2002; Aboh et al., 2009). In EDXRF method, the sample on the filter is irradiated with a high intensity beam of X-rays. This causes electrons to be ejected from the inner shell orbital of atoms in the sample and the vacancies created are filled by electrons in the outer atomic shells. The excess energy is released in the form of (fluorescence) X-rays which is seen as a line in a spectrum whose energy is characteristic of the elements present. The intensity of the characteristic (fluorescent) X-rays is proportional to the concentration of the element in the sample. The X-ray fluorescence emitted from the irradiated sample is captured by the semiconductor detector and processed by the signal processing unit. University of Ghana http://ugspace.ug.edu.gh 66 Some advantages of EDXRF over other analytical techniques are:  It is non-destructive; therefore samples are left intact after analysis so they can be submitted for additional analyses by other methods as needed.  It requires little or no sample preparation or operator time after the samples are loaded into the analyser.  It is fast and can be used to simultaneously quantify the concentrations of elements with atomic numbers ranging from 11 (Na) to 92 (U). A basic X-ray spectrometer system consists of a source that generates X-rays, a solid state detector that measures the different energies of the characteristic radiation from the sample, a sample holder and a signal processing unit that records the emission or fluorescence energy signals and calculates the elemental concentrations in the sample. 3.4.2.1 ANALYSIS OF ELEMENTAL COMPOSITION The elemental concentrations of PM10 and PM2.5 samples from all monitoring sites in Accra were quantified by energy dispersive X-ray fluorescence spectrometry analysis (EDXRF) using a Shimadzu EDX-700HS spectrometer (Shimadzu Corp., Japan) at the Institute of Astronomy, Geophysics and Atmospheric Science, University of Sao Paulo, Brazil. The spectrometer used a low power Rh-target tube and operated with a voltage varying from 5 to 50 kV and a current from 1 to 1000 μA (Andrade et al., 2010). The characteristic X-ray radiation was detected by a Si (Li) detector. University of Ghana http://ugspace.ug.edu.gh 67 3.4.2.2 METHOD DETECTION LIMIT (DL) OF EDXRF The detection limit (DL) is usually defined as the amount of analytes that gives a net intensity peak equal to three times the standard deviation of the background intensity. In this work the MDL was calculated as three times the standard deviation of the field blanks. It depends upon the specific element and the sample matrix. An element was said to be detected if the following condition in equation 3.4 was satisfied: √ where Np = the number of counts measured on the peak Nb = the number of counts measured on the background. 3.5 TRAFFIC COUNT METHODOLOGY AND MEASUREMENTS Local variations in vehicular traffic volume were measured at the same intersection where air monitoring occurred using manual traffic count method at all sites (both residential and traffic) in the four Accra neighbourhoods. Manual traffic counts were conducted on 15-minute intervals with a 10-minute break after each interval for a 12- hour period (6:00 am to 6:00 pm) for two weeks (Monday to Sunday) on all roads at each monitoring site. Vehicular traffic counts were measured for each direction separately and the data were aggregated on a direction by direction basis. The vehicles were classified into 4 categories, namely: cars, buses/4WD, trucks, and University of Ghana http://ugspace.ug.edu.gh 68 motorbikes/scooters (Figure 3.5). Manual counts were used to provide detailed information about the vehicle categories and diurnal variation. Diurnal patterns of ambient PM concentrations were examined using the traffic and air quality data measured. Manual counts usually require trained observers to collect specific information that cannot be efficiently obtained through automated means. These may include vehicle occupancy, pedestrians, turning movements and vehicle classification. To conduct manual traffic count, tally sheets, mechanical count boards and electronic count boards are the most commonly used method. For this study, a manual traffic count using the tally sheet method was adopted because of the lack of access to a mechanical or electronic counting board and also because it is the least expensive tool for manual data collection. Manual traffic counting for this study was performed by two teams consisting of two observers in each team. Prior to commencement of the data collection, the observers were trained on how to conduct a manual traffic count using a specially prepared tally sheet, proper identification of traffic lanes and directions to be counted, and the identification of the four categories of vehicles to be counted. University of Ghana http://ugspace.ug.edu.gh 69 Cars - passenger cars, taxis, mini-vans Buses/4WD - trotros, school buses, large buses, 4WD vehicles (Jeeps, SUVs, and small pick-up trucks) Trucks - large trucks, multi-axle pick-up trucks Motorbikes/Scooters - scooters, motorbikes, zoomlions Figure 3.5 Classification of vehicles counted University of Ghana http://ugspace.ug.edu.gh 70 Pilot /trial runs were conducted after the training by the observers for a period of one week on selected roads within the various neighbours to test the counting methodology. Observers were positioned at a vantage point at each location away from the edge of the roadway so that their views were not blocked by trucks, buses, parked cars or signs and also for safety considerations. Counting of traffic was done simultaneously in the same direction by a team of two observers. The first team of two observers counted traffic from 6:00 am to 12:00 pm while the second team counted from 12:00 pm to 6:00 pm daily. Vehicle counts per unit time were determined visually at each road for each day of sampling. Traffic data from the tally sheets were entered onto a spreadsheet. These daily counts were then averaged across days at each site and expressed in counts per hour for each vehicle category. The use of a tally sheet involves the observer to make a tick mark for every vehicle in a given classification or movement. The tally sheet was prepared prior to going into the field allowing space for all information to be marked easily once the count began. The tally sheet was used in conjunction with a stopwatch to time the desire interval. Once these counts are collected, they are tallied and summarized. The average daily traffic (ADT) which is the sum of all vehicles that pass through a street during a day (unit = vehicles/day) averaged over the observed period was calculated. For this study, daily traffic flow data are based on daily average observations including all weekdays and weekends. University of Ghana http://ugspace.ug.edu.gh 71 3.6 ENRICHMENT FACTOR Enrichment factor (EF) analyses can be used to distinguish between elemental composition of aerosol particles originating from either anthropogenic activities or natural sources and to also evaluate the extent of anthropogenic influence (Duce et al., 1975; Zhang et al., 2008; Yongjie et al., 2009). EF compares the ratios for various elements in measured PM concentrations to the corresponding ratios in geological material as a means of confirming the man-made and natural contributions. Although there is no fixed rule for choosing the reference element, elements such as Aluminium (Al), Silicon (Si), Titanium (Ti), Iron (Fe) are most commonly used as reference element because they all have an abundant natural content in the soil, good chemical stability and less affected by anthropogenic pollution (Basha et al., 2010). In this study, Fe was used as reference, and the upper continental compositions of the earth’s crust were taken from Taylor and McLennan (1985). The enrichment factor (EF) for the each element in PM2.5 and PM10 at the four sampling sites was calculated using the following formula in equation 4.1: ( ) ( ) where X and E refer to the concentration (in µg m-3) of the element of interest X and the reference element respectively. University of Ghana http://ugspace.ug.edu.gh 72 The EF of an element is usually taken as being from a crustal or natural source if EF < 10. However, if EF > 10 then it is considered to indicate that a significant fraction of the element is contributed from non-crustal or anthropogenic source (Braga et al., 2005; Chimidza et al., 2001; Kothai et al., 2011). EF values >100 and 1000 are taken as highly and heavily enriched respectively. Table 3.5 shows the average of the crustal elements that were used in the study. Table 3.3 Averages of crustal elements used to calculate enrichment factors of PM at all sampling sites in Accra Atomic Number Element Crustal Average (ppm) Crustal average (μg/m3) 11 Na 23600 22.19 12 Mg 23300 23.16 13 Al 82300 90.82 14 Si 281500 323.36 15 P 1050 1.33 16 S 260 0.34 17 Cl 130 0.19 19 K 20900 33.42 20 Ca 41500 68.03 22 Ti 5700 11.16 23 V 135 0.28 24 Cr 100 0.21 25 Mn 950 2.1 26 Fe 56300 128.59 29 Cu 55 0.14 30 Zn 70 0.19 35 Br 2.5 0.0082 38 Sr 375 1.34 40 Zr 165 0.62 82 Pb 12.5 0.11 Source: (Taylor and McLennan, 1985) University of Ghana http://ugspace.ug.edu.gh 73 CHAPTER 4: RESULTS AND DISCUSSIONS 4.1 PM2.5 AND PM10 MASS CONCENTRATIONS Statistics for mass concentrations of PM2.5 and PM10 samples at five monitoring sites in Accra for the period September 1, 2007 – August 17, 2008 are shown in Table 4.1. The statistics include the average, standard deviation, minimum and maximum. The standard deviation of the concentration values represents the dispersion of the data around the average value (i.e. data variability during the investigated period). Overall, average PM2.5 and PM10 at the residential site, JT-R, had higher concentrations than other sites, 74.8 and 134.8 µg m-3, respectively. The next highest mass PM2.5 and PM10 concentrations were recorded at the only traffic site NM-T. JT-R site is located on a secondary road in James Town/Ussher Town with less traffic than NM-T. The higher mass concentration observed at the JT-R site may likely be associated with the widespread use of biomass fuels, especially firewood, for home and small-commercial cooking purposes in a very densely populated area (Dionisio et al., 2010). JT-R site is about 0.5 km from the James Town beach located along the coast of the Atlantic Ocean where fish smoking and goat roasting are carried out by burning wood and used rubber tires (Obiri-Danso et al., 2008). The lowest PM2.5 concentrations were obtained at the EL-R site. PM2.5 concentrations at this site were, on the average, 39% less than at JT-R. However, there were no significant differences observed on average PM10 levels among residential sites EL-R, AD-R and NM-R. This indicates that PM10 levels at these sites are influenced by similar sources in urban area. University of Ghana http://ugspace.ug.edu.gh 74 JT-R had slightly higher overall PM2.5/PM10 ratio (0.55) than the other sites (0.47-0.51). PM2.5/PM10 ratios > 0.5 have been observed in many urban areas in developed countries (US-EPA, 2002; Tecer et al., 2008). Qian et al. (2001) found PM2.5/PM10 ratios of 0.51- 0.72 during a study of four urban areas in China. The lower ratio in Accra could be attributable to coarse fraction contributions from re-suspended dust and geologic and marine sources which has higher coarse fraction (Arku et al., 2008). Table 4.1 Statistical summary of average PM10 and PM2.5 in Accra during entire sampling period. Average PM10 and PM2.5 concentrations in (µg m -3) PM2.5/ Sites Type n Average SD Min Max PM10 ratio AD-R PM10 44 108.2 121.2 29.4 655.4 0.47 PM2.5 46 50.4 64.1 14.2 399.9 EL-R PM10 46 96.9 118.1 22 581.3 0.47 PM2.5 46 45.9 60.2 10.7 298.73 JT-R PM10 38 134.8 121.6 52.9 652.6 0.55 PM2.5 39 74.8 61.6 31.3 320.4 NM-R PM10 51 93.9 108.98 27.3 569.6 0.51 PM2.5 51 48.3 56.8 16.3 319.8 NM-T PM10 57 111.9 108.9 41 593.44 0.51 PM2.5 55 57.5 60 22.4 348.4 n = Number of samples, SD =Standard deviation Figures 4.1(a) and (b) illustrates the temporary variations of daily mass concentration of PM2.5 and PM10 at the five sites during the investigation period. Peaks in the data suggest the influence of meteorology during the study period. It was observed that PM concentration peaked sharply in all neighbourhoods between December and February. University of Ghana http://ugspace.ug.edu.gh 75 This period corresponds to the seasonal Harmattan season when the northeast trade winds blow large amount of dust from the Sahara desert laden with emissions from dry- season bushfires towards the Gulf of Guinea in a south-westerly direction at an altitude of about 1500 m above sea level (Afeti and Resch, 2000). It is clear that the large PM concentrations measured during the few days of Harmattan period may distort information on the actual levels of aerosol concentrations. Aboh et al. (2009) separated total particulate matter concentrations and compositions data into seasonal Harmattan and non-Harmattan conditions to examine the extent to which the extreme conditions influence the identification of aerosol particle sources at a semi-rural site in Accra. Olic et al. (1999) identified fingerprints of individual sources by investigating aerosols particles using seasonal haze and non-haze conditions in Singapore. University of Ghana http://ugspace.ug.edu.gh 76 Figure 4.1 Time series plots of (a) PM2.5 and (b) PM10 mass concentrations during study period. (a) (b) University of Ghana http://ugspace.ug.edu.gh 77 4.1.1 NON-HARMATTAN CONDITIONS, PM2.5 AND PM10 FRACTIONS The arithmetic annual mean and standard deviation of PM2.5 and PM10 mass concentrations and elemental compositions at the five neighbourhood sites are presented in Table 4.2 and Table 4.3 for non-Harmattan period. The highest mean for both PM2.5 and PM10 mass concentrations during this period were obtained at JT-R site (49 and 82 µg m-3, respectively) which was closely followed by NM-T site (34 and 70 µg m-3, respectively). The results shows that during non-Harmattan conditions, PM2.5 and PM10 at neighbourhood sites ranged from 22 to 49 µg m-3 and 45 to 70 µg m-3 respectively. Currently Ghana has no standard limits for PM2.5 levels but the PM10 levels were just within the EPA-Ghana standard (70 µg m-3). However these levels are considerably higher than the World Health Organization (WHO) Air Quality Guidelines (AQG) of 10 µg m-3for PM2.5 and 20 µg m -3 for PM10. Moreover, in some cases these levels observed are even higher than the WHO Interim Target 1 (IT-1) of 35 µg m-3 (PM2.5) and 75 µg m-3(PM10). The high average aerosol particles values denoted a severe urban air pollutant problem in these neighbourhoods. However, the measured concentrations in this study are consistent with other studies in sub-Saharan African, such as PM10 concentrations of 40-100 µg m-3 at urban site of Addis Ababa, and PM2.5 and PM10 concentrations of 86 µg m-3 and 97 µg m-3, respectively, in South Africa (Engelbrecht et al., 2001; Etyemezian et al., 2005). These annual mean levels are also comparable to or slightly less than concentrations in large cities in South and East Asia and in the Eastern University of Ghana http://ugspace.ug.edu.gh 78 Mediterranean, but much higher than those in Latin America and high-income countries (Cohen et al., 2004; Dionisio et al., 2010). Table 4.2 Average concentration of total PM2.5 mass (µg m -3) and its elemental components (ng m-3) at five monitoring sites during the non-Harmattan months. Species AD-R (n =36) EL-R (n = 37) JT-R (n = 29) NM-R (n= 41) NM-T (n = 44) Mass SD Mass SD Mass SD Mass SD Mass SD Total mass 28 8 22 9 49 12 27 8 34 8 Na 244 216 221 175 456 349 271 246 181 134 Mg 103 87 144 100 166 86 132 87 90 96 Al 532 528 643 656 689 748 569 591 795 604 Si 1190 1356 1430 1597 1554 1882 1258 1457 1702 1447 P 15 7 14 9 18 6 13 7 18 8 S 662 256 639 229 833 270 662 230 662 225 Cl 247 274 145 115 1443 886 359 265 379 220 K 586 229 619 231 1841 410 843 250 930 218 Ca 242 181 247 210 380 276 282 186 451 218 Ti 33 31 49 34 42 46 36 35 54 37 V 1 1 1 1 1 1 1 1 1 1 Cr 2 1 1 1 1 1 1 1 1 1 Mn 7 4 9 4 8 6 6 5 8 5 Fe 365 280 412 334 400 381 361 291 537 304 Cu 5 4 2 2 5 3 5 4 5 5 Zn 37 15 18 18 44 16 28 12 32 10 Br 14 8 22 20 20 12 20 17 17 14 Sr 4 5 3 5 5 3 