UNIVERSITY OF ZIMBABWE

FACULTY OF ENGINEERING DEPARTMENT OF CIVIL ENGINEERING

Integration of physicochemical assessment of water quality with Remote sensing techniques for the Dikgathong Dam in

By

KAGISO MOSIMANEGAPE

M.Sc. THESIS IN IWRM

HARARE, MAY 2016

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

In collaboration with

UNIVERSITY OF ZIMBABWE

DEPARTMENT OF CIVIL ENGINEERING

Integration of physicochemical assessment of water quality with Remote sensing techniques for the Dikgathong Dam in Botswana

By

KAGISO MOSIMANEGAPE

MASTER OF SCIENCE THESIS IN INTEGRATED WATER RESOURCES MANAGEMENT

A thesis submitted in partial fulfilment of the requirements for the Degree of Master of Science in Integrated Water Resources Management of the University of Zimbabwe

SUPERVISORS: Eng. Z. Hoko Mr W. Gumindoga

JULY 2016 Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

DECLARATION I, Kagiso Mosimanegape, declare that this research report is my own work. It is submitted for the degree of Master of Science in Integrated Water Resources Management (IWRM) at the University of Zimbabwe. It has not been submitted before for any other degree of examination at any other University. The findings, interpretations and conclusions expressed in this study neither reflect the views of the University of Zimbabwe, Department of Civil Engineering nor those of the individual members of the MSc Examination Committee, nor of their respective employers.

Signature: ______

Date: ______

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE I

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

CONTENTS DECLARATION ...... i TABLE OF CONTENTS ...... ii LIST OF TABLES ...... vi APPENDICES ...... vii LIST OF ABBREVIATIONS AND ACRONYMS ...... viii ACKNOWLEDGEMENTS ...... ix ABSTRACT ...... x CHAPTER 1 ...... 1 1.0 Introduction ...... 1 1.1 Background ...... 1 1.2 Problem Statement ...... 3 1.3 Justification ...... 4 1.4 Objectives ...... 5 1.4.1 Main objective ...... 5 1.4.2 Specific objectives ...... 5 2.0 LITERATURE REVIEW ...... 6 2.1 Water quality monitoring ...... 6 2.2.2 Turbidity and Total Suspended Solids ...... 7 2.2.3 Electrical conductivity and Total Dissolved Solids ...... 8 2.2.4 Total Hardness, Calcium and Magnesium ...... 8 2.2.5 Sodium and Potassium ...... 8 2.2.6 Chloride ...... 9 2.2.7 Nitrates and Total Nitrogen ...... 9 2.2.8 Total Phosphate and Orthophosphate ...... 9 2.2.9 Sulphates ...... 10 2.2.10 Alkalinity ...... 10 2.2.11 Chemical Oxygen Demand ...... 10 2.2.12 Algae ...... 11 2.3 Water quality challenges ...... 11 2.3.1 Agricultural use of fertilizer and runoff ...... 11 2.3.2 Population growth, uncontrolled disposal of human and industrial wastes ...... 12

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE II

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

2.3.3 Mining activities ...... 12 2.4 Land use impacts on water quality ...... 13 2.5 Climate change ...... 13 2.6 Remote sensing of water quality ...... 14 2.6.1 Availability of satellite data ...... 15 2. 6.2 Empirical Determination of water quality variables ...... 16 2.6.3 Analytical models for water quality monitoring ...... 16 Quasi Analytical Algorithm ...... 16 CHAPTER 3 ...... 20 3.0 STUDY AREA ...... 20 3.1 Location ...... 20 3.1.1 Population ...... 21 3.1.2 Rainfall ...... 21 3.1.3 Soils, geology and vegetation ...... 21 3.1.4 Socio-economic issues ...... 21 3.2 Potential point and non-point sources of pollution in the Dikgathong Catchment Area...... 21 CHAPTER 4 ...... 23 4.0 MATERIALS AND METHODS ...... 23 4.1 Land use and Land cover Classification for 2010 and 2015 ...... 23 4.1.1 Image acquisition and processing...... 23 4.1.2 Validation of the classification output ...... 23 4.2 Physicochemical analysis of water quality parameters ...... 24 4.2.1 Selection of study site ...... 24 4.2.2 Selection of sampling sites ...... 24 4.2.3 Selected of parameters to be analysed ...... 26 4.2.4 Methods of sampling and frequency ...... 26 4.2.5 Water quality testing ...... 28 4.3 Near real-time retrieval of Chl_a and TSM using Quasi Analytical Algorithms (QAA) ...... 28 4.3.1 Downloading MODIS images ...... 28 4.3.2 Image processing and extraction of reflectance values at specific points ...... 29 4.3.3 Input of MODIS reflectance into the QAA ...... 29 4.3.4. The Water Colour Simulator (WASI) ...... 30