5 4 3 4 Zr 3 3 2 5 3 2 2 2 2 3 Pb 14 8 11 7 12 7 13 7 15 14 University of Ghana http://ugspace.ug.edu.gh 79 Table 4.3 Average concentration of total PM10 mass (µg m -3) and its elemental components (ng m-3) at five monitoring sites during the non-Harmattan months. Species AD-R (n = 33) EL-R (n = 36) JT-R (n = 28) NM-R (n = 41) NM-T (n = 45) Mass SD Mass SD Mass SD Mass SD Mass SD Total mass 56 19 45 19 82 23 51 20 70 21 Na 985 613 570 459 1966 831 687 500 762 526 Mg 287 182 248 201 455 220 248 200 308 196 Al 1654 1248 1834 1330 1720 1478 1679 1340 2591 1496 Si 4137 3260 4283 3290 4310 3793 4002 3316 5736 3454 P 26 10 22 10 30 16 24 11 41 15 S 862 280 695 232 1115 299 750 246 885 252 Cl 2031 930 1161 570 4930 1407 1696 651 2152 753 K 886 334 851 331 2215 515 1111 357 1318 335 Ca 1118 611 911 555 1642 779 1183 685 1920 825 Ti 126 87 129 91 130 101 124 92 203 112 V 2 1 3 1 2 1 2 2 3 2 Cr 3 1 3 1 4 2 3 2 5 2 Mn 20 12 17 13 21 15 17 13 26 14 Fe 1491 773 1463 790 1291 869 1346 848 2034 892 Cu 8 5 2 2 7 4 6 3 9 5 Zn 53 17 25 13 67 24 39 18 54 16 Br 20 18 28 24 23 17 25 28 30 33 Sr 4 4 7 6 6 4 7 5 6 5 Zr 2 2 5 4 5 3 5 4 4 4 Pb 19 18 20 14 16 11 17 12 15 7 The three most abundant PM10 crustal species are Si (5736 to 4002 ng m -3), Al (1654 to 2591 ng m-3), and Fe (1291 to 2034 ng m-3). PM10 calcium concentrations are low, in the range of 911 to 1920 ng m-3 at the four sites. PM2.5 crustal species are less than a quarter of the corresponding PM10 concentrations. This is consistent with the notion that a majority of crustal material is in the coarse particle size fraction. University of Ghana http://ugspace.ug.edu.gh 80 The PM2.5 and PM10 contributions from P, S, Cu, Zn and Pb did not exhibit significant variations between the sites. However, these elements in PM2.5 are emitted mainly by anthropogenic sources. JT-R had relatively higher sodium (Na) and chlorine (Cl) concentration. In comparison with the other sites, JT-R is the closest to the ocean; hence it received more particles from sea spray. JT-R is about 0.5 km from the ocean compared to approximately 3 km for AD-R, 4.5 km for NM and 9 km for EL-R. 4.1.2 HARMATTAN CONDITIONS, PM2.5 AND PM10 FRACTIONS The data of average concentrations of PM2.5 and PM10 mass (µg m -3) and their elemental components (ng m-3) at five monitoring sites during the Harmattan conditions are given in Tables 4.4 and Table 4.5, respectively. During the Harmattan period, which in Accra, Ghana started on 27th December 2007 and lasted till 7th February 2008, the average wind speed was 6.5 m s-1 (Dionisio et al., 2010). It is interesting to note that both PM2.5 and PM10 mass concentrations during the Harmattan period were found to be approximately 5 times higher than non-Harmattan period. Aboh et al. (2009) found large differences in mass and elemental levels between both the coarse and fine particles fraction during Harmattan as compared to non-Harmattan conditions. They observed similarly as found in this study that mass concentrations increased by more than five times during the Harmattan season. University of Ghana http://ugspace.ug.edu.gh 81 Table 4.4 Average concentration of total PM2.5 mass (µg m -3) and its elemental components (ng m-3) at five monitoring sites during the Harmattan months. Species AD-R (n = 10) EL-R (n = 9) JT-R (n = 10) NM-R (n = 10) NM-T (n = 11) Mass SD Mass SD Mass SD Mass SD Mass SD Total mass 132 104 143 83 149 86 134 86 152 84 Na 97 82 117 50 106 95 84 44 108 62 Mg 1004 685 1345 751 837 620 982 611 1162 653 Al 6338 4303 7902 4071 5649 3796 6217 3706 7257 3718 Si 16447 2793 21367 12231 14802 11058 15711 10450 18480 10979 P 25 11 24 13 32 19 27 4 33 15 S 1142 174 988 192 1251 165 1103 180 1160 197 Cl 528 499 660 603 1393 593 724 489 772 416 K 2698 1501 2799 1216 3891 1225 3125 1532 3271 1369 Ca 2387 2337 3239 2230 2384 1898 2487 1934 3023 1991 Ti 417 359 526 311 365 269 406 285 486 293 V 6 4 8 5 6 4 6 3 7 4 Cr 5 5 7 5 5 4 6 4 7 5 Mn 69 68 85 56 61 50 65 52 77 54 Fe 3727 3266 4736 2873 3301 2466 3686 2577 4320 2624 Cu 10 8 9 4 13 12 11 9 12 6 Zn 70 52 53 48 104 84 61 44 74 47 Br 34 39 64 107 58 33 48 32 37 35 Sr 33 29 22 23 25 14 60 41 34 30 Zr 21 16 15 12 17 9 37 25 23 21 Pb 24 12 24 19 32 19 36 24 26 22 University of Ghana http://ugspace.ug.edu.gh 82 Table 4.5 Average concentration of total PM10 mass (µg m -3) and its elemental components (ng m-3) at five monitoring sites during the Harmattan months. Species AD-R (n = 11) EL-R (n = 10) JT-R (n = 10) NM-R (n = 10) NM-T (n = 11) Mass SD Mass SD Mass SD Mass SD Mass SD Total mass 266 160 284 137 281 164 270 146 289 148 Na 273 229 173 136 546 426 126 113 171 134 Mg 1951 931 2271 957 1898 958 1891 847 2008 955 Al 12297 6050 14234 5918 11599 981 12708 5653 13404 5791 Si 33311 18762 37526 17629 31647 18606 33130 16550 34710 17791 P 38 10 35 16 64 26 56 32 65 27 S 1408 156 1198 118 1734 192 1333 166 1433 119 Cl 2093 777 1530 574 4625 1197 2199 597 2275 562 K 3898 1821 4150 1768 5198 1612 4568 2041 4528 1782 Ca 6485 4085 7154 4021 6949 4467 7172 4154 7825 4217 Ti 947 574 1099 577 909 591 991 546 1076 549 V 14 8 18 11 13 8 15 9 15 8 Cr 13 9 16 10 14 10 15 9 17 9 Mn 156 110 170 101 149 108 155 98 168 102 Fe 8733 5406 10121 5552 8119 5389 9333 5403 9769 5172 Cu 21 13 16 10 24 16 24 14 24 11 Zn 100 64 76 64 140 108 107 70 114 64 Br 44 35 51 43 35 27 100 135 48 36 Sr 42 22 25 26 28 24 54 44 28 21 Zr 28 13 15 12 20 19 33 28 18 11 Pb 39 19 32 19 29 17 46 27 34 26 University of Ghana http://ugspace.ug.edu.gh 83 4.2 PM2.5 TO PM10 RATIOS Table 4.6 and Table 4.7 presents annual average PM2.5 and PM10 mass concentrations with standard deviations and their ratios for the five sites during non-Harmattan and Harmattan conditions respectively. The annual average mass concentrations for PM2.5 and PM10 during non-Harmattan showed a 5-fold increment during Harmattan conditions. However, the ratios of PM2.5 to PM10 during non-Harmattan and Harmattan conditions showed only marginal variations. The average PM2.5/PM10 ratio observed in this study is much lower than those found by other studies. He et al. (2001) and Ye et al. (2003) reported an average PM2.5/PM10 ratio 0.64 and 0.55 for their entire period. Table 4.6 PM2.5 and PM10 annual average mass concentration [± SD (standard deviation)] for the five monitoring sites during non-Harmattan conditions. Urban sites Mean Annual Concentrations Ratio PM2.5/PM10 PM2.5 (µg m -3) PM10 (µg m -3) Mean SD Mean SD AD-R 28 8 56 19 0.50 EL-R 22 9 45 19 0.50 JT-R 49 12 82 23 0.55 NM-R 27 8 51 20 0.54 NM-T 34 8 70 21 0.49 University of Ghana http://ugspace.ug.edu.gh 84 Table 4.7 PM2.5 and PM10 annual mean mass concentration [± SD (standard deviation)] for the five monitoring sites during Harmattan conditions. Urban sites Mean Annual Concentrations Ratio PM2.5/PM10 PM2.5 (µg m -3) PM10 (µg m -3) Mean SD Mean SD AD-R 132 104 266 160 0.50 EL-R 143 83 284 137 0.50 JT-R 149 86 281 162 0.53 NM-R 134 86 270 146 0.50 NM-T 152 84 289 148 0.50 It was noted that during the non-Harmattan season between 49% and 55% of PM10 mass was in the fine of PM2.5 fraction at different sites. However, 50% to 53% of PM10 mass was in fine fraction during the Harmattan season. The average PM2.5/PM10 ratio of elemental components at five monitoring sites during the non-Harmattan period is given in Table 4.8. The quantity of mass in PM2.5 fraction was less than 40% for crustal elements such as Al, Si, Ca, Ti and Mn. A similar trend was noted for Na and Cl which are principal constituents of sea salt. This trend indicates that these elements were mainly in the form of coarse particles. Other elements such as K, Cu, Zn, Br and Pb had 60% or more of their mass in the fine fraction. These elements are associated with anthropogenic sources such as biomass burning, solid waste burning, motor vehicles emissions and industrial emissions. University of Ghana http://ugspace.ug.edu.gh 85 Table 4.8 Average PM2.5/PM10 ratio of PM elemental components at five monitoring sites during the non-Harmattan period. Species AD-R EL-R JT-R NM-R NM-T Na 0.25 0.39 0.23 0.39 0.24 Mg 0.36 0.58 0.36 0.53 0.29 Al 0.32 0.35 0.39 0.34 0.31 Si 0.29 0.33 0.36 0.31 0.30 P 0.59 0.66 0.59 0.56 0.43 S 0.77 0.92 0.75 0.83 0.75 Cl 0.12 0.13 0.29 0.21 0.18 K 0.66 0.73 0.83 0.76 0.71 Ca 0.22 0.27 0.23 0.24 0.23 Ti 0.26 0.38 0.33 0.29 0.27 V 0.59 0.35 0.54 0.38 0.33 Cr 0.51 0.39 0.39 0.35 0.29 Mn 0.35 0.56 0.38 0.32 0.32 Fe 0.24 0.28 0.31 0.27 0.26 Cu 0.55 0.79 0.64 0.84 0.55 Zn 0.70 0.70 0.66 0.70 0.59 Br 0.71 0.77 0.88 0.82 0.56 Sr 1.22 0.47 0.84 0.67 0.49 Zr 1.57 0.53 0.67 0.42 0.45 Pb 0.75 0.56 0.74 0.77 0.97 University of Ghana http://ugspace.ug.edu.gh 86 4.3 METEOROLOGICAL INFLUENCES ON THE CONCENTRATION OF PARTICULATE MATTER MASS 4.3.1 INFLUENCE OF METEOROLOGY ON PM MASS Meteorology plays a very important role in the distribution of aerosol particles. Several studies have shown that particulate matter mass concentrations are influenced by various meteorological parameters such as precipitation, temperature, wind speed and relative humidity (Karar et al., 2005; Bhaskar et al., 2010). These meteorological variables also affect the emissions of aerosol from ground surface, their residence time in the ambient air and the formation of secondary pollutants. Hence there is a need to understand the physical processes leading to an observed concentration of PM at a given location. Precipitation is one of the reasons for low aerosol particles in the non- Harmattan season as the particles are washed out by rain. Wet deposition by precipitation or wet removal is one of the major mechanisms for removal of aerosols from the atmosphere (Bhaskar et al., 2010). Daily average of temperature, wind speed, cumulative precipitation and relative humidity were obtained from the Ghana Meteorological Authority and are summarized in Table 2.1 of Chapter 2. The daily average wind speed ranged from 3 to 12 m s-1 while daily average temperature ranged from 23 °C to 30 °C during the study period. The daily mean relative humidity also ranged from 18% to 92 %, with the lower value usually recorded in the Harmattan season. The cumulative precipitation during the study period varied from zero during the Harmattan season to 62.8 mm during the non-Harmattan season. University of Ghana http://ugspace.ug.edu.gh 87 The correlations of the different meteorological parameters with PM2.5 mass concentrations measured during the period of investigation were examined. 4.3.1.1 RELATIONSHIP WITH PRECIPITATION The PM2.5 mass concentration was consistently low during precipitation occurrences (Figure 4.2). The best line obtained to explain the relationship between PM2.5 mass concentrations and rainfall was linear. PM2.5 mass concentrations and rainfall were negatively correlated but it is not significant. It was found that rainfall explained less than10% of the variance in PM2.5 concentration for the overall period of monitoring time. Figure 4.2 Relationship between PM2.5 and total daily cumulative precipitation. y = -0.2566x + 46.243 R² = 0.0023 0 30 60 90 120 150 0 10 20 30 40 50 60 70 P M 2 .5 co n c (µ g m -3 ) Precipitation (mm) University of Ghana http://ugspace.ug.edu.gh 88 4.3.1.2 RELATIONSHIP WITH TEMPERATURE The variation of PM2.5 mass concentration with temperature is shown in Figure 4.3. It was observed that concentration was high at higher temperature. PM2.5 mass concentration was positively correlated with temperature but this correlation is weak. It means that PM2.5 increases or reduces as temp increases or reduces. The influence of temperature over PM is attributed to the fact that more favourable atmospheric dispersion conditions occur under a warm air than cold air mass (Owoade et al., 2012). Figure 4.3 Relationship between PM2.5 and total daily average temperature. y = 4.3989x - 76.162 R² = 0.0142 0 50 100 150 200 20 22 24 26 28 30 P M 2 .5 c o n c. ( µ g m -3 ) Temperature (°C) University of Ghana http://ugspace.ug.edu.gh 89 4.3.1.3 RELATIONSHIP WITH WIND SPEED Wind speed had a negative relationship with PM2.5 (Figure 4.4). Karar and Gupta (2006) suggest that such an inverse relationship between PM2.5 and wind speed indicates a dominance of local sources. Wind speed plays a leading role in cleansing the atmosphere of PM. Strong winds disperse particulate matter out of the atmosphere, whereas low winds allow particles levels to increase. Figure 4.4 Relationship between PM2.5 mass concentrations and daily wind speed. y = -1.2345x + 55.466 R² = 0.0022 0 50 100 150 200 2 4 6 8 10 12 P M 2 .5 c o n c. ( µ g m -3 ) Wind speed (m s-1) University of Ghana http://ugspace.ug.edu.gh 90 4.3.1.4 RELATIONSHIP WITH RELATIVE HUMIDITY Relative humidity showed an inverse relationship with PM2.5 mass concentration. Relative humidity factor influences particles to gather mass and settle down on the ground rather than getting airborne. Hence, relative humidity largely removes pollutants from the atmosphere. Figure 4.5 Relationship between PM2.5 mass concentrations and relative humidity. y = -0.4328x + 78.602 R² = 0.0068 0 50 100 150 20 40 60 80 100 P M 2 .5 c o n c. ( µ g m -3 ) Relative humidity (%) University of Ghana http://ugspace.ug.edu.gh 91 4.4 ENRICHMENT FACTOR ANALYSIS Enrichment factor (EF) calculations were employed as a tool for interpretation of tracer selection in each source category especially for trace element contributions. Enrichment factor analysis was used to provide an initial indication on the extent of the anthropogenic contributions to the measured elemental composition of PM2.5 and PM10 at the five urban sites in Accra. Average concentrations of all elements detected in PM samples were used for the calculations of EFs. By convention, elements with EF < 10 is usually taken as being from a crustal or natural source, while EF > 10 is usually considered to indicate that a significant fraction of the element is contributed from non- crustal or anthropogenic source (Braga et al., 2005; Chimidza et al., 2001; Kothai et al., 2011). EF values >100 and 1000 could be seen as highly and heavily enriched respectively. Enrichment factors were calculated according to equation 3.5 and aluminium was used as the reference element (see section 3.6 of Chapter 3.0). Results of enrichment factor analysis for each of the five sampling sites (AD-R, EL-R, JT-R, NM-R and NM-T) PM2.5 elemental concentrations are presented in Table 4.9. University of Ghana http://ugspace.ug.edu.gh 92 Table 4.9: Calculated enrichment factor (EFs) (using Al as reference) for all analysed elements in PM2.5. EFs were calculated using the average concentration of all samples from each site (see equation 3.5). Crustal concentrations were taken from Taylor and McLennan (1985). AD-R EL-R JT-R NM-R NM-T Na 1.9 1.4 2.7 1.9 0.9 Mg 0.8 0.9 0.9 0.9 0.4 Al 1 1 1 1 1 Si 0.6 0.6 0.6 0.6 0.6 P 2.0 1.5 1.8 1.6 1.5 S 331.5 264.6 322.2 291.1 221.6 Cl 223.2 108.8 1009.2 303.8 229.6 K 3.0 2.6 7.3 4.0 3.2 Ca 0.6 0.5 0.7 0.7 0.8 Ti 0.5 0.6 0.5 0.5 0.6 V 0.8 0.5 0.5 0.5 0.4 Cr 1.3 0.8 0.9 0.8 0.8 Mn 0.5 0.6 0.5 0.4 0.4 Fe 0.5 0.5 0.4 0.4 0.5 Cu 5.6 1.9 4.2 5.9 4.0 Zn 33.8 13.3 31.1 23.6 19.6 Br 302.8 373.9 321.5 400.2 236.8 Sr 0.6 0.4 0.5 0.6 0.2 Zr 1.0 0.6 0.7 0.5 0.4 Pb 23.0 15.0 14.8 20.3 15.8 Figure 4.6 shows the averages of the EFs of elements in PM2.5 for all sampling sites in Accra. In the category of EF<10 for PM2.5 are Na, Mg, Al, Si, P, K, Ca, Ti, V, Cr, Mn, Fe, Cu, Sr and Zr at all sampling sites. The EF results displayed almost similar values at each sampling site. These elements are typical of crustal origin and they were attributable predominantly to crustal or soil dust. Although these elements were mainly University of Ghana http://ugspace.ug.edu.gh 93 from natural sources, the influences of anthropogenic sources cannot be ignored. K is emitted from wood and biomass burning; Ca, a component of some detergents used in engines lubricating oils, is emitted from motor vehicles; Sr is emitted from vehicle brake wear; Fe is emitted as from wear and tear of brake pads and other automobile parts (Schauer, 2006). In the PM2.5 fraction Mn and Cu are also attributed to brake wear and vehicle tailpipe emission (Cadle et al., 1999; Garg et al., 2000). Among the various sites NM-R represented higher EFs value for Cu, i.e., 5.9, while the lowest was shown by EL-R with value of 1.9. In the group of EF>10 but less than 100 were the elements Zn and Pb at all sampling sites. It indicated that Pb and Zn were moderately enriched. Pb and Zn are traditional tracers of vehicle emissions and their EF value suggested a large contribution from vehicle emission. Besides fuel and motor oil combustion, brake wear, industry emission, long-range transported dust and the re-suspended soil containing the deposition of those from previously emitted leaded gasoline could be the important sources of Pb (Miguel et al., 1997; Yele et al., 2006). It has been reported (Miguel et al., 1997) that Zn compounds have been used extensively as antioxidants (e.g., zinc carboxylate complexes and zinc sulphonates) and as detergent or dispersant improvers for lubricating oils. University of Ghana http://ugspace.ug.edu.gh 94 Figure 4.6 Enrichment factors for PM2.5 chemical components for all sites in Accra. Mechanical abrasion of vehicles, metal processing industrial activities, and the wear and tear of vulcanized vehicle tires can contribute significantly to Zn of urban dust (Ellis and Revitt, 1982; Jiries et al., 2001; Li et al, 2001). Elements with EF>100 were S, Cl and Br. These elements are highly enriched with emission sources likely from traffic and refuse burning, respectively. The S concentration is important because it includes the sulphates and most toxic trace metals exist in the atmosphere in the form of sulphates and/or nitrates. In the group with EF>1000 was only the element Cl at the 0 1 10 100 1000 N a M g A l Si P S C l K C a Ti V C r M n Fe C u Zn B r Sr Zr P b En ri ch m en t Fa ct o rs Elements Enrichment factors for PM2.5 AD-R EL-R JT-R NM-R NM-T University of Ghana http://ugspace.ug.edu.gh 95 JT-R site. The heavily enriched Cl at the JT-R is not surprising since the JT-R site was the closest to the Atlantic Ocean. Cl is mainly derived from sea salt. The average EFs for PM10 of the various elements at all sampling locations are shown in Table 4.10. In the PM10 fraction, Na, Mg, Al, Si, P, K, Ca, Ti, V, Cr, Mn, Fe, Cu, Sr, Zr, Pb (all sites) and Zn (EL-R and NM-T) all had their enrichment factors below 10 (Figure 4.7). For these elements, their predominant source could be classified as soil dust (or other crustal matter, such as road dust). University of Ghana http://ugspace.ug.edu.gh 96 Table 4.10: Calculated enrichment factor (EFs) (using Al as reference) for all analysed elements in PM10. EFs were calculated using the average concentration of all samples from each site (see equation 3.5). Crustal concentrations were taken from Taylor and McLennan (1985). AD-R EL-R JT-R NM-R NM-T Na 2.4 1.3 4.7 1.7 1.2 Mg 0.7 0.5 1.0 0.6 0.5 Al 1 1 1 1 1 Si 0.7 0.7 0.7 0.7 0.6 P 1.1 0.8 1.2 1.0 1.1 S 138.8 100.9 172.6 119.0 91.0 Cl 591.4 304.9 1380.8 486.6 400.3 K 1.5 1.3 3.5 1.8 1.4 Ca 0.9 0.7 1.3 0.9 1.0 Ti 0.6 0.6 0.6 0.6 0.6 V 0.4 0.5 0.4 0.4 0.4 Cr 0.8 0.7 0.9 0.8 0.8 Mn 0.5 0.4 0.5 0.4 0.4 Fe 0.6 0.6 0.5 0.6 0.6 Cu 3.3 0.9 2.7 2.4 2.2 Zn 15.5 6.6 18.9 11.4 10.1 Br 137.6 171.4 146.8 165.6 128.8 Sr 0.2 0.3 0.2 0.3 0.2 Zr 0.2 0.4 0.4 0.4 0.2 Pb 9.8 9.4 8.0 8.9 5.0 University of Ghana http://ugspace.ug.edu.gh 97 Figure 4.7 Enrichment factors for PM10 chemical components for all sites in Accra. In the group of moderately enriched elements with EF>10 but less than 100 were Zn (AD-R, JT-R and NM-R) and S (EL-R). This gives an indication that these elements have contributions from anthropogenic sources. Zn is emitted from vehicle tire wear, non-ferrous industry emissions as well refuse burning (Schauer et al., 1996; Bullock et al., 2008). Elements in the highly enriched group with EF>100 are S (AD-R, JT-R and NM-R), and Cl (AD-R, EL-R, NM-R and NM-T) and Br (AD-R, EL-R, JT-R, NM-R and NMN-T). In the category of heavily enriched element with EF>1000 is Cl in JT-R, which is similar to the PM2.5 results. 0 1 10 100 1000 10000 N a M g A l Si P S C l K C a Ti V C r M n Fe C u Zn B r Sr Zr P b En ri ch m en t Fa ct o rs Elements Enrichment factors for PM10 AD-R EL-R JT-R NM-R NM-T University of Ghana http://ugspace.ug.edu.gh 98 4.5 SOURCE IDENTIFICATION AND APPORTIONMENT USING POSITIVE MATRIX FACTORIZATION Source identification and apportionment of PM mass by PMF was performed using the EPAPMF version 3.0 modelling software in accordance with the User’s Guide (USEPA, 2008). Signal-to-noise ratio (S/N) of an individual species can have a significant influence on a receptor model. S/N ratio shows whether the variability in measured concentrations is real or within the noise of the data. Hence, Paatero and Hopke (2003) strongly recommend that species with S/N ratio <0.2 be categorized as a “Bad” species that should be excluded from the analysis, 0.2 < S/N < 2.0 as “Weak” that can be down-weighted (i.e., its uncertainty increased), and S/N > 2.0 as “Strong” species that should be included in the analysis. The S/N ratios for the measured species at each of the 5 sites are presented in Table 4.11 for PM2.5 and Table 4.12 for PM10 respectively. Concentrations of species below MDL were replaced by half of the MDL, and the corresponding uncertainty was set as 5/6 of MDL. Missing concentration was replaced by the geometric mean of the measured concentration, and its uncertainty was set to 4 times this geometric mean value. The measured PM mass concentrations were included as an independent variable in the PMF model analysis so as to obtain the mass apportionment directly without the usual multilinear regression. The PM mass values were downweighted in the model analysis by setting the uncertainties to 4 times the mass concentrations. University of Ghana http://ugspace.ug.edu.gh 99 Table 4.11 S/N values for the PM2.5 data at AD-R, EL-R, JT-R, NM-R and NM-T Species AD-R EL-R JT-R NM-R NM-T Na 2.67 2.30 4.20 3.00 1.77 Mg 2.56 3.16 3.20 2.82 2.58 Al 16.68 16.95 17.05 16.81 17.03 Si 18.96 18.97 18.98 18.97 18.98 P 4.61 4.58 5.15 4.05 5.35 S 18.97 18.97 18.98 18.97 18.97 Cl 17.48 17.36 17.52 17.49 17.49 K 18.99 18.99 19.00 19.00 19.00 Ca 18.98 18.98 18.99 18.98 18.99 Ti 18.12 18.47 18.51 18.25 18.56 V 2.36 1.06 1.15 0.91 1.06 Cr 2.03 1.39 1.75 1.36 1.70 Mn 9.14 10.80 10.68 8.59 10.37 Fe 18.99 18.99 18.99 18.99 18.99 Cu 6.70 3.28 5.80 7.07 7.38 Zn 17.66 16.03 18.00 16.83 17.19 Br 13.08 16.33 15.20 15.95 14.91 Sr 5.62 5.61 5.10 5.04 4.36 Zr 3.91 5.57 3.53 2.85 3.32 Pb 6.89 5.50 5.74 6.43 8.44 The initial results were then normalized by the apportionment of PM mass concentrations so that the quantitative source contributions for each source were obtained. The number of species (including particulate mass concentrations) and the number of valid samples used in the PMF analysis at the five sites are shown in Table 4.13. The choice of valid sample depends on the quality of available species measured. University of Ghana http://ugspace.ug.edu.gh 100 Table 4.12 S/N values for the PM10 data at AD-R, EL-R, JT-R, NM-R and NM-T Species AD-R EL-R JT-R NM-R NM-T Na 6.03 4.92 6.87 5.32 5.55 Mg 5.03 4.87 6.00 4.87 5.23 Al 17.40 17.42 17.42 17.41 17.46 Si 19.00 19.00 19.00 19.00 19.00 P 7.60 6.56 9.11 7.16 10.98 S 18.98 18.97 18.99 18.98 18.98 Cl 17.52 17.52 17.52 17.52 17.52 K 19.00 19.00 19.00 19.00 19.00 Ca 19.00 19.00 19.00 19.00 19.00 Ti 18.92 18.92 18.93 18.92 18.96 V 2.96 3.25 2.56 3.10 3.95 Cr 4.00 3.87 4.40 4.08 5.94 Mn 16.11 15.59 16.54 15.78 17.10 Fe 19.00 19.00 19.00 19.00 19.00 Cu 9.70 3.21 8.68 7.43 10.27 Zn 18.27 16.54 18.55 17.84 18.29 Br 16.06 17.25 16.24 17.27 17.70 Sr 5.03 7.83 5.87 7.19 6.45 Zr 2.83 5.13 4.89 5.52 5.00 Pb 10.24 9.62 7.97 8.56 6.89 Table 4.13 Dataset of number of valid samples and species used in the PMF model. Sites AD-R EL-R JT-R NM-R NM-T PM size fraction PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 Number of valid samples 36 33 37 36 29 28 41 41 44 45 Number of species 21 21 21 21 21 21 21 21 21 21 University of Ghana http://ugspace.ug.edu.gh 101 4.5.1 NON-HARMATTAN SEASON PMF model applied to the non-Harmattan months’ data for both PM2.5 and PM10 identified six potential factors (sources) for AD-R and five potential factors (sources) for EL-R, JT-R, NM-R and NM-T. Although many factors were similar, their importance varied based on the differing composition among sites. The potential sources were found based on several methods including the closeness of Q values to the theoretical Q value, FPEAK value of 0.1, comparison of the measured total mass of the sample to the total mass predicted by the model, the observation that the scaled residuals for each element were fairly normally distributed between 3 standard deviations (Amato, 2009), and by checking the results to see whether they are realistic for the area around the sites. 4.5.1.1 FINE PARTICULATE MATTER The reconstructed PM2.5 mass concentrations were estimated from the sum of the contributions of the PMF resolved sources. The scatter plots to compare the reconstructed PM2.5 mass concentrations using predicted results from the model and the measured PM2.5 mass concentrations are shown in Figure 4.8 to Figure 4.12. The high values of r2 (all greater than 0.73) indicate that resolved sources effectively accounted for most of the variation in PM2.5 mass concentrations at the five sites. The performance of PMF model was also evaluated using Table A-1 to Table A-5 in Appendix A by comparing the calculated values with measured concentrations for each species used in the model. The largest uncertainties such as (22.7% for Mg, 18.9% for V, 23% for Cr, University of Ghana http://ugspace.ug.edu.gh 102 19.7% for Sr, 18% for Zr in AD-R), (20.02% for Cr, 28.82% for Cu, 30.34% for Zn, 27.29 for Sr, 44.52% for Zr in EL-R), (19.77% for Na, 20.76% for Mg, 20.57% for Cr, 22.65% for Sr, 29.77% for Zr in NM-R), and (26.19% for Sr, 19.57% for Zr in NM-T) together with low correlation coefficients, can be attributed to data quality, since species with missing values were replaced by their median values and associated uncertainties assigned to be four times the mean values. However, in general, good correlation coefficient was obtained between measured and calculated concentrations with significant correlation coefficients. Overall, the performance of PMF on the source apportionment of PM2.5 at the five study sites was found to be satisfactory. The average source contributions are summarized in Table 4.14. Figure 4.8 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at AD-R. y = 1.0272x - 1.1762 r² = 0.8874 0 10 20 30 40 50 60 0 10 20 30 40 50 60 P re d ic te d P M 2 .5 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM2.5 Mass Concentration (µg m -3) Asylum Down (AD-R) University of Ghana http://ugspace.ug.edu.gh 103 Figure 4.9 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at EL-R. Figure 4.10 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at JT-R. y = 1.0733x -1.9459 r² = 0.9114 0 10 20 30 40 50 0 10 20 30 40 50 P re d ic te d P M 2 .5 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM2.5 Mass Concentration (µg m -3) East Legon (EL-R) y = 1.0293x - 1.9079 r² = 0.8769 0 10 20 30 40 50 60 70 80 90 0 20 40 60 80 100 P re d ic te d P M 2 .5 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM2.5 Mass Concentration (µg m -3) James Town/Ussher Town (JT-R) University of Ghana http://ugspace.ug.edu.gh 104 Figure 4.11 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at NM-R. Figure 4.12 Scatter plots of predicted PM2.5 mass concentrations and measured PM2.5 mass concentrations at NM-T. y = 0.8868x + 2.5324 r² = 0.7344 0 10 20 30 40 50 60 0 10 20 30 40 50 60 P re d ic te d P M 2 .5 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM2.5 Mass Concentration (µg m -3) Nima Residence (NM-R) y = 1.0737x - 3.3003 r² = 0.7882 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 P re d ic te d P M 2 .5 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM2.5 Mass Concentration (µg m -3) Nima Traffic (NM-T) University of Ghana http://ugspace.ug.edu.gh 105 Table 4.14 Average contributions of identified sources to PM2.5 concentrations. Source 1: A source with high loading mass fractions for sodium (Na), chlorine (Cl) and Sulphur (S) was identified by PMF model at all five sites with the highest contribution at JT-R. The source profiles are shown in Figure 4.13 and the source contributions shown in Figure 4.14. Source profile with dominant Na and Cl as identified in various source apportionment studies conducted in coastal areas was classified as sea salt (Wu et al., 2007; Guo et al., 2009). Sea salt aerosols are also sources of sulphur (Watson et al., 2004). The geographical location of the Accra along the coast of the Atlantic Ocean makes the area susceptible to large-scale influence of sea salt aerosols. The distances of the monitoring sites to the ocean coast were: JT-R 0.5 km, AD-R 3 km, NM 4.5 km, and EL-R 9 km. Thus, source 1 was classified as “sea salt”. AD-R EL-R JT-R NM-R NM-T Sea Salt 3.87% (1.05 μg m-3) 11.45% (2.51 μg m-3) 18.86% (9.21 μg m-3) 3.13% (0.84 μg m-3) 11.62% (3.86 μg m-3) Biomass burning 18.85% (5.14 μg m-3) 38.76% (8.50 μg m-3) 29.41% (14.36 μg m-3) 41.89% (11.22 μg m-3) 37.68% (12.52 μg m-3) Solid waste burning 11.98% (3.27 μg m-3) 5.38% (1.18 μg m-3) 21.49% (10.50 μg m-3) 13.23% (3.54 μg m-3) 0.75% (0.25 μg m-3) Soil Dust 23.18% (6.33 μg m-3) 44.12% (9.67 μg m-3) 15.25% (7.45 μg m-3) 28.94% (7.75 μg m-3) 34.73% (11.54 μg m-3) Re-suspended dust 31.26% (8.53 μg m-3) 0.29% (0.06 μg m-3) 15.01% (7.33 μg m-3) Traffic/Industry 10.86% (2.96 μg m-3) 12.81% (3.43 μg m-3) 15.22% (5.06 μg m-3) University of Ghana http://ugspace.ug.edu.gh 106 Figure 4.13 PM2.5 source profiles for sea salt factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Sea Salt C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 107 Figure 4.14 Time series of PM2.5 source contributions for sea salt factors. University of Ghana http://ugspace.ug.edu.gh 108 The sea salt contributions to PM2.5 were 3.9, 11.5, 18.9, 3.1 and 11.6% at AD-R, EL-R, JT-R, NM-R, and NM-T, respectively. Present in this source are 87, 93, 97, 50 and 50% of total Cl and 36, 34, 91, 100, 100% of total Na at AD-R, EL-R, JT-R, NM-R, and NM-T, respectively. Time series plot shows higher concentrations in November and August. The highest contribution of the source occurred in August. In order to understand the origins of the air masses arriving in Accra on 17 August 2008 (the day with the highest contribution of sea salt), five-day backward air masses trajectories were calculated using the NOAA HYSPLIT model. These trajectories were computed at several altitudes (500, 1500 and 3000 m a.g.l.). Figure 4.15 reveals that air masses reached Accra from the Atlantic Ocean. This shows that probably the sea salt influence detected in Accra is due to the transport of marine aerosols from the ocean during the dry season. University of Ghana http://ugspace.ug.edu.gh 109 Figure 4.15 Five-day HYSPLIT backward trajectories of air masses reaching Accra on 17 August 2008. Source 2: The second source with high loading for potassium (K), sulphur (S) and phosphorus (P) was identified at all sites. The source profiles are shown in Figure 4.16 and the time series of contributions are presented in Figure 4.17. It is well known that K is a unique tracer of a biomass burning source (Song et al., 2006). The source contributions to PM2.5 mass were about 19, 39, 29, 42 and 38% at AD-R, EL-R, JT-R, University of Ghana http://ugspace.ug.edu.gh 110 NM-R, and NM-T, respectively. However in absolute terms, biomass combustion contributed more particle pollution in James Town/Ussher Town and Nima, where the density of households who use biomass fuels is significantly higher, than in Asylum Down and East Legon (Zhou et al., 2013). Therefore, biomass burning accounted for 14.4 µg m-3 of PM2.5 mass at JT-R, 11.2 µg m -3 at NM-R and 12.5 µg m-3 at NM-T as compared to 5.14 µg m-3 at AD-R and 8.5 µg m-3 at EL-R (Table 4.14). James Town/Ussher Town and Nima are urban slums where many residents use charcoal and especially wood for cooking and heating in their homes. Others use these biomass fuels for small-commercial cooking within the neighbourhoods. For example in James Town/Ussher Town, fish smoking and goat roasting are done by burning wood or old rubber tires (Obiri-Danso et al., 2008, Dionisio et al., 2010, Adam et al., 2013). Trash/wood/grass