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE III

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

4.4 Relationship between satellite and ground measured water quality parameters ...... 31 4.5 Methods of data analysis and interpretation ...... 32 4.5.1 Analysis of the water quality parameters ...... 32 A one-way Analysis of Variance (ANOVA) ...... 32 Cluster Analysis ...... 32 Principal Component Analysis ...... 33 5.0 RESULTS AND DISCUSSION ...... 34 5.1.0 Landuse classification ...... 34 5.1.1 Validation of the classification output ...... 37 5.2 Water quality of the Dikgathong Dam ...... 38 5.1.1 Chemical Oxygen Demand ...... 38 5.1.2 Electrical Conductivity ...... 39 5.1.3 Turbidity ...... 40 5.1.4 Total Suspended Solids ...... 41 5.1.5 Total Hardness ...... 42 5.1.6 Calcium ...... 43 5.1.7 Magnesium ...... 43 5.1.8 Total Alkalinity ...... 44 5.1.9 Nitrates ...... 45 5.1.10 Sulphates ...... 45 5.2 Water quality parameters retrieved from MODIS images ...... 50 5.3 Relationship between satellite and ground measured water quality parameters ...... 53 CHAPTER 6 ...... 56 6.0 CONCLUSIONS AND RECOMMENDATIONS ...... 56 6.1 CONCLUSIONS ...... 56 6.2 RECOMMENDATIONS ...... 56 REFERENCES ...... 58

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE IV

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

LIST OF FIGURES

Figure 3.0: The Dikgathong Dam and its tributaries...... 20 Figure 4.0. Map illustrating sampling points for the Dikgathong Dam...... 25 Figure 4.1: Some of the equipment used include cooler box and polypropylene bottles...... 27 Figure 4.2: Water field meter YSI 650 for measurement of physical water quality parameters...... 28 Figure 5.0 Classified thematic map for 2010 and 2015 ...... 34 Figure 5.1: 2010 and 2015 land use classification ...... 35 Figure 5.2: Spatial variation of COD in the Dikgathong Dam ...... 39 Figure 5.4: Spatial distribution of Turbidity in the Dam...... 41 Figure 5.5: Spatial variation of average values of TSS in the Dikgathong Dam ...... 42 Figure 5.6: Temporal variation of average values of TH in the Dikgathong Dam ...... 43 Figure 5.7: Temporal variation of average values of Ca in the Dikgathong Dam ...... 43 Figure 5.8: Temporal variation of average values of Mg in the Dikgathong Dam ...... 44 Figure 5.9: Temporal variation of average values of TA in the Dikgathong Dam ...... 44

Figure 5.10: Temporal variation of averages values of NO3 in the Dikgathong Dam ...... 45 Figure 5.11: Temporal variation of average values of SO4 in the Dikgathong Dam ...... 46 Figure 5.12: Spatial variation of Chl_a in the Dikgathong Dam ...... 52 Figure 5.13: Spatial variation of TSS in the Dikgathong Dam ...... 53