burning is common in many places. Biomass burning is a major component of PM2.5 in Accra. The relatively high source contributions (Figure 4.17) in August and December may be due to the usual bushfires that occur during the dry season. University of Ghana http://ugspace.ug.edu.gh 111 Figure 4.16 PM2.5 source profiles for biomass factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Biomass Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Biomass Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Biomass Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Biomass Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Biomass Burning C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 112 Figure 4.17 Time series of PM2.5 source contributions for biomass burning factors. University of Ghana http://ugspace.ug.edu.gh 113 Source 3: A source with large contributions from bromine (Br) and lead (Pb) was identified by PMF model at all five sites. Minor amounts of other crustal elements were also observed. The third source was attributed to solid waste burning based on the high loadings of Br with Pb. The source profiles and source contributions are shown in Figure 4.18 and Figure 4.19, respectively. The source contributions do not show seasonal variations. Pb and Br have been linked to incineration and/or traffic in several studies. Solid waste burning is characterized by Br (Vehlow et al., 2003). Br is used as the building block for some of the most effective flame retarding agents available to the plastics industry today. Moreover, burning of waste electrical and electronic equipment (WEEE) plastic have been found result in the formation of toxic brominated by-product in most combustion systems (Nnorom and Osibanjo, 2008). The use of leaded petrol in Ghana has been phased out since 2004. The remaining Pb in airborne particulate matter may therefore be a result of resuspended dust, residual lead in petrol, and oil and solid waste burning (Hosiokangas et al., 1995; Arku et al., 2008). On average, the solid waste burning source accounted for 12%, 5%, 22%, 42% and 4% of the total PM2.5 mass measured at AD-R, EL-R, JT-R, NM-R and NM-T, respectively. This higher mass concentration of this source in James Town/Ussher Town and Nima may be due to small-scale manufacturers of aluminium cooking pots in during the course of smelting add Pb from dry cell batteries to lower the melting point of the aluminium in order to cast the pots. It was observed that the JT-R site had about 3 times more solid waste burning contributions than the NM-R site (Table 4.14). University of Ghana http://ugspace.ug.edu.gh 114 Figure 4.18 PM2.5 source profiles for solid waste burning factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Solid Waste Burning C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 115 Figure 4.19 Time series of PM2.5 source contributions for solid waste burning factors. University of Ghana http://ugspace.ug.edu.gh 116 Source 4: Figure 4.20 shows source profiles of the fourth source which was characterized with large amounts of Al, Si, Ca, Ti, and Fe, suggesting crustal or soil dust aerosols. Sources with similar chemical composition resolved by other source apportionment studies were classified as resuspended soil dust (Gatari et al., 2005), crustal (Aboh et al., 2009) or soil dust (Ofosu et al., 2012). The soil-derived elements Al, Si, Ca, Ti, and Fe also exhibited variable concentrations at the various sites, signifying localized sources. Soil dust contributed 23, 44, 15, 29 and 35% to the PM2.5 concentrations at the AD-R, EL-R, JT-R, NM-R and NM-T monitoring sites, respectively. Soil dust sources had a larger proportion of total particle mass at EL-R, the northernmost neighbourhood site. Despite having a larger share from the soil dust source, lower total particle mass at the EL-R site meant that the absolute contribution of soil dust particles was only slightly higher than other neighbourhoods, e.g., 9.7 μg m-3 in PM2.5 compared to 7.8 μg m -3 at NM-R. The higher contribution may have been because more spread-out low-rise homes do not block the windblown local dust. Soil dust was a major source of airborne particulate matter identified at all the sites. The only roadside site monitoring site in this study, NM-T, had the highest mass concentration (11.5 μg m-3). However, small fractions of other elements, such as P, K, V, Cr, Mn, Zn, and Pb, were also associated with this source at NM-T, probably because of the resuspension of urban soil containing these elements. The source contributions (Figure 4.21) show high peaks in the dry season when rainfall is at minimum. University of Ghana http://ugspace.ug.edu.gh 117 Figure 4.20 PM2.5 source profiles for soil dust factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Soil Dust C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 118 Figure 4.21 Time series of PM2.5 source contributions for soil dust factors. University of Ghana http://ugspace.ug.edu.gh 119 Source 5: A source with significant concentrations of Mg, Al, S, K, Ca, Ti, Mn, Fe, Zn, and Pb was identified at AD-R, EL-R and JT-R (Figure 4.22), but not at NM-R and NM-T. This source was identified as resuspended dust. The resuspended dust source contributions to PM2.5 mass were 31, 0.29 and 15% at AD-R, EL-R and JT-R respectively. Resuspended road dust often consists of deposition of vehicle exhausts and industrial exhausts, tire and brake wears, dust from paved roads or potholes, and dust from construction sites. Figure 4.22 PM2.5 source profiles for resuspended dust factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Resuspended Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Resuspended Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Resuspended Dust C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 120 Iron (Fe) is emitted as a result of wear and tear of brake pads and other vehicular parts. Zinc (Zn) may have its origin from automobile sources such as wear and tear of vulcanized rubber tires, lubricating oil and corrosion of galvanized vehicular parts. Manganese (Mn) has been used as an additive in vehicle fuel (Karar et al., 2006). Al, Mn and K have also been used as tracers for paved road dust (Gupta et al., 2012). Time series plots (Figure 4.23) show higher concentrations in November, December and April, the dry season period. Figure 4.23 Time series of PM2.5 source contributions for resuspended dust factors. University of Ghana http://ugspace.ug.edu.gh 121 Source 6: Source profiles (Figure 4.24) with high loadings for Manganese (Mn), Copper (Cu) and Zinc (Zn) were identified at AD-R, NM-R and NM-T, but not at EL-R and JT-T. Cu and Zn are usually related to vehicular traffic such as motor vehicle exhaust and industrial sources from nonferrous smelting processes (Gao et al., 2002). Traffic produces road dust and air turbulence that can stir up road dust. The presence of Mn in this source profile may be due to the addition of methylcyclopentadienyl manganese tricarbonyl (MMT), a manganese-based gasoline antiknock fuel additive which has been in use in Ghana since January 2004 after the ban on the use of unleaded gasoline in the country. The source contributions (Figure 4.25) at AD-R show seasonal variations with higher concentrations during rainy season and lower concentrations during the dry season. In contrast, the source contributions at the Nima sites (NM-R and NM-T) show no seasonal variations. The higher source contributions observed at the AD-R site could be attributed to vehicle traffic diversion through the Asylum Down neighbourhood from Ring Road Central which is one of the largest, busiest and congested roads in Accra. Ring Road Central experienced major flooding during the rainy season. Distance to the nearest primary road at AD-R, NM-R and NM-T sites were 166 m, 188 m, and 25 m respectively (see Table 3.1). Traffic count conducted on the road in front of the AD-R site recorded an average daily traffic (ADT) of 3,114 vehicles/day. Roads in front of the two Nima sites (NM-R and NM-T) were 471 and 7,193 vehicles/day respectively. University of Ghana http://ugspace.ug.edu.gh 122 Figure 4.24 PM2.5 source profiles for traffic/industry factors. Higher source concentration at NM-R may be due to diversion of vehicular traffic (especially trotros and taxis) through the Nima residential area mostly on Wednesdays and Fridays which are Nima market days. This results in traffic congestion with idling engines. Road traffic congestion increases PM concentration as compared to free flowing traffic. Amato et al. (2010) found a factor with high mass concentrations of Zn, Pb and Mn and attributed it to industrial emissions. The location of Cu or Zn smelter within these neighbours needs further investigation. However, due to improper zoning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Traffic/Industry 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Traffic/Industry 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Traffic/Industry C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 123 in Accra, it is common to find some small and medium-scale industries in residential neighbours of Accra. Figure 4.25 Time series of PM2.5 source contributions for traffic/industry dust factors. University of Ghana http://ugspace.ug.edu.gh 124 4.5.1.2 COARSE PARTICULATE MATTER The reconstructed PM10 mass concentrations were estimated from the sum of the contributions of the PMF resolved sources as in the case of the PM2.5 mass concentrations. In Figure 4.26 to Figure 4.30, a comparison of the reconstructed PM10 contributions from all sources with measured PM10 concentrations shows that the PMF resolved sources effectively reproduce the measured total mass. In addition, the PMF derived for most of the variation in the PM10 concentrations had high linear correlation coefficient r2 (all greater than 0.93) at the five sites. The performance of PMF model was also evaluated by comparing the calculated values with measured concentrations for each species used in the model. These are shown in Table A-6 to Table A-10 in Appendix A. The average source contributions for all sites are summarized in Table 4.15. University of Ghana http://ugspace.ug.edu.gh 125 Figure 4.26 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at AD-R. Figure 4.27 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at EL-R y = 1.0357x - 2.3718 r² = 0.9687 0 20 40 60 80 100 120 0 20 40 60 80 100 120 P re d ic te d P M 1 0 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM10 Mass Concentration (µg m -3) Asylum Down (AD-R) y = 1.0041x - 0.4448 r² = 0.9731 0 20 40 60 80 100 0 20 40 60 80 100 P re d ic te d P M 1 0 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM10 Mass Concentration (µg m -3) East Legon (EL-R) University of Ghana http://ugspace.ug.edu.gh 126 Figure 4.28 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at JT-R. Figure 4.29 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at NM-R. y = 0.9993x - 0.3605 r² = 0.9368 0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160 P re d ic te d P M 1 0 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM10 Mass Concentration (µg m -3) James Town/Ussher Town (JT-R) y = 1.0041x - 0.7542 r² = 0.9357 0 20 40 60 80 100 120 0 20 40 60 80 100 120 P re d ic te d P M 1 0 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM10 Mass Concentration (µg m -3) Nima Residence (NM-R) University of Ghana http://ugspace.ug.edu.gh 127 Figure 4.30 Scatter plots of predicted PM10 mass concentrations and measured PM10 mass concentrations at NM-T. Source 1: The first source was identified as sea salt aerosol due to the high loadings of Na and Cl at all five sites. This source contributes 14, 8, 33, 12 and 14% on an average at AD-R, EL-R, JT-R, NM-R, and NM-T, respectively. The source profiles are shown in Figure 4.31 and the source contributions shown in Figure 4.32. Across sites, average contributions of the sea salt source in the PM10 were higher at JT-R (26.62 μg m -3) than at the other sites in Accra (3.44 μg m-3 - 10.02 μg m-3). As expected, sea salt was more dominant in the James Town/Ussher Town neighbourhood as a result of its proximity to the Atlantic Ocean than in the other three neighbours. Similar to the PM2.5 fraction discussed earlier, the highest contribution of this source to PM10 occurred in August. y = 1.0039x - 0.6498 r² = 0.9439 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 P re d ic te d P M 1 0 M as s C o n ce n tr at io n ( µ g m -3 ) Observed PM10 Mass Concentration (µg m -3) Nima Traffic University of Ghana http://ugspace.ug.edu.gh 128 From backward trajectories computed for 17 August 2008, it appears that this source is largely due to the tropical maritime air mass originating from the Atlantic Ocean. Table 4.15 Average contributions of identified sources to PM10 concentrations. AD-R EL-R JT-R NM-R NM-T Sea Salt 14.23% (7.85 μg m-3) 7.68% (3.44 μg m-3) 32.50% (26.62 μg m-3) 11.69% (5.90 μg m-3) 14.38% (10.02 μg m-3) Biomass burning 31.73% (17.50 μg m-3) 36.91% (16.50 μg m-3) 30.25% (24.78 μg m-3) 27.86% (14.06 μg m-3) 48.06% (33.49 μg m-3) Solid waste burning 3.11% (1.72 μg m-3) 5.21% (2.33 μg m-3) 5.10% (4.18 μg m-3) 4.83% (2.44 μg m-3) 3.54% (2.47 μg m-3) Soil Dust 34.00% (18.75 μg m-3) 38.89% (17.39 μg m-3) 25.67% (21.03 μg m-3) 41.57% (20.98 μg m-3) 25.95% (18.09 μg m-3) Resuspended dust 6.13% (3.38 μg m-3) 11.31% (5.06 μg m-3) 6.48% (5.31 μg m-3) Traffic/Industry 10.80% (5.96 μg m-3) 14.05% (7.09 μg m-3) 8.07% (5.63 μg m-3) University of Ghana http://ugspace.ug.edu.gh 129 Figure 4.31 PM10 source profiles for sea salt factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Sea Salt C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 130 Figure 4.32 Time series of PM10 source contributions for sea salt factors. University of Ghana http://ugspace.ug.edu.gh 131 Source 2: A biomass burning source was identified at all sites largely by the presence of high loadings of potassium (K) in the estimated source profiles as shown in Figure 4.33. This source accounted for 31.7, 36.9, 30.3, 27.9 and 48.1% of the apportioned PM10 mass at Ad-R, EL-R, JT-R, NM-R and NM-T, respectively. Biomass burning is a major component of PM10 in all the five neighbours. Relatively elevated source contributions were observed during the dry season as shown in Figure 4.34. This could be attributed to long-range transport of particles from wild bushfires, clearing or preparation land for farming by the process of burning. Source 3: A source with high loadings for Br and Pb and minor quantities of crustal species was identified at the 5 sites in Accra. This source can be attributed to solid waste burning (Vehlow et al., 2003; Nnoroma and Osibanjo, 2008). The source profiles and source contributions for this source are shown in Figure 4.35 and Figure 4.36, respectively. Similar sources were found in the PM2.5 fractions at all the sites. The burning of solid waste is common in residential and open areas in the city of Accra. The solid waste burning source accounted for 3.1%, 5.2%, 5.1%, 4.8% and 3.5% of the total PM10 mass measured at AD-R, EL-R, JT-R, NM-R and NM-T, respectively. The source contributions for solid waste burning show no seasonal variations. University of Ghana http://ugspace.ug.edu.gh 132 Figure 4.33 PM10 source profiles for biomass burning factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Biomass Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Biomass Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Biomass Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Biomass Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Biomass Burning C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 133 Figure 4.34 Time series of PM10 source contributions for biomass burning factors. University of Ghana http://ugspace.ug.edu.gh 134 Figure 4.35 PM10 source profiles for solid waste burning factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Solid Waste Burning C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 135 Figure 4.36 Time series of PM10 source contributions for solid waste burning factors. University of Ghana http://ugspace.ug.edu.gh 136 Source 4: The fourth source at all the five sites in Accra was identified as soil dust based on the high loadings of Mg, Al, Si, Ca, Ti, Mn and Fe in the source profiles (Figure 4.37). The contribution of soil dust to total PM10 varies from 26-42% and is therefore a significant contributor to PM10. In addition as expected, the contribution of soil dust in PM10 is greater than in PM2.5 since large diameter particles such as those found in soil dust are expected to be big contributors to PM10. In Accra, soil dust contributes significantly to both PM2.5 and PM10. The source contributions of the soil dust source for the five sites showed high peaks on 9th December 2007, 20th March 2008 and 13th April 2008 (Figure 4.38). In order to understand the source region of the soil dust particles, backward trajectories of the air mass movement on these dates were constructed at heights of 500, 1500 and 3000 m above ground level over 96 h using NOAA HYSPLIT (Draxler and Rolph, 2003), and the result displayed as Figure 4.39 showed that the source contributions on these days had similar backward trajectories originating from both local and long-range transported dust. University of Ghana http://ugspace.ug.edu.gh 137 Figure 4.37 PM10 source profiles for soil dust factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Soil Dust C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 138 Figure 4.38 Time series of PM10 source contributions for soil dust factors. University of Ghana http://ugspace.ug.edu.gh 139 (a) (b) (c) Figure 4.39 NOAA HYSPLIT model of air mass trajectories over Accra (location shown in star★) on (a) 9th December 2007, (b) 20th March 2008 and (c) 13th April 2008. Backward trajectory was derived from the NOAA website: www.arl.noaa.gov/ready/ University of Ghana http://ugspace.ug.edu.gh 140 Source 5: The same resuspended dust source like that found in the PM2.5 size range was identified for the PM10 fraction. The fifth source has significant loadings of Al, Si, K, Ti, Zn and Fe. It appeared to be a mixture of a resuspension of mineral dust or traffic related particles. This source was identified at AD-R, EL-R and JT-R but not at NM-R and NM-T. The resuspended dust source contributed 6.1, 11.3 and 6.5% to PM10 mass at AD-R, EL-R and JT-R respectively. The source profiles and source contributions for this source are shown in Figure 4.40 and Figure 4.41, respectively. It was expected that source contributions for this source would be higher in the PM10 fraction than the PM2.5 size range. However, with the exception of the EL-R site, the resuspended dust PM10 source contributions were lower in the other two neighbourhoods. It is not clear why this occurred, continuous monitoring could be able to explain this occurrence. Time series plots for PM10 showed higher concentrations in November, December and April, the dry season period. Source 6: The sixth source contributed about 10.8, 14.1 and 8.1% of the PM10 mass concentrations at AD-R, NM-R and NM-T. PMF model did not resolve this source for EL-R and JT-R sites. The sixth source had high concentrations Cr, Cu, Zn and Pb. These were the key indicator elements for industry and traffic emission sources attributed to wear of brake linings and tires as well as from lubricating oils (Sternbeck et al., 2002). This source was identified as traffic/industry source and is similar to that in the PM2.5 fraction. The traffic/industry source was mixed with loadings from crustal elements (Al, Si, Ca and Ti) and biomass combustion emissions (K). Time series plots (Figure 4.43) also show similar variations like the PM2.5 fractions discussed earlier. University of Ghana http://ugspace.ug.edu.gh 141 Figure 4.40 PM10 source profiles for resuspended dust factors. Figure 4.41 Time series of PM10 source contributions for resuspended dust factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Resuspended Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb EL-R Resuspended Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb JT-R Resuspended Dust C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 142 Figure 4.42 PM10 source profiles for traffic/industry factors. Figure 4.43 Time series of PM10 source contributions for traffic/industry factors. 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb AD-R Traffic/Industry 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-R Traffic/Industry 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb NM-T Traffic/Industry C o n ce n tr at io n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 143 4.5.2 HARMATTAN SEASON The Harmattan data for the five sites were pooled to increase the sample size. This was carried out after observing that during the non-Harmattan period source profiles were similar across sites but source contributions varied (Zhou et al., 2013). When PMF was applied to the pooled data for the Harmattan period, five sources in both the PM2.5 and PM10 size fractions were resolved by the model. The five sources identified in both fractions were Sea Salt, Biomass Burning, Soil Dust, Solid Waste Burning and Resuspended Dust. The source profile for the PM2.5 and PM10 fractions are shown in Figure 4.44 and Figure 4.46, respectively. The corresponding source contributions for PM2.5 and PM10 are respectively displayed in Figure 4.45 and Figure 4.47. The five sources accounted for 6.6, 44.3, 38.4, 0.4, and 44.3% respectively of the total PM2.5 mass. They also accounted for 9.3, 23.1, 59.0, 3.3, and 5.3.0% respectively of total PM10 mass. The first source identified by the PMF model in PM2.5 and PM10 was sea salt. This source had the highest mass fractions of sodium (Na) and chlorine (Cl). In source apportionment studies conducted in coastal areas, Guo et al. (2009) and Ofosu et al. (2012) identified a similar source with dominant Na and Cl in the source profiles and classified it as sea salt. The sea salt source contributed more to the PM10 mass concentration (25.8 μg m-3) as compared to PM2.5 (9.3 μg m -3) during this period. The second source with high loadings of sulphur (S) and potassium (K) in both PM2.5 and PM10 was attributed to biomass combustion. This source also reflects regional or University of Ghana http://ugspace.ug.edu.gh 144 long-range transport. The source contributions were relatively high throughout the Harmattan period. It was rather surprising that biomass burning source contributed 4 times more mass concentration to PM10 (63.97 μg m -3) than to PM2.5 (14.6 μg m -3). Continuous monitoring of PM during the Harmattan season may provide reasons for this anomaly. The third source includes most of the crustal elements with high concentrations of Mg, Al, Si, K, Ca, Ti, Mn and Fe. These elements are the major constituents of airborne soil and usually make an important contribution to PM10. As expected, soil dust contribution to PM10 (166.67 μg m -3) was 3 times more than that to PM2.5 (54.1 μg m -3). The fourth source with large contributions from bromine (Br) and lead (Pb) was identified as solid waste burning. PM2.5 and PM10 source contributions do not show any significant variations. This source had the lowest contributions to the two PM size fractions; 0.55 μg m-3 to PM2.5 and 9.2 μg m -3 to PM10, respectively. Solid waste burning source was not a significant contributor to PM during this period. The fifth source was characterized by high loadings of Al, Si, K, Ca, V, Ti, Cr, Mn, Fe, Cu, Zn and Pb. It was attributed to resuspended dust source and it contributed 44.3% (62.3 μg m-3) of PM2.5 mass and 5.3% (14.71 μg m -3) of PM10 mass, respectively. The high contribution to PM2.5 is probably because many roads are unpaved and there is traffic congestion throughout the city. University of Ghana http://ugspace.ug.edu.gh 145 Figure 4.44 PM2.5 source profiles for peak Harmattan months (25th December 2007 to 7th February 2008). 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Biomass 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Resuspended Dust C o n ce n tr a ti o n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 146 Figure 4.45 Time series of PM2.5 source contributions for peak Harmattan months (25th December 2007 to 7th February 2008). University of Ghana http://ugspace.ug.edu.gh 147 Figure 4.46 PM10 source profiles for peak Harmattan months (25th December 2007 to 7th February 2008). 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Sea Salt 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Biomass 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Soil Dust 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Solid Waste Burning 0.0001 0.001 0.01 0.1 1 Na Mg Al Si P S Cl K Ca Ti V Cr Mn Fe Cu Zn Br Sr Zr Pb Resuspended Dust C o n ce n tr a ti o n o f sp ec ie s (µ g m -3 ) University of Ghana http://ugspace.ug.edu.gh 148 Figure 4.47 Time series of PM10 source contributions for peak Harmattan months (25th December 2007 to 7th February 2008). University of Ghana http://ugspace.ug.edu.gh 149 CHAPTER 5: CONCLUSIONS AND RECOMMENDATIONS 5.1 CONCLUSIONS This work has investigated the levels, elemental composition and sources of ambient particulate matter from five different receptor sites in four locations, which span from poor to wealthy neighbourhoods in Accra, Ghana. To our knowledge, this is one of the first studies to examine the chemical composition and sources of air pollution in multiple neighbourhoods in a city in a developing country in a year-long campaign which accounts for seasonal differences; data from sub-Saharan Africa, the world’s fastest urbanizing region are even more limited. Particulate matter in two size fractions, PM2.5 and PM10 were determined at all sites between September 2007 and August 2008 by simultaneous 48 h measurement every six days. The mean mass concentration values for PM2.5 obtained at the five sites during the study period were 50.4 µg m-3, 45.9 µg m-3, 74.8 µg m-3, 48.3 µg m-3 and 57.5 for AD-R, EL- R, JT-R, NM-R AND NM-T, respectively. The mean mass PM10 mass concentration values were 108.2 µg m-3, 96.9 µg m-3, 134.8 µg m-3, 93.9 µg m-3 and 111.9 µg m-3at AD-R, EL-R, JT-R, NM-R and NM-T, respectively. These levels are all substantially higher than the WHO, USEPA and EU air quality standards. These results are consistent with the few studies in sub-Saharan Africa such as in urban Addis Ababa that found PM10 levels in the range of 40 to 100 µg m -3 and in South Africa that measured PM2.5 and PM10 levels of 86 and 97 µg m -3 respectively. University of Ghana http://ugspace.ug.edu.gh 150 Outside the Harmattan period, the mean annual PM2.5 and PM10 at neighbourhood sites ranged from 22 to 49 µg m-3, respectively. PM2.5 and PM10 during the Harmattan period (late December 2007 to early February 2008) were found to be 5 times higher than non- Harmattan period. Harmattan period results revealed that Harmattan dust contributes significantly to ambient particulate pollution. EDXRF technique was used to identify a total of 21 elements ranging from Al – Pb from the loaded filters. Between-neighbourhood and within-neighbourhood differences in the concentration of many trace element species were observed in this study. These differences could be related to specific local or regional sources affecting the receptor sites. During non-Harmattan months, the three most abundant PM10 crustal species at the five sites were Si (5736 to 4002 ng m-3), Al (1654 to 2591 ng m-3), and Fe (1291 to 2034 ng m-3). During the Harmattan months these same crustal species in PM10 were highly elevated, Si (31467 to 37526 ng m-3), Al (11599 to 14234 ng m-3) and Fe (8119 to10121 ng m-3). During the non-Harmattan period between 49% and 55% of PM10 mass was found in the fine of PM2.5 fraction at the various sites. During Harmattan period, it was observed that about 50% to 53% of PM10 mass was found in PM2.5 fraction. Elemental species such as K, Cu, Zn, Br and Pb found to have 60% or more of their mass in the fine fraction. These elements are mainly anthropogenic in origin and could originate from sources such as biomass burning, solid waste burning, motor vehicles emissions and industrial emissions. University of Ghana http://ugspace.ug.edu.gh 151 Enrichment factor results of PM2.5 and PM10 chemical components for all sites in Accra indicate that elements of anthropogenic origin such as Zn, S, Cl, Br and Pb were highly enriched with respect to crustal elements. With the exception of temperature which shows a positive relationship with PM2.5, the other weather parameters (rainfall, relative humidity and wind speed) had a negative relationship during the study period. However, these relationships were weak and not significant during the period of study. The receptor model PMF was used to identify sources and their contributions to ambient airborne particulate matter at the five sites in Accra. During non-Harmattan months, PMF resolved for both PM2.5 and PM10 six sources for AD-R and five sources for EL-R, JT-R, NM-R and NM-T, respectively. Sea salt, biomass burning, solid waste burning and soil dust were ubiquitous sources and were identified at the all five sites. Resuspended dust was identified at three sites. Traffic/industry emissions were identified as a source at three sites near traffic routes. It was not possible to separate of motor vehicle emissions from traffic into diesel and gasoline vehicle contributions due to the lack of data on elemental carbon and organic carbon. In general, anthropogenic related sources were found to be dominant in all neighbourhoods with the exception of East Legon. Contributions of pollutant from natural sources were 27, 56, 34, 32, and 46% at AD-R, EL-R, JT-R, NM-R and NM-T, respectively. Contributions from anthropogenic sources were 73, 44, 66, 68, and 54% at AD-R, EL-R, JT-R, NM-R and NM-T, respectively. University of Ghana http://ugspace.ug.edu.gh 152 PMF model identify five sources in both PM2.5 and PM10 pooled data. The five sources were sea salt, biomass burning, soil dust, solid waste burning and resuspended dust. These sources were similar to those resolved during non-Harmattan period at all the sites but much higher contributions from crustal sources during Harmattan as there had been in other months. 5.2. RECOMMENDATIONS From the results, discussions and conclusions of this research, it is recommended that:  Permanent PM monitoring stations are established at rural, urban and industrial areas to conduct regular monitoring of key indicator pollutants.  Short duration studies are routinely implemented to estimate PM2.5/PM10 ratio for different cities and seasons.  Black carbon analysis is conducted in future studies in these neighbourhoods to understand their levels and health effects.  Attention is given to land use and agriculture policies in addition to wildfire management. This is with respect to the finding that there are contributions from regional long-range transport of particulate matter due to wildfire and the annual bush burning that take place in preparation for the farming period.  Set up effective source-based control strategies and policy on the role of urban biomass burning.  Regulate the importation of over-aged second hand vehicles and engines by limiting their numbers or through taxation. University of Ghana http://ugspace.ug.edu.gh 153  Low-emission mass transportation system is established to decongest the roads, reduce air pollution and fuel consumption.  Increase community awareness and education on the effect of air pollution as far as particulate matter are concerned. University of Ghana http://ugspace.ug.edu.gh 154 REFERENCES 1. Aboh, I. J. K. (2009). Air quality over southern Ghana: mass, black carbon, elemental concentrations and sources of air particulate matter at Kwabenya, near Accra-Ghana. PhD Thesis. Department of Physics, University of Ghana, Accra, Ghana. 2. Aboh, I. J. K., Henriksson, D., Laursen, J., Lundin, M., Ofosu, F. G., Pind, N., Selin Lindgren, E., Wahnström, T. (2009a). Identification of aerosol particle sources in semi‐rural area of Kwabenya, near Accra, Ghana, by EDXRF techniques. X‐Ray Spectrometry 38, 348–353. 3. Adam, I., Okyere, D., Teye, M. (2013). Assessment of Heavy Metal Residues in Hides of Goats Singed with Tyres, and the Effect of Boiling on the Heavy Metal Concentrations in the Hides. Journal of Veterinary Advances, 3(5), 165-169. 