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE V

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

LIST OF TABLES Table 2.0 Satellite information of different selected sensors...... 15 Table 2.1 Steps of the QAA to Derive Absorption and Backscattering Coefficients ...... 18 Table 4.0 Coordinates of sampling points in the Dikgathong Dam ...... 25 Table 4.1: Analyzed parameters and method used ...... 26 Table 4.2 QAA wavelengths corresponding to MODIS band ...... 29 Table 5.0 Land use change in area from 2010 to 2015 ...... 35 Table 5.1 Results for LULC accuracy assessment ...... 37 Table 5.2: Descriptive statistics for average water quality parameters from ten sampling points ...... 38 Table 5.3: Cluster membership ...... 47 Table 5.4: Final cluster centers ...... Error! Bookmark not defined. Table 5.5: Eigenvalues for principal components ...... 49 Table 5.6: Contribution of the variables (%) ...... 49 Table 5.7: Derived IOPs and Chl_a for 10th March 2016 ...... 51 Table 5.8. Spearman’s correlation between chl_a and ground measured water quality ...... 54 Table 5.9. Spearman’s correlation between TSM and ground measured turbidity and TSS ...... 55

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE VI

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

APPENDICES LIST OF APPENDICES ...... 72 Appendix 1: Results of physical, chemical and microbiological sampled on the 07th April 2016 ...... 72 Appendix 2: Results of physical, chemical and microbiological sampled on the 23rd March 2016 ...... 72 Appendix 3: Results of physical, chemical and microbiological sampled on the 10th March 2016 ...... 73 Appendix 4: Results of physical, chemical and microbiological sampled on the 24th February 2016...... 73 Appendix 5: Results of physical, chemical and microbiological sampled on the 11th February 2016...... 74 Appendix 6: Results of physical, chemical and microbiological sampled on the 27th January 2016...... 74 Appendix 7: Results of physical, chemical and microbiological sampled on the15th January 2016...... 75 Appendix 8: Standards and Guidelines for surface waters ...... 75 Appendix 9: Botswana rainfall map ...... 76 Appendix 10: Rainfall data for airport from January to April 2016 ...... 77 Appendix 11: derived iops at 440nm and chl_a for 13th january 2016 ...... 78 Appendix 12: derived iops at 440nm and chl_a for 26th january 2016 ...... 78 Appendix 13: derived iops 440 nm and chl_a for 11th february 2016...... 79 Appenidx 14: derived iops and chl_a for 05th april 2016 ...... 80

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE VII

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

LIST OF ABBREVIATIONS AND ACRONYMS

BOBS Botswana Bureau of Standards DWA Department of Water Affairs EPA Environmental Protection Agency GWP Global Water Partnership IOP Inherent Optical Properties IWRM Integrated Water Resources Management LRB Limpopo River Basin MODIS Moderate Resolution Imaging Spectroradiometer PACN Pan African Chemistry Network SADC Southern African Development Community SPSS Statistical Package for Social Sciences UN United Nations USAID United States Agency International Development WHO World Health Organization WUC Water Utilities Corporation

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE VIII

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

ACKNOWLEDGEMENTS

I forward my appreciation to WaterNet for funding and making this research work possible. I extend my gratitude to my supervisors Eng. Z. Hoko and Mr W. Gumindoga for their guidance, advice, patience and encouragement throughout the research period. Many thanks go to the rest of the lecturers at University of Zimbabwe and University of Malawi; and to my employer, the Department of Water Affairs for their support, with a special gratitude extended to Mr Keletso Kgosana, who worked with me tirelessly throughout the entire sampling campaign.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE IX