4. Afeti, G.M. and Resch, F.J. (2000). Physical characteristics of Saharan dust near the Gulf of Guinea. Atmospheric Environment. 34, 1273–1279. 5. Agyemang-Bonsu W.K., Tutu-Benefoh D. and Asiamah, H. (2007). Ghana Vehicular Emission Report, Accra, Ghana. Environmental Protection Agency, Accra. 6. Akbar, S. and Kojima, M. (2002). What do we know about air pollution? – India case study: UNDP/World Bank Energy Sector management Assistance Program. 7. Alfarra, M. R. (2004). Insights into atmospheric organic aerosols using an aerosol mass spectrometer, PhD Thesis, Department of Chemical Engineering, UMIST, Manchester, UK. 8. Allen, G., Oh, J.A., Koutrakis, P., Sioutas, C. (1999). Techniques for high- quality ambient coarse particle mass measurements. J Air Waste Manage Assoc.49, 133–141. University of Ghana http://ugspace.ug.edu.gh 155 9. Amato, F., Pandolfi, M., Escrig, A., Querol, X., Alastuey, A., Peya, J., Perez, N., Hopke, P.K., (2009). Quantifying road dust resuspension in urban environment by Multilinear Engine: A comparison with PMF2. Atmos. Environ. 43, 2770- 2780. 10. Andah, W. E. I., van de Giesen, N., and Biney, C. A. (2003). Water, climate, food, and environment in the Volta Basin. Contribution to the ADAPT Project, Adaptation strategies to changing environments. Institute for Environmental Studies, Free University of Amsterdam. http://www.weap21.org/downloads/AdaptVolta.pdf. Last accessed 18/04/2013. 11. Andrade, M.F., Miranda, R.M., Fornaro, A., Kerr, A., Oyama, B., Andre, P.A., Saldiva, P. (2010). Vehicle emissions and PM2.5 mass concentrations in six Brazilian cities. Air Qual Atmos Health 5, 79–88 40. doi:10.1007/s11869-010- 0104-5. 12. Aneja, V. P., Wang, B., Tong, D. Q., Kimball, H. and Steger, J. (2006). Characterization of Major Chemical Components of Fine Particulate Matter in North Carolina. J. Air & Waste Manage. Assoc. 56, 1099–1107. 13. Arku, R. E., Vallarino, J., Dionisio, K. L., Willis, R., Choi, H., Wilson, J. G., Hemphill, C., Agyei-Mensah, S., Spengler, J. D., Ezzati, M. (2008). Characterizing air pollution in two low-income neighborhoods in Accra, Ghana. Science of The Total Environment. 402, 217–231. 14. Bailis, R., Ezzati, M., and Kammen, D. M. (2005). Mortality and greenhouse gas impacts of biomass and petroleum energy futures in Africa. Science, 308:98- 103. University of Ghana http://ugspace.ug.edu.gh 156 15. Bandhu, H.K., Puri, S., Garg, M.L., Singh, B., Shahi, J.S., Mehta, D., Swietlicki, E., Dhawan, D.K., Mangal, P.C., Singh, N., (2000). Elemental composition and sources of air pollution in the city of Chandigarh, India, using EDXRF and PIXE techniques. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms. 160, 126–138. 16. Barnes, D. F., Krutilla, K., and Hyde, W. F. (2005). The Urban Household Energy Transition: Social and Environmental Impacts in the Developing World Washington, DC: RFF Press. 17. Basha, S., Jhala, J., Thorat, R., Goel, S., Trivedi, R., Shah, K. Menon, G., Gaur, P., Mody, K.H., Jha, B. (2010). Assessment of heavy metal content in suspended particulate matter of coastal industrial town, Mithapur, Gujarat, India. Atmospheric Research. 97, 257–265. 18. Bates, T.S., Lamb, B. K., Guenther, A., Dignon, J. and Stoiber, R. E. (1992). Sulfur emissions to the atmosphere from natural sources. J. Atmos. Chem., 14, 315-337. 19. Bauer, H., Schueller, E., Weinke, G., Berger, A., Hitzenberger, R., Marr, I. L., Puxbaum, H. (2008). Significant contributions of fungal spores to the organic carbon and to the aerosol mass balance of the urban atmospheric aerosol. Atmospheric Environment, 42(22), 5542–5549. 20. Belis, C.A., Karagulian, F., Larsen, B.R., Hopke, P.K. (2013). Critical review and meta-analysis of ambient particulate matter source apportionment using receptor models in Europe. Atmospheric Environment. 69, 94–108. 21. Benkowitz, C. M., Scholtz, M.T., Pacyna, J., Tarrason, L., Dignon, J., Voldner, E. C., Spiro, P.A., Logan, J.A. and Graedel, T.E. (1996) Global gridded inventories of anthropogenic emissions of sulfur and nitrogen. J. Geophys. Res., 101, 29239-29253. University of Ghana http://ugspace.ug.edu.gh 157 22. Benneh, G. and Agyapong, G. T. (1990). Land Degradation in Ghana. London: Commonwealth Secretariat, and Legon: Department of Geography and Resource Development. University of Ghana. 23. Benneh, G., Songsore, J., Nabila, J., Amuzu, A., Tutu, K., Yangyuoru, Y. and McGranahan, G. (1993). Environmental Problems and the Urban Household in the Greater Accra Metropolitan Area (GAMA)-Ghana. Stockholm Environment Institute, Stockholm. 24. Bennet, C., Jonsson, P., Lingren, E. S. (2005). Concentrations and sources of trace elements in particulate air pollution, Dar es Salaam, Tanzania, studied by EDXRF. X-Ray Spectrometry. 34, 1–6. 25. Bhaskar, B.V., Mehta, V.M. (2010). Atmospheric Particulate Pollutants and their Relationship with Meteorology in Ahmedabad. Aerosol and Air Quality Research. 10, 301–315. 26. Boadi, K. O. and Kuitunen, M. (2005). Factors affecting the choice of cooking fuel, cooking place, and respiratory health in Accra metropolitan area, Ghana. J. Biosoc. Sci. 38, 403–412. 27. Braga, C. F., Teixeira, E. C., Meira, L., Wiegand, F., Yoneama, M. L., Dias, J. F. (2005). Elemental composition of PM10 and PM2.5 in urban environment in South Brazil. Atmospheric Environment 39, 1801-1815. 28. Breslow, L. and Goldsmith, J. (1958). Health effects of air pollution. American Journal of Public Health and the Nations Health, 48(7), 913–917. 29. Breuning-Madsen, H. and Awadzi, W.A. (2005). Harmattan dust deposition and particle size in Ghana. Catena. 63, 23–28. University of Ghana http://ugspace.ug.edu.gh 158 30. Bullock, K.R., Duvall, R.M., Norris, G.A., McDow, S.R., Hays, M.D. (2008). Evaluation of the CMB and PMF models using organic molecular markers in fine particulate matter collected during the Pittsburgh air quality study. Atmospheric Environment 42 (29), 6897-6904. 31. Buzorius, G., Hameri, K., Pekkanen, J. and Kulmala, M. (1999). Spatial variation of aerosol number concentration in Helsinki city. Atmospheric Environment 33, 553–565. 32. Cadle, S.H., Mulawa, P.A., Hunsanger, E.C., Nelson K., Ragazzi, R.A., Barrett, R., Gallagher, G.L., Lawson, D.R., Knapp, K.T., Snow, R. (1999). Composition of light-duty motor vehicle exhaust particulate matter in the Denver, Colorado area. Environ Sci Technol. 33, 2328–2339. 33. Canadian Chemical Producers’ Association (CCPA). (2001). Ambient Particulate Matter Characterization Guidelines. Ottawa, ON. Canada. 34. Cao, J. J., Lee, S. C., Zheng, X. D. et al., (2003). Characterization of dust storms to Hong Kong in April 1998. J. Water, Air, and Soil Pollution Focus. 3, 213- 229. 35. Central Intelligence Agency (CIA) (2011). The World Factbook: Estimate July 2011.https://www.cia.gov/library/publications/the-world-actbook/fields/ 2119. html. Last accessed 26/09/2011. 36. CEPA/FPAC Working Group on Air Quality Objectives and Guidelines. (1998). National ambient air quality objectives for particulate matter: Part 1: Science Assessment Document. Ottawa, Ont: Working Group, Health Canada. 37. Chalfin, B. (2008). Cars, the Customs Service, and Sumptuary Rule in Neoliberal Ghana. Comparative Studies in Society and History 50(2), 424-453. University of Ghana http://ugspace.ug.edu.gh 159 38. Chao, C. Y., Wong, K. K. (2002). Residential indoor PM10 and PM2.5 in Hong Kong and the elemental composition. Atmospheric Environment. 36, 265-277. 39. Chimidza, S., Viksna, A., Lindgren, E.S. (2001). EDXRF and TXRF analysis of aerosol particles and the mobile fraction of soil in Botswana. X-Ray Spectrometry 30, 301-307. 40. Chow, J.C. (1995). “Critical Review: Measurement Methods to Determine Compliance of Ambient Air Quality Standards for Suspended Particles.” J. Air and Waste Management Assoc., 45, 320–385. 41. Chow, J.C. (1995). Measurement methods to determine compliance with ambient air quality standards for suspended particles. Journal of Air and Waste Management Association 45, 320–382. 42. Cohen, A., Anderson, R., Ostro, B., Pandey, K.D., Krzyzanowski, M., Künzli, N., et al. (2004). Urban ambient air pollution. In: Comparative Quantification of Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk Factors (Ezzati, M., Lopez, A.D., Rodgers, A., Murray, C.J.L., eds). Geneva: World Health Organization, 1353–1433. 43. Cohen, D. D., Garton, D., Stelcer, E., Wang, T., Poon, S., Kim, J., et al. (2002). Characterisation of PM25 and PM10 Fine Particle Pollution in Several Asian Regions. 16th Int Clean Air Conf. Christchurch, NZ. 44. Davidson, C. I., Lin, S. F., Osbom, J. F. (1986). Indoor and outdoor air pollution in the Himalayas. J. Environ. Sci. Technol. 20, 561–567. 45. DeGaetano, A.T., Doherty, O.M. (2004). Temporal, spatial and meteorological variations in hourly PM2.5 concentration extremes in New York City. Atmos Environ. 38, 1547–58. University of Ghana http://ugspace.ug.edu.gh 160 46. Dickson, K. B. and Benneh, G. (1988). A new geography of Ghana. London: Longman. 47. Dionisio, K.L., Arku, R.E., Hughes, A.F., Vallarino, J., Carmichael, H., Spengler, J.D., Agyei-Mensah, S., Ezzati, M. (2010). Air pollution in Accra neighborhoods: spatial, socioeconomic, and temporal patterns. Environmental Science and Technolog. 44(7), 2270-6. 48. Dionisio, K.L., Rooney, M.S., Arku, R.E., Friedman, A.B., Hughes, A.F., Vallarino, J., Agyei-Mensah, S., Spengler, J.D., Ezzati, M. (2010).Within- Neighborhood Patterns and Sources of Particle Pollution: Mobile Monitoring and Geographic Information System Analysis in Four Communities in Accra, Ghana. Environ. Health Perspect. 118, 607-613. 49. Dockery, D. and Pope, A. (1996). Epidemiology of acute health effects: summary of time series studied. In: Wilson, R., Spengler, J. D. (Eds.), Particles in Our Air: Concentration and Health Effects. Harvard University Press, Cambridge, MA, USA, pp. 123–147. 50. Dockery, D. W., Pope, C. A. III, (1994). Acute Respiratory Effects of Particulate Air Pollution. Annu. Rev. Public Health. 15, 107. 51. Dockery, D.W., Pope, C. A., Xu, X. P., Spengler, J. D., Ware, J. H., Fay, M. E., Ferris, B. G. and Speizer, F. E. (1993). An association between air-pollution and mortality in six United-States cities, New England Journal of Medicine. 329, 1753-1759. 52. Draxler, R.R., Rolph, G.D. (2003). HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model access via NOAA ARL READY Website (http://www.arl.noaa.gov/ready/hysplit4.html). NOAA Air Resources Laboratory, Silver Spring, MD. University of Ghana http://ugspace.ug.edu.gh 161 53. Driver and Vehicle Licensing Authority (DVLA) (2010). Population of vehicles registered in Ghana by category. Driver and Vehicle Licensing Authority,Accra- Ghana. http://www.mrt.gov.gh/ststistics/DVLA/dvla_vlicence2010. Last accessed 15/07/12. 54. Duce, R.A., Hoffman, G.L., Zoller, W.H., (1975). Atmospheric trace metals at remote northern and southern hemisphere sites: pollution or natural? Science 187(4171), 59–61 55. Ellis, J.B., Revitt, D.M. (1982). Incidence of heavy metals in street surface sediments: solubility and grain size studies. Water Air Soil Pollut. 17, 87–100. 56. Engelbrecht D.R., Cahill, T.A., Feeney, P.J. (1980). Electrostatic effects on gravimetric analysis of membrane filters. JAPCA. 30, 391-392. 57. Engelbrecht JP, Swanepoel L, Chow JC, Watson JG, Egami RT. (2001). PM2.5 and PM10 concentrations from the Qalabotjha low-smoke fuels macro-scale experiment in South Africa. Environ. Monit. Assess. 69, 1–15. 58. EPA (2002). State of Environment Report 2001. Environmental Protection Agency, Accra, Ghana. 59. Etyemezian, V., Tesfaye, M., Yimer, A., Chow, J.C., Mesfin, D., Nega, T., et al. (2005). Results from a pilot-scale air quality study in Addis Ababa, Ethiopia. Atmos Environ. 39, 7849–7860. 60. Fabretti, J. F., Sauret, N., Gal, J. F., Maria, P. C., Schärer, U. (2009). Elemental characterization and source identification of PM2.5 using Positive Matrix Factorization: the Malraux road tunnel, Nice, France. Atmospheric Research 94, 320–329. University of Ghana http://ugspace.ug.edu.gh 162 61. Faiz, A. and Sturm, P., (2000). New directions: air pollution and road traffic in developing countries. Atmospheric Environment. 34(27): 4745–4746. 62. Feeney, P.J., Cahill, T.A., Olivera, J., Guidara, R. (1984). Gravimetric determination of mass on lightly loaded membrane filters. JAPCA 34, 376-378. 63. Finlayson-Pitts, B. J. and Pitts, J. N. (1986). Atmospheric Chemistry. Wiley, New York. 64. Finlayson-Pitts, B. J. and Pitts, J. N. (2000). Chemistry of the upper and lower atmosphere. Theory, experiments and applications. Academic Press, London, pp. 349–440. 65. Fourn, L. and Fayomi, E.B. (2006). Air pollution in urban area in Cotonou and Lokossa, Benin. Bull. Soc. Pathol. Exot. 99, 264–268. 66. Friend, A., Ayoko, G., Jayaratne, R., Jamriska, M., Hopke, P., Morawska, L., (2012). Source apportionment of ultrafine and fine particle concentrations in Brisbane, Australia. Environmental Science and Pollution Research, 19(7), pp. 2942-2950. 67. Garg, B.D., Cadle, S.H., Mulawa, P.A., Groblicki, P.J., Laroo, C., Parr, G.A. (2000). Brake wear particulate matter emissions. Environ Sci Technol 34, 4463– 4469. 68. Gatari, M., Boman, J. and Maina, D. (2005). Trace element categorization of pollution sources in the equator town of Nanyuki, Kenya. X-Ray Spectrom. 34, 118–123. 69. GEF (2007). Ghana Urban Transport Project. Global Environment Facility. http://www.theGEF.org. Last accessed 11/12/2012. University of Ghana http://ugspace.ug.edu.gh 163 70. Ghana Homepage (2014).Ghana Administrative Map. http://cdn.ghanaweb.com/GHP/img/pics/59387424.gif. Last accessed 20/02/2014. 71. Ghana Statistical Service (GSS) (2010). Population and Housing Census Final Results http://www.statsghana.gov.gh/docfiles/2010phc/2010_POPULATION_AND_H OUSING_CENSUS_FINAL_RESULTS.pdf. Last accessed 16/12/2012. 72. Godish, T., (2004). Air Quality. 4th Edition. Boca Raton: Lewis. 73. Gordon, C., Nukpezah, D., Tweneboah-Lawson, E., Ofori, B.D., Yirenya- Tawiah, D., Pabi, O., Ayivor, J. S., Koranteng, S., Darko, D. and Mensah, A.M. (2013). West Africa – Water Resources Vulnerability Using a Multidimensional Approach: Case Study of Volta Basin. Climate Vulnerability: Understanding and Addressing Threats to Essential Resources. Elsevier Inc., Academic Press, 283–309. 74. Guo, H., Ding, A.J., So, K.L., Ayoko, G., Li, Y.S. and Hung, W.T. (2009). Receptor modelling of source apportionment of Hong Kong aerosols and the implication of urban and regional contribution. Atmos. Environ. 43, 1159–1169. 75. Gupta, I., Salunkhe, A., Kumar, R., (2012). Source Apportionment of PM10 by Positive Matrix Factorization in Urban Area of Mumbai, India. Scientific World Journal 2012. 76. Hänninen, O.O., Koistinen, K.J., Kousa, A., Keski-Karhu, J., Jantunen, M.J. (2002). Quantitative analysis of environmental factors in differential weighing of blank Teflon filters. J Air Waste Manage Assoc. 52, 134-139. University of Ghana http://ugspace.ug.edu.gh 164 77. Harrison, R. M. and Yin, J. X. (2000). Particulate matter in the atmosphere: Which particle properties are important for its effects on health? Science and Total Environment, 249: 85-101. 78. He, K., Yang, F., Ma, Y., Zhang, Q., Yao, X., Chan, C. K., Cadle, S., Chan, T., and Mulawa, P. (2001). The characteristics of PM2.5 in Beijing, China, Atmos. Environ. 35, 4959–4970. 79. Health Effects Institute (HEI) (2010). Traffic related air pollution: A critical review of the literature on emissions, exposure and health effects, (HEI Special Report 17). 80. Hinds, W.C. (1982). Aerosol technology: properties, behaviour, and measurement of airborne particles. New York, Wiley. 81. Hopke P. K., Ito, K., Mar, T., Christensen, W.F., Eatough, D. J., Henry, R. C., Kim, E., Laden, F., Lall, R., Larson, T. V., Liu, H., Neas, L., Pinto, J., Stölzel, M., Suh, H., Paatero, P., Thurston, G. D. (2006). PM source apportionment and health effects: 1. Intercomparison of source apportionment results. J. Exposure Sci. Environ. Epidemiol. 16, 275–286. 82. Hosiokangas, J., Kikas U., Pekkanen, J., Ruuskanen, J., Tammet, H. (1995). Identifying and quantifying air pollution sources in Kuopio by receptor modeling. Journal of Aerosol Science. 26, 423–424. 83. Huang, S. and Conte, M. H. (2009). Source/process apportionment of major and trace elements in sinking particles in the Sargasso sea. Geochimica et Cosmochimica Acta. 73, 65-90. 84. International Atomic Energy Agency (IAEA) (1992). Sampling and Analytical Methodologies for Instrumental Neutron Activation Analysis of Airborne Particulate Matter. IAEA-TCS-4, IAEA, Vienna. University of Ghana http://ugspace.ug.edu.gh 165 85. Jeong, C.-H., Lee, D.-W., Kim, E., Hopke, P. K. (2004). Measurement of real- time PM2.5 mass, sulfate, and carbonaceous aerosols at the multiple monitoring sites. Atmospheric Environment 38, 5247–5256. 86. Jiries, A.G., Hussein, H.H., Halash, Z. (2001). The quality of water and sediments of street runoff in Amman. Jordan. Hydrol Proced. 15, 815–824. 87. Johnson, T.M., Guttikunda, S., Wells, G.J., Artaxo, P., Bond, T.C., Russell, A.G. (2011). Tools for Improving Air Quality Management – A Review of Top- down Source Apportionment Techniques and Their Application in Developing Countries. Energy Sector Management Assistance Program (ESMAP) Formal Report 339/11. The International Bank for Reconstruction and Development/THE WORLD BANK GROUP. 88. Kam, W., Liacos, J. W., Schauer, J. J., Delfino, R. J., and Sioutas, C. (2012). On-road emission factors of PM pollutants for light-duty vehicles (LDVs) based on urban street driving conditions. Atmospheric Environment. 61, 378-386. 89. Karar, K. and Gupta, A.K. (2006). Seasonal variations and chemical characterization of ambient PM10 at residential and industrial sites of an urban region of Kolkata (Calcutta), India. Atmos. Res. 81, 36–53. 90. Karar, K., Gupta, A. K., Kumar, A., Biswas, A. K., and Devotta, S. (2006). Statistical interpretation of weekday/weekend differences of ambient particulate matter, vehicular traffic and meteorological parameters in an urban region of Kolkata, India. Indoor and Built Environment, 15(3), 235-245. 91. Karar, K., Gupta, A.K., Kumar, A., Biswas, A.K. and Devotta, S. (2005). Statistical Interpretation of Week Day/week End Differences of Ambient Gaseous Pollutant, Vehicular Traffic and Meteorological Parameter in Urban Region of Kolkatta. J. Environ. Sci. Eng. 47, 164–175. University of Ghana http://ugspace.ug.edu.gh 166 92. Kgabi, N. (2010). An assessment of common atmospheric particulate matter sampling and toxic metal analysis methods. African Journal of Environmental Science and Technology. 4(11), 718–728. 93. Koistinen, K.J., Kousa, A., Tenhola, V., Hänninen, O., Jantunen, M., Oglesby, L., Künzli, N., Georgoulis, L. (1999). Fine particle (PM2.5) measurement methodology, quality assurance procedures and pilot results of the EXPOLIS study. J Air Waste Manage Assoc. 49, 1212–1220 . 94. Kojima, M., Lovei, M. (2001). Urban Air Quality Management: Coordinating Transport, Environment, and Energy Policies in Developing Countries. World Bank Technical Paper No. 508, Washington DC, 56pp. 95. Kothai, P., Saradhi, I. V., Pandit, G. G., Markwitz, A., Puranik, V. D. (2011). Chemical characterization and source identification of particulate matter at an urban site of Navi Mumbai, India. Aerosol and Air Quality Research 11, 560– 569. 96. Kothai, P., Saradhi, I.V., Prathibha, P., Hopke, P. K., Pandit, G. G., and Puranik, V. D. (2008). Source apportionment of coarse and fine particulate matter at Navi Mumbai, India. Aerosol and Air Quality Research, 8(4):423–436. 97. Krzyzanowski, M., Kuna-Dibbert, B. and Schneider, J. (Eds). (2005). Health Effects of Transport-related Air Pollution. Copenhagen: World Health Organization, Regional Office for Europe. http://www.euro.who.int/document/e86650.pdf. Last accessed 25/07/2009. 98. Kulkarni, P., Baron, P. A., Willeke, K., (2011). Introduction to Aerosol Characterization. In: Aerosol Measurement: Principles, Techniques, and Applications, 3rd Edition. Kulkarni, P., Baron, P. A. and Willeke, K. (Eds.), Wiley Interscience, New Jersey. University of Ghana http://ugspace.ug.edu.gh 167 99. Kupiainen, K. (2007). Road dust from pavement wear and traction sanding. Monographs of the Boreal Environment Research. 26, p. 50. 100. Lawless, P.A., Rodes, C.E. (1999). Maximizing data quality in the gravimetric analysis of personal exposure sample filters. J Air Waste Manage Assoc. 49, 1039–1049. 101. Li, X.D., Poon, C., Liu, P.S. (2001). Heavy metal contamination of urban soils and street dusts in Hong Kong. Applied Geochemistry 16, 1361–1368. 102. Markowicz, A., Aleta, A., Smodis, B., (2002). IAEA activities in support of air pollution monitoring in Asia and Pacific Region. Better Air Quality in Asian and Pacific Rim countries, 16 – 18 Dec 2002, Hong Kong, SAR, PS-16, 1- 7. http://www.cse.polyu.edu.hk/~activi/BAQ2002/BAQ2002_files/Proceedings/Post erSession/16.pdf. 103. Marple, V.A., and Willeke, K. (1976). Impactors design. Atmos. Environ.10, 891– 896. 104. Marple, V.A., Rubow, K.L., Turner, W. and Spengler, J.D. (1987). Low Flow Rate Sharp Cut Impactors for Indoor Air Sampling: Design and Calibration. Japca, 37, 1303–1307. 105. Mathys, P. (2002). WP7 Report: Method. www.ecrhs.org, Last accessed 21/10/2013. 106. Mayer, H. (1999). Air pollution in cities. Atmospheric Environment. 33, 3977– 3988. 107. Miguel, E.D., Llmas, J.F., Chacon, E., Berg, T., Larssen, S., Rbyset, O. (1997). Origin and patterns of distribution of trace elements in street dust: unleaded petrol and urban lead. Atmospheric Environment 31(17) 2733 –2740. University of Ghana http://ugspace.ug.edu.gh 168 108. Miller, S. L., Anderson, M. J., Daly, E. P., and Milford, J. B. (2002). Source apportionment of exposures to volatile organic compounds. I. Evaluation of receptor models using simulated exposure data. Atmospheric Environment, 36, 3629–3641. 109. Miranda, F. and Tomaz, E. (2008). Characterization of Urban Aerosol in Campinas, Sao Paulo, Brazil. Atmospheric Research 87: 147-157. 110. Mkoma, S.L. and Ibrahimu, M.C. (2011). Influence of Meteorology on Ambient Air Quality in Morogoro, Tanzania International Journal of Environmental Sciences 1(6), 1114–1122. 111. Mkoma, S. L. (2008). Physico-chemical characterisation of atmospheric aerosols in Tanzania, with emphasis on carbonaceous aerosol components and on chemical mass closure. PhD thesis. Ghent University. (ISBN 978-90-5989-230- 9). 112. Mkoma, S. L., Maenhaut, W., Chi, X., Wang, W., and Raes, N. (2009). Characterisation of PM10 Atmospheric Aerosols for the Wet Season 2005 at Two Sites in Tanzania, East Africa., Atmos. Environ., 43, 631–639 . 113. Mkoma, S.L., Tungaraza, C., Maenhaut, W. and Raes, N. (2010). Elemental Composition and Sources of Atmospheric Particulate Matter in Dar es Salaam, Tanzania. Ethiopian Journal of Environmental Studies and Management. 3(1), 20–29. 114. Møller-Jensen, L., Kofie, R.Y., Yankson, P.W.K. (2005). Large-area urban growth observations – a hierarchical kernel approach based on image texture. Geografisk Tidsskrift, Danish Journal of Geography 105(2), 39-47. 115. Nerquaye-Tetteh, E. (2009). Urban Air Quality Monitoring in Accra – Case Study. Environmental Protection Agency – Ghana: Accra, Ghana. University of Ghana http://ugspace.ug.edu.gh 169 http://www.unep.org/urban_environment/PDFs/BAQ09_ghanacasestudy.pdf. Last accessed 05/11/2009. 116. Nesamani, K.S., Chu, L., McNally, M.G., and Jayakrishnan, R., (2007). Estimation of vehicular emissions by capturing traffic variations. Atmospheric Environment 41, 2996–3008. 117. Nnorom, I. C., and Osibanjo, O. (2008). Sound management of brominated flame retarded (BFR) plastics from electronic wastes: State of the art and options in Nigeria. Resources, Conservation and Recycling. 52(12), 1362-1372. 118. Obiri-Danso, K., Hogarh, J.N., Antwi-Agyei, P. (2008). Assessment of contamination of singed hides from cattle and goats by heavy metals in Ghana. African J. Environ. Sci. Technol. 2(8), 217–221. 119. Obuobie, E., Keraita, B., Danso, G., Amoah, P., Cofie, O. O., Raschid-Sally, L., and Drechsel, P. (2006). Irrigated urban vegetable production in Ghana. Characteristics, benefits and risks. CSIR-INSTI, Printing Division, Accra pp.118-132. 120. OECD (2012). OECD Environmental Outlook to 2050: The Consequences of Inaction - Key Findings on Health and Environment. 121. Ofosu, F.G., Hopke, P. K., Aboh, I. J.K., Bamford, S. A. (2012). Characterization of fine particulate sources at Ashaiman in Greater Accra, Ghana. Atmospheric Pollution Research. 3, 301‐310. 122. Ojo, O. (1977). The Climates of West Africa. Heinemann Educational Books Ltd. 123. Olic, I., Wen, X., Ng, T.H. and Tang, S.M. (1999). Two years of aerosol pollution monitoring in Singapore: a review. Nuclear Instruments and Methods in Physics Research B 150, 457 – 464. University of Ghana http://ugspace.ug.edu.gh 170 124. Oluyemi, E.A and Asubiojo, O.I. (2001). Ambient air particulate matter in Lagos, Nigeria: a study using receptor modelling with X-ray fluorescence analysis. Bulletin of the Chemical Society of Ethiopia. 15, 97-108. 125. Ondráček, J., Schwarz, J., Ždímal, V., Andělová, L., Vodička, P., Bízek, V., Tsai, C.-J., Chen, S.-C., Smolik, J. (2011). Contribution of the road traffic to air pollution in the Prague city (busy speedway and suburban crossroads) Atmospheric Environment. 45, 5090–5100. 126. Oppong, R. A. and Badu, E. (2013). Building Material Preferences in Warm- Humid and Hot-Dry Climates in Ghana. Journal of Science and Technology (Ghana), 32(3), 24-37. 127. Owoade, O.K., Olise, F.S., Ogundele, L.T., Fawole, O.G., Olaniyi, H.B. (2012). Correlation between particulate matter concentrations and meteorological parameters at a site in Ile-Ife, Nigeria. Ife Journal of Science. 14(1), 83 – 93. 128. Paatero, P. and Hopke, P.K. (2003). Discarding or downweighting high-noise variables in factor analytic models. Anal. Chim. Acta. 490, 277-289. 129. Paatero, P. and Tapper, U. (1994). Positive Matrix Factorisation: A Non-negative Factor Model with Optimal Utilization of Error Estimates of Data Value. Environmetrics 5, 111–126. 130. Paatero, P., (1997). Least squares formulation of robust non-negative factor analysis. Chemometrics and Intelligent Laboratory Systems. 37, 23–35. 131. Pandis, S. N., Harley, R. A., Cass, G. R., Seinfeld, J. H., (1992). Secondary organic aerosol formation and transport. Atmospheric Environment 26A, 2269– 2282. University of Ghana http://ugspace.ug.edu.gh 171 132. Pant, P. and Harrison, R.M. (2012). Critical review of receptor modelling for particulate matter: A case study of India. Atmospheric Environment. 49, 1–12. 133. Polissar, A. V., Hopke, P. K., Paatero, P. (1998). Atmospheric aerosol over Alaska-2. Elemental composition and sources. Journal of Geophysical Research Atmospheres. 103, 19045–19057. 134. Pollice, A. (2009). Recent statistical issues in multivariate receptor models. Environmetrics 22, 35–41. 135. Pongkiatkul, P., and Oanh, N. T. K. (2012). Receptor Modeling for Air Pollution Source Apportionment Study. Integrated Air Quality Management: Asian Case Studies, 63. 136. Pope, C. A., Burnett, R. T., Thun, M. J., Calle, E. E., Krewski, D., Ito, K., Thurston, G. D. (2002). Lung cancer, cardio-pulmonary mortality and long-term exposure to fine particulate air pollution. Journal American Medical Association 287, 1132–1141. 137. Pope, C. A. and Dockery, D. W. (1992). Acute Health Effects of PM10 Pollution on Symptomatic and Asymptomatic Children. Am. Rev. Respir. Dis. 145, 1123- 1128. 138. Posfai, M. and Molnar, A., (2000). Atmospheric aerosol particles: A mineralogical introduction. In EMU Notes in Mineralogy, Vol. 2, Vaughan, D. J. and Wogelius, R. A. (Eds.), Eotvos University Press, Budapest, pp. 197–252. 139. Pui, D.Y.and Liu, B.Y. (1988). Advances in instrumentation for atmospheric aerosol measurement. Physica Scripta. 37, 252–269. 140. Qian, Z., Zhang, J., Wei, F., Wilson, W.E. and Chapman, R.S. (2001). Long-term ambient air pollution levels in four Chinese cities: inter-city and intra-city University of Ghana http://ugspace.ug.edu.gh 172 concentration gradients for epidemiological studies. J. Expo. Anal. Environ. Epidemiol. 11, 341–51. 141. Raes, F., Bates, T., McGovern, F. and Van Liedekerke, M., (2000). The second Aerosol Characterization Experiment (ACE-2): General overview and main results. Tellus 52, 111–125. 142. Rajkumar, W.S. and Chang, A.S. (2000). Suspended particulate matter concentrations along the East-West Corridor, Trinidad, West Indies. Atmospheric Environment 34, 1181-1187. 143. Ray, M. R., Basu, C. and Lahiri, T. (2003). Haematogilogical changes and up- regulation of p-selection expression in circulating platelets if Indian women chronically exposed to emissions from biomass fuels. Comp. Clin. Path., 12:151– 154. 144. Ricci, P. F., Catalano, K. M. D., (1996). Time series (1963-1991) of mortality and ambient air pollution in California: an assessment with annual data. Inhalation Toxicology 9, 95-106. 145. Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., and Simoneit, B.R.T. (1993). Sources of Fine Organic Aerosol. 3. Road dust, tire debris, and Organometallic Brake Lining Dust: Roads as Sources and Sinks. Environ. Sci. Technol. 27, 1892–1904. 146. Salm, S. J., and Falola, T. (2002). Culture and customs of Ghana. Greenwood Publishing Group. 147. Sarnat, J. A., Demokritou, P., Koutrakis, P. (2003). Measurement of fine, coarse and ultrafine particles. Ann Ist Super Sanità 39(3), 351-355. University of Ghana http://ugspace.ug.edu.gh 173 148. Schauer, J.J., Lough, G.C., Shafer, M.M., Christensen, W.F., Arndt, M.F., DeMinter, J.T., Park, J.S. (2006). Characterization of Metals Emitted from Motor Vehicles. Research Report 133. Health Effects Institute, Boston, MA. 149. Schauer, J.J., Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R. (1996). Source apportionment of airborne particulate matter using organic compounds as tracers. Atmospheric Environment 30, 3837-3855. 150. Schwela, D. (2010). Global Atmospheric Pollution (GAP) Forum Air Pollution Monitoring Manual, Stockholm Environment Institute (SEI). http://sei- international.org/rapidc/gapforum/html/reports /Techical/Monitoring /GAP_Forum_Monitoring_Manual%20_2010.pdf 151. Seinfeld, J. H., Pandis, S. N. (1998). Atmospheric Chemistry and Physics: From Air Pollution to Climate Change. Wiley, New York, pp. 51, 429–433. 152. Sillanpää, M. (2006). Chemical and source characterization of size-segregated urban air particulate matter, PhD Thesis, Department of Chemistry, University of Helsinki, Finland. 153. Smith, K. R. and Akbar, S. (2003). Health-damaging air pollution: a matter of scale. In: McGranahan, G., Murray, F. (Eds.), Air Pollution and Health in Rapidly Developing Countries. Earthscan, Canada. 154. Smith, S. J., Van Aardenne, J., Klimont, Z., Andres, R. J., Volke, A. and Delgado Arias, S. (2011). Anthropogenic sulfur dioxide emissions: 1850–2005. Atmos. Chem. Phys, 11(3), 1101-1116. 155. Song, Y., Xie, S., Zhang, Y., Zeng, L., Salmon, L.G., Zheng, M., (2006). Source apportionment of PM2.5 in Beijing using principal component analysis/absolute principal component scores and UNMIX. Science of the Total Environment 372, 278–286. University of Ghana http://ugspace.ug.edu.gh 174 156. Songsore, J., McGranahan, G. (1998). The political economy of household environmental management: gender, environment and epidemiology. World Dev. 26(3), 395–412. 157. Stelson, W. T. and Seinfeld, J. H. (1982). Relative humidity and temperature dependence of the ammonium nitrate dissociation constant. Atmospheric Environment 16, 983–992. 158. Taylor, S.R., McLenna, S., (1985). The continental crust: Its composition and evolution. Geoscience Texts. Blackwell Scientific Publications. 159. Tecer, L. H., Süren , P., Alagha, O., Karaca, F. and l Tuncel, G. (2008). Effect of Meteorological Parameters on Fine and Coarse Particulate Matter Mass Concentration in a Coal-Mining Area in Zonguldak, Turkey, Journal of the Air and Waste Management Association, 58(4), 543–552. doi:10.3155/1047- 3289.58.4.543. 160. Thorpe, A. and Harrison, R.M. (2008). Sources and properties of non-exhaust particulate matter from road traffic: a review. Sci. Total. Environ. 400, 270–282. 161. Tsai, Y.I. and Cheng, M.T. (2004). Characterization of Chemical Species in Atmospheric Aerosols in a Metropolitan Basin. Chemosphere. 54, 1171-1181. 162. UN Habitat (2010). The State of African Cities 2010: Governance, Inequality and Urban Land Markets. United Nations Human Settlements Programme, Nairobi. 163. UNEP (2012). Global Environmental Outlook 5. United Nations Environment Programme, Nairobi. 164. UNFPA (2007). State of world population 2007 - Unleashing the Potential of Urban Growth. University of Ghana http://ugspace.ug.edu.gh 175 165. UN-Habitat (2010). The State of African Cities 2010: Governance, Inequality and Urban Land Markets. United Nations Human Settlements Programme, Nairobi. 166. United Nations Environment Program (UNEP) (2010). Urban Air Pollution. http://www.unep.org/urban_environment/issues/urban_air.asp. Last accesses 14/06 /2013. 167. US EPA (1996). Air quality criteria for particulate matter. US Environmental Protection Agency, Research Triangle Park, NC. 168. US EPA (1999b). Chemical species analysis of filter-collected suspended particulate matter (SPM).Compendium of methods for the determination of inorganic compounds in ambient air, chapter IO-3. EPA/625/R-96/010a. Office of Research and Development. U.S. Environmental Protection Agency, Cincinnati, OH. http://www.epa.gov/ttn/amtic/files/ambient/inorganic/overvw3.pdf. Last accessed 08/17/2013. 169. US EPA, (1999a). Determination of metals in ambient particulate matter using neutron activation analysis (NAA) gamma spectrometry. Compendium of methods for the determination of inorganic compounds in ambient air. Compendium Method IO-3.7. EPA/625/R-96/010a. Office of Research and Development. U.S. Environmental Protection Agency, Cincinnati, OH. http://www.epa.gov/ttn/amtic/files/ambient/inorganic/mthd-3-7.pdf. Last accessed 08/17/2013. 170. US-EPA (2002). Third External Review Draft of Air Quality Criteria for Particulate Matter April 2002 (EPA/600/p-99/002aC). Washington, DC: US EPA Office of Research and Development. 171. USEPA (2004). Air Quality Criteria for Particulate Matter. Washington DC, USA: United States Environmental Protection Agency. University of Ghana http://ugspace.ug.edu.gh 176 172. USEPA (2004). EPA CMB 8.2 User Manual. USEPA Office of Air Quality Planning and Standards. 173. USEPA (2007). EPA Unmix 6.0 Fundamentals and User Guide. USEPA Office of Research and Development. 174. USEPA (2008). EPA Positive Matrix Factorization (PMF) 3.0 Fundamentals and User Guide, USEPA Office of Research and Development. 175. US-EPA. (1999). Compendium Method IO-3.3: Determination of Metals in Ambient Particulate Matter Using X-ray Fluorescence (XRF) Spectroscopy. Center for Environmental Research Information, Office of Research and Development, U.S. Environmental Protection Agency. Cincinnati, OH. http://epa.gov/ttn/amtic /files/ambient/inorganic/mthd-3-3.pdf. Last accessed 08/17/2013. 176. Vallero, D. (2008). Fundamentals of Air Pollution. 4th Edition. Elsevier Academic Press. Amsterdam. 177. Vallius, M. (2005). Characteristics and sources of fine particulate matter in urban air. PhD Thesis. National Public Health Institute, University of Kuopio, Helsinki, Finland. 178. van Vliet, E. and Kinney, P. L. (2007). Impacts of roadway emissions on urban particulate matter concentrations in sub-Saharan Africa: New evidence from Nairobi, Kenya. Environ. Res. Lett. 2. doi:10.1088/1748-9326/2/4/045028. 179. Vardoulakis, S., Fisher, B.E.A., Pericleous, K., Gonzalez-Flesca, N. (2003). Modelling air quality in street canyons: a review. Atmos Environ 37, 155-182. University of Ghana http://ugspace.ug.edu.gh 177 180. Vehlow, J., Bergfeldt, B., Hunsinger, H., Seifert, H. and Mark, F. E. (2003). Bromine in waste incineration: partitioning and influence on metal volatilisation Environ. Sci. Pollut. Res.Int. 10, 329–334. 181. Villeneuve, P. J., Burnett, R. T., Shi, Y., Krewski, D., Goldberg, M. S., Hertzman, C., Chen, Y., Brook, J. (2003). A time-series study of air pollution, socioeconomic status, and mortality in Vancouver, Canada. Journal of Exposure Analytical Environmental Epidemiology 13, 427–435. 182. Wahlin, P., Berkowicz, R., Palmgren, F. (2006). Characterisation of traffic- generated particulate matter in Copenhagen. Atmospheric Environment. 40 (12), 2151 – 2159. 183. Wang, G., Wang, H., Yu, Y., Gao, S., Feng, J., et al. (2003). Chemical characterization of water-soluble components of PM10 and PM25 atmospheric aerosols in five locations of Nanjing, China. Atmospheric Environment. 37:2893– 2902. 184. Watson, J.G. (1984). Overview of receptor model principles. J. Air Pollut. Control Assoc. 34(6), 619–623. 185. Watson, J.G. (2002). Visibility: science and regulation. Journal of the Air and Waste Management Association 52, 628–713. 186. Watson, J.G. and Chow, J.C. (2004). Receptor models for air quality management. EM 9(Oct.), 27–36. 187. Watson, J.G., Chow, J.C., Frazier, C.A., (1999). X-ray fluorescence analysis of ambient air samples. In: Landsberger, S., Creatchman, M. (Eds.), Elemental Analysis of Airborne Particles. Gordon and Breach, Newark, NJ, pp. 67–96. University of Ghana http://ugspace.ug.edu.gh 178 188. Whitby, K.T., Svendrup, G.M. (1980). California aerosols: Their physical and chemical characteristics. Adv. Environ. Sci. Tech. 10: 477. 189. WHO (2009). Global Health Risks: Mortality and Burden of Disease Attributable to Selected Major Risks. World Health Organization, Geneva. http://whqlibdoc.who.int/publications/2009/9789241563871_eng.pdf. Last accessed 12/03/2013. 190. WHO (2012). Air quality and health http://www.who.int/mediacentre/factsheets/fs313/en/index.html. Last accessed 02/12/2013. 191. Wilson, R. and Spengler, J.D., (1996). Particles in Our Air: Concentrations and Health Effects, Harvard University Press: Harvard. 192. Wilson, W.E., Chow, J.C., Claiborn, C., Wei, F.S., Engelbrecht, J., Watson, J.G. (2002). Monitoring of particulate matter outdoors. Chemosphere. 49, 1009–1043. 193. Winchester, J. W. (1981). Fine and coarse aerosol composition from a rural area in north China. Atmospheric Environment. 15, 933-937. 194. World Bank (2012). Review of urban air quality in Sub-Saharan Africa region - air quality profile of SSA countries. http://climatechange.worldbank.org/reports/review-urban-air-quality-sub- saharan-africa-region-air-quality-profile-ssa-countries. Last accessed 02/04/2012. 195. World Health Organization (WHO). (2005) Air quality guidelines. http://www.euro.who.int/_data/assets/pdf_file/0005/78638/E90038.pdf. 196. Wu, C.F., Larson, T.V., Wu, S.Y., Williamson, J., Westberg, H.H. and Liu, L.J.S. (2007). Source apportionment of PM2.5 and selected hazardous air pollutants in Seattle. Sci. Total Environ. 386, 42–52. University of Ghana http://ugspace.ug.edu.gh 179 197. Ye, B. M., Ji, X. L., Yang, H. Z., Yao, X. H., Chan, C. K., Cadle, S. H., Chan, T., and Mulawa, P. A. (2003). Concentration and chemical composition of PM2.5 in Shanghai for a 1-year period. Atmos. Environ. 37(4), 499–510. 198. Yele, S., Guoshun, Z., Wenjie, Z., Ying,W., Yahui, Z. (2006). Characteristics and sources of lead pollution after phasing out leaded gasoline in Beijing. Atmospheric Environment. 40 (16), 2973–2985. 199. Yongjie,Y., Yuesi, W., Tianxue, W., Wei, L., Ya'nan, Z., Liang, L., (2009). Elemental composition of PM2.5 and PM10 at Mount Gongga in China during 2006. Atmospheric Research. 93, 801–810 200. Zhang, Y.X., Quraishi, T. and Schauer, J.J. (2008). Daily Variations in Sources of Carbonaceous Aerosol in Lahore, Pakistan during a High Pollution Spring Episode. Aerosol Air Qual. Res. 8,130–146. 201. Zhou, Z., Dionisio, K.L., Arku, R.E., Quaye, A., Hughes, A.F., Vallarino, J., Spengler, J.D., Hill, A., Agyei-Mensah, S., and Ezzati, M. (2011). Household and community poverty, biomass use, and air pollution in Accra, Ghana. Proc. Natl Acad. Sci. USA 108, 11028–33. 202. Zhou, Z., Dionisio, K.L., Verissimo, T.G., Kerr, A.S., Coull, B., Arku, R.E., Koutrakis, P., Spengler, J.D., Hughes, A.F., Vallarino, J., Agyei-Mensah, S. and Ezzati, M. (2013). Chemical composition and Sources of Particle Pollution in Affluent and Poor Neighbourhoods of Accra, Ghana. Environmental Research Letters, (8) doi:10.1088/1748-9326/8/4/044025. 203. Zhou, Z., Dionisio, K.L., Verissimo, T.G., Kerr, A.S., Coull, B., Howie, S., Arku, R.E., Koutrakis, P., Spengler, J.D., Fornance, K., Hughes, A.F., Vallarino, J., Agyei-Mensah, S. and Ezzati, M. (2014). Chemical Characterization and Source Apportionment of Household Fine Particulate Matter in Rural, Peri-urban, and Urban West Africa. Environmental Science & Technology 48 (2), 1343-1351. University of Ghana http://ugspace.ug.edu.gh 180 APPENDIX A Table A1: Comparison between measured and PMF calculated PM2.5 concentrations (in ng m-3) except for PM (in μg m-3) at AD-R site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM2.5 Mass (μg m -3) 27.29 27.63 0.89 1.2 Elements Na 222.16 243.68 0.81 8.8 Mg 79.74 103.19 0.40 22.7 Al 519.06 532.08 0.99 2.4 Si 1132.97 1189.98 0.99 4.8 P 15.09 15.42 0.74 2.2 S 663.33 662.28 0.95 0.2 Cl 242.89 246.54 0.99 1.5 K 568.67 585.63 0.87 2.9 Ca 241.89 242.20 0.97 0.1 Ti 33.05 33.14 0.99 0.3 V 1.10 1.36 0.27 18.9 Cr 1.28 1.67 0.09 23.0 Mn 6.68 6.81 0.96 1.9 Fe 370.15 364.51 0.99 1.5 Cu 4.58 4.66 0.95 1.8 Zn 35.19 37.08 0.78 5.1 Br 14.00 14.49 0.93 3.4 Sr 3.59 4.47 0.36 19.7 Zr 2.81 3.43 0.29 18.0 Pb 12.82 14.25 0.67 10.0 University of Ghana http://ugspace.ug.edu.gh 181 Table A2: Comparison between measured and PMF calculated PM2.5 concentrations (in ng m-3) except for PM (in μg m-3) at EL-R site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM2.5 Mass (μg m -3) 21.98 22.26 0.92 1.24 Elements Na 186.85 220.83 0.49 15.39 Mg 126.45 144.28 0.71 12.36 Al 620.74 643.31 0.99 3.51 Si 1341.27 1430.32 0.98 6.23 P 13.80 14.26 0.86 3.24 S 632.79 638.95 0.94 0.96 Cl 144.61 145.24 0.99 0.44 K 602.54 619.18 0.90 2.69 Ca 245.24 246.60 0.99 0.55 Ti 48.45 49.19 0.99 1.52 V 0.85 0.90 0.68 6.06 Cr 0.98 1.22 0.14 20.02 Mn 8.66 9.31 0.71 6.96 Fe 426.87 411.95 0.99 3.62 Cu 1.40 1.97 0.09 28.82 Zn 12.26 17.60 0.00 30.34 Br 21.46 21.64 1.00 0.83 Sr 2.49 3.43 0.22 27.29 Zr 1.38 2.48 0.19 44.52 Pb 9.69 11.28 0.38 14.12 University of Ghana http://ugspace.ug.edu.gh 182 Table A3: Comparison between measured and PMF calculated PM2.5 concentrations (in ng m-3) except for PM (in μg m-3) at JT-R site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM2.5 Mass (μg m -3) 48.85 49.30 0.88 0.92 Elements Na 420.91 455.63 0.86 7.62 Mg 148.93 165.69 0.39 10.12 Al 671.00 688.92 1.00 2.60 Si 1476.88 1553.64 0.99 4.94 P 16.80 17.93 0.17 6.26 S 799.67 833.33 0.88 4.04 Cl 1408.42 1442.96 0.95 2.39 K 1831.38 1840.75 0.90 0.51 Ca 372.07 380.41 0.95 2.19 Ti 41.97 42.43 1.00 1.08 V 0.97 1.04 0.59 6.50 Cr 1.19 1.39 0.31 14.06 Mn 7.61 7.92 0.96 4.01 Fe 407.33 399.89 0.99 1.86 Cu 4.59 4.60 0.98 0.40 Zn 38.89 44.15 0.17 11.93 Br 19.69 19.92 0.98 1.17 Sr 4.34 5.05 0.24 14.09 Zr 2.92 3.46 0.17 15.81 Pb 11.11 11.87 0.59 6.40 University of Ghana http://ugspace.ug.edu.gh 183 Table A4: Comparison between measured and PMF calculated PM2.5 concentrations (in ng m-3) except for PM (in μg m-3) at NM-R site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM2.5 Mass (μg m -3) 26.78 27.27 0.73 1.83 Elements Na 217.42 271.01 0.62 19.77 Mg 104.58 131.99 0.29 20.76 Al 550.51 569.41 0.99 3.32 Si 1191.70 1258.44 0.98 5.30 P 12.42 13.23 0.53 6.09 S 606.20 622.38 0.92 2.60 Cl 354.68 359.07 0.99 1.22 K 815.22 842.66 0.73 3.26 Ca 277.81 282.13 0.96 1.53 Ti 36.02 35.80 1.00 0.60 V 0.72 0.80 0.50 10.26 Cr 0.86 1.08 0.23 20.57 Mn 5.42 5.60 0.96 3.29 Fe 370.12 361.32 0.99 2.44 Cu 5.21 5.29 0.96 1.48 Zn 22.87 27.65 0.13 17.31 Br 17.65 20.50 0.73 13.89 Sr 3.59 4.65 0.16 22.65 Zr 1.47 2.10 0.19 29.77 Pb 12.34 13.50 0.58 8.53 University of Ghana http://ugspace.ug.edu.gh 184 Table A5: Comparison between measured and PMF calculated PM2.5 concentrations (in ng m-3) except for PM (in μg m-3) at NM-T site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM2.5 Mass (μg m -3) 33.23 34.04 0.79 2.38 Elements Na 152.85 181.10 0.57 15.60 Mg 76.92 90.20 0.65 14.72 Al 776.89 795.17 1.00 2.30 Si 1597.07 1701.65 0.98 6.15 P 17.11 17.69 0.80 3.28 S 650.16 661.65 0.96 1.74 Cl 376.51 378.97 0.99 0.65 K 899.13 929.89 0.57 3.31 Ca 434.54 450.53 0.88 3.55 Ti 54.04 54.11 0.99 0.13 V 0.95 1.02 0.63 6.64 Cr 1.16 1.40 0.19 17.30 Mn 7.99 8.23 0.93 2.92 Fe 539.81 536.69 0.98 0.58 Cu 4.82 4.95 0.97 2.68 Zn 29.00 32.08 0.08 9.61 Br 16.89 16.94 0.99 0.28 Sr 2.14 2.89 0.48 26.19 Zr 1.58 1.96 0.51 19.57 Pb 12.28 14.67 0.60 16.29 University of Ghana http://ugspace.ug.edu.gh 185 Table A6: Comparison between measured and PMF calculated PM10 concentrations (in ng m-3) except for PM (in μg m-3) at AD-R site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM10 Mass (μg m -3) 55.14 55.65 0.97 0.91 Elements Na 959.38 984.60 0.96 2.56 Mg 272.62 287.18 0.92 5.07 Al 1627.32 1654.45 1.00 1.64 Si 4016.63 4137.18 0.99 2.91 P 25.41 25.92 0.90 1.99 S 838.21 861.83 0.91 2.74 Cl 2026.10 2030.75 0.98 0.23 K 857.41 886.20 0.92 3.25 Ca 1113.91 1118.34 0.97 0.40 Ti 124.97 125.60 0.99 0.50 V 2.17 2.28 0.79 4.94 Cr 3.10 3.26 0.54 4.97 Mn 19.15 19.55 0.99 2.01 Fe 1505.08 1491.33 0.97 0.92 Cu 8.45 8.50 0.98 0.55 Zn 48.24 52.86 0.29 8.74 Br 20.42 20.48 1.00 0.30 Sr 3.65 3.68 0.98 0.96 Zr 2.11 2.19 0.93 3.61 Pb 13.74 18.92 0.22 27.36 University of Ghana http://ugspace.ug.edu.gh 186 Table A7: Comparison between measured and PMF calculated PM10 concentrations (in ng m-3) except for PM (in μg m-3) at EL-R site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM10 Mass (μg m -3) 44.71 45.00 0.97 0.64 Elements Na 545.03 569.90 0.96 4.36 Mg 229.67 248.34 0.93 7.52 Al 1830.21 1834.19 1.00 0.22 Si 4176.72 4282.65 0.99 2.47 P 20.22 21.76 0.39 7.07 S 689.03 694.63 0.97 0.81 Cl 1165.27 1160.59 0.98 0.40 K 832.26 850.76 0.94 2.17 Ca 900.53 910.85 0.98 1.13 Ti 129.04 129.32 0.99 0.22 V 2.48 2.57 0.76 3.44 Cr 3.02 3.14 0.56 3.90 Mn 16.16 16.54 0.99 2.31 Fe 1459.67 1462.86 0.94 0.22 Cu 1.99 2.49 0.10 20.01 Zn 24.96 25.12 0.99 0.67 Br 23.04 28.28 0.67 18.54 Sr 6.17 7.28 0.41 15.33 Zr 4.22 4.73 0.43 10.80 Pb 18.99 20.06 0.91 5.31 University of Ghana http://ugspace.ug.edu.gh 187 Table A8: Comparison between measured and PMF calculated PM10 concentrations (in ng m-3) except for PM (in μg m-3) at JT-R site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM10 Mass (μg m -3) 81.92 82.42 0.94 0.60 Elements Na 1940.70 1965.98 0.95 1.29 Mg 439.68 455.07 0.94 3.38 Al 1707.95 1720.22 1.00 0.71 Si 4225.87 4310.29 1.00 1.96 P 29.53 30.21 0.96 2.26 S 1059.66 1114.95 0.64 4.96 Cl 4910.24 4929.92 0.96 0.40 K 2173.59 2215.20 0.84 1.88 Ca 1617.07 1641.68 0.97 1.50 Ti 129.84 130.12 0.99 0.21 V 1.82 1.94 0.69 6.22 Cr 3.44 3.51 0.64 2.09 Mn 20.59 20.81 0.99 1.03 Fe 1298.71 1291.06 0.99 0.59 Cu 6.26 7.25 0.33 13.64 Zn 60.99 67.07 0.29 9.06 Br 22.84 22.72 1.00 0.50 Sr 5.26 6.04 0.29 12.84 Zr 4.68 5.16 0.51 9.33 Pb 15.74 16.08 0.96 2.09 University of Ghana http://ugspace.ug.edu.gh 188 Table A9: Comparison between measured and PMF calculated PM10 concentrations (in ng m-3) except for PM (in μg m-3) at NM-R site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM10 Mass (μg m -3) 50.47 50.87 0.94 0.78 Elements Na 656.98 686.73 0.96 4.33 Mg 224.80 248.35 0.86 9.48 Al 1655.98 1679.48 1.00 1.40 Si 3895.90 4001.70 0.98 2.64 P 22.71 23.65 0.66 3.97 S 732.61 750.36 0.94 2.37 Cl 1699.05 1696.07 0.98 0.18 K 1101.36 1111.18 0.91 0.88 Ca 1181.10 1182.53 0.97 0.12 Ti 124.66 124.10 0.99 0.45 V 2.08 2.14 0.83 2.78 Cr 2.93 3.10 0.48 5.77 Mn 17.17 17.42 0.99 1.45 Fe 1369.62 1346.08 0.98 1.75 Cu 5.36 6.31 0.14 15.05 Zn 39.03 39.33 0.98 0.78 Br 22.23 25.01 0.74 11.11 Sr 5.90 6.97 0.35 15.26 Zr 3.64 5.00 0.14 27.24 Pb 15.66 17.45 0.47 10.28 University of Ghana http://ugspace.ug.edu.gh 189 Table A10: Comparison between measured and PMF calculated PM10 concentrations (in ng m-3) except for PM (in μg m-3) at NM-T site for studied chemical species. Species PMF modelled Measured r2 Uncertainty PM10 Mass (μg m -3) 69.69 70.30 0.94 0.86 Elements Na 742.92 761.69 0.97 2.46 Mg 293.49 307.89 0.92 4.68 Al 2573.85 2590.71 0.99 0.65 Si 5650.19 5736.35 0.99 1.50 P 40.10 41.25 0.82 2.80 S 841.37 884.95 0.73 4.92 Cl 2133.14 2152.32 0.95 0.89 K 1283.18 1317.76 0.73 2.62 Ca 1869.12 1919.56 0.92 2.63 Ti 200.13 202.73 0.96 1.28 V 3.05 3.09 0.82 1.55 Cr 4.61 4.76 0.68 3.19 Mn 25.73 26.11 0.98 1.46 Fe 2020.19 2033.75 0.97 0.67 Cu 8.92 8.97 1.00 0.50 Zn 51.02 53.96 0.51 5.44 Br 29.98 30.01 1.00 0.09 Sr 4.68 5.94 0.21 21.30 Zr 3.04 4.34 0.09 29.85 Pb 13.73 15.19 0.28 9.61 University of Ghana http://ugspace.ug.edu.gh