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

ABSTRACT

Water quality has become a global concern due to ever increasing population and developmental activities that are polluting water resources. Botswana’s water resources are threatened by various pollution sources such as agricultural runoff, industrial and domestic effluents. This study was carried out to assess the water quality of Dikgathong Dam in Botswana using physicochemical analysis of water quality parameters and remote sensing techniques. The study first assessed landuse patterns before construction (2010) and after construction (2015) to establish dominant landuse in order to select water quality parameters related to the principal landuse in the catchment. Images for 2010 and 2015 were acquired from Landsat and were classified using the supervised classification through the Maximum Likelihood algorithm. Results showed that forest and shrubs were the dominant landuse covering 73.7 % of the total area, followed by settlements (21.1 %) and agricultural fields (2.76 %). Chl_a, COD, EC, TP, TN, TSS, NO3 and NO2 were selected for testing and analysis based on their relationship with forest, settlements and agricultural fields. For assessment of water quality, ten points were sampled in the dam from 15th January to 07th April 2016. Temperature, pH, EC, COD, TDS, TSS, turbidity, chloride, nitrates, sodium, potassium, calcium, magnesium, sulphates, phosphates, total phosphorus, alkalinity, hardness and algae were tested and analysed according to standard methods. Only COD, turbidity and TSS exceeded the limits set by Environmental Protection Agency (EPA) surface water standards of 2001, making Dikgathong Dam slightly polluted. One way ANOVA showed significant variations (p<0.05) between water quality values in all sampling points only for NO3, SO4, pH, algae and Na. Five different groups of sites were identified from ten sites using cluster analysis. The principal component analysis identified ten parameters (COD, EC, turbidity, TSS, Ca, Mg, NO3, SO4, total hardness and alkalinity) based on similarities of water quality characteristics. The Water Utilities Corporation, which is responsible for the dam, can therefore monitor water quality at five points focusing mainly on ten parameters found to be principal. This study also investigated the likelihood of integrating remote sensing and in-situ measurements to assess the water quality status of the dam. Quasi analytical algorithms and MODIS data were used to quantify Chl_a and TSS concentrations in the dam. Values for Chl_a were between 1.74 and 24.4 mg/m3, while TSS ranged from 2.34 mg/l to 59.2 mg/l. Based on chlorophyll concentrations the dam can be classified as

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE X

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

both oligotrophic and mesotrophic as per the EPA 2001 standard. The QAA and MODIS can therefore be deployed as a mechanism for near real time monitoring of water quality in Botswana reservoirs. Spearman’s correlation was used to test whether satellite retrieved water quality parameters relate to in-situ measurements. Strong positive significant correlation was observed

between chl_a and turbidity (r=0.794 and 0.830), TSS (r = 0.819 and 0.770), SO4 COD (r=0.781

and 0.769).), SO4 (r= 0.851 and 0.646) and alkalinity (r= 0.847). Moderate positive and non- significant relationship is observed for temp (r= 0.055), pH (r= 0.587), EC (r= 0.409), TDS (r=0.348), Na (r= 0.406) and Cl (r= 0.394). Strong positive and significant correlation was observed between remote sensing retrieved TSS and in-situ measured TSS (r= 0.733) and turbidity (r= 0.867). This study concludes that there is strong positive correlation between parameters retrieved through remote sensing and in-situ measurements and therefore can be used in monitoring and assessment of the water quality in the lake at any point in time.

Key Words: Dikgathong Dam, water quality, remote sensing, geographic information systems, land use, Quasi Analytical Algorithms.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE XI

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

CHAPTER 1 1.0 Introduction

1.1 Background

Population growth, urbanization and industrialization have led to the decline of quality of surface waters globally (Trevor, 2009; Martinez et al., 2011; Walakira and Okot-okumu, 2011; Owa, 2013). The quality of surface water has deteriorated in many countries in the past few decades due extensive anthropogenic inputs of nutrients and sediments (Tessema et al., 2014). Most rivers in the urban areas of the developing world are the end points of discharges from the municipal and industrial treatment facilities (Bernard, 2010; Suthar et al., 2010; Ljee, 2011) rendering surface waters the most polluted water resources (Kadewa et al., 2005). According to the Pan African Chemistry Network (PACN), Africa’s population exceeded 1 billion in 2009 and continues to increase at a rate of 2.4 % annually (PACN, 2010). Of this population, 341 million lack access to clean drinking water and a further 589 million have no access to adequate sanitation, resulting in loss of productivity due to water-related illnesses (PACN, 2010).

Water for human use require sustaining an adequate water quality standards and changes in water quality threatens human health (Massoud, 2012). The effects of environmental factors such as climate change make the challenge of conserving water resources even more difficult (PACN, 2010). Climate change has a major impact on water quality and water management. Increases in water temperature produce unfavourable changes in surface-water quality, which has detrimental effects on human and ecosystem health (IPCC, 2008).

Africa’s water resources are being degraded due to discharge of untreated wastewater from industrial and domestic sources (Corcoran et al., 2010). Pollution from natural and anthropogenic processes also threatens available fresh water resources in Southern Africa (Nyakungu et al., 2013). In Zimbabwe, Lake Chivero is polluted due to nutrient loadings from sewage discharges through its main tributary, Manyame River ( Masere et al., 2012; Nyakungu et al., 2013; Kibena et al., 2014). Ngerengere River in Wami/Ruvu basin in Tanzania is polluted due to agricultural and industrial wastewater from upstream sources (Mero, 2011) and with most affected communities located downstream of the catchment (Mero, 2011).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 1

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Botswana’s water resources are also under threat from various pollution sources that mainly include pit latrines, solid waste, agriculture, industrial and domestic wastes (DWA, 2013). River water quality is deteriorating due to pollutant loads from point and nonpoint sources, such as agricultural runoff, solid waste disposal, sewage and industrial effluent (Gbadegesin, 2015). Motloutse River has been polluted by discharges from mine effluent (Kgathi and Masamba, 2006). Notwane River is heavily polluted with sewage effluent from Glen Valley and Mochudi wastewater treatment plants (Mladenov et al., 2005) rendering water from these rivers unfit for human use and in some cases not conducive for irrigation and environmental consumption. The Dikgathong Dam, located within the Limpopo River Basin (LRB), in the north east part of Botswana, is also under pollution threat from anthropogenic activities (EHES, 2002). The LRB main pollution threat come from anthropogenic activities such as agriculture, mining and urbanization (MRC, 2009).

Water quality monitoring is essential for identifying sources of contaminants entering water resources (PACN, 2010). Monitoring offers key information for detecting and dealing with water quality problems, by identifying trends over time and comparisons between different water bodies (UNEP, 2010). The quality of water bodies is determined by its physicochemical features, hence the importance of long and short term analysis of water quality status (Adakole et al.,2003). There is inadequate information for water quality around the world, especially in developing countries. The little available information has challenges of inconsistency, rendering it not very useful for those who want to use it (UNEP, 2010). Water quality monitoring methods includes biological indicators, physicochemical analysis, water quality indices and use of remote sensing techniques.

The use of GIS and remote sensing in water quality monitoring has been long acknowledged (Usali and Ismail, 2010). The strength of remote sensing techniques is their capability to capture the spatial and temporal variability of water quality parameters (Mohamed, 2015) and has been demonstrated to be effective at reduced cost (Schaeffer et al., 2013). Traditional methods of water quality monitoring are costly and time consuming. Traditional methods provide accurate measurements because of the direct measurements, but it is only at discrete points, not covering the entire water body (He et al., 2008). Thus it is advocated for the use of GIS and remote sensing to improve water quality monitoring (Papoutsa and Hadjimitsis, 2008).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 2

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Besides climate change effects mentioned earlier, landuse changes have potential great impacts on water resources (Wagner et al., 2013). Land use classifications can be analysed to identify changes over a past period of time (Wagner et al., 2013). Understanding the relationship between upstream landuse and water quality is useful for identifying principal threats to water quality (Ding et al., 2015). Landuse usually defines the concentrations of contaminants that flow into rivers and lakes (Larkin, 2014). Pollution is related to anthropogenic practices which can be measured in terms of population density and land use type (Yaakub et al., 2012). It is important to constantly determine point and nonpoint source of pollution loads for proper management of water quality (Yaakub et al., 2012).

1.2 Problem Statement Urbanization, agricultural runoff, industrial and sewage disposal have affected the quality of water around the world and making it unfit for domestic purpose ( Kadewa et al., 2005; Gupta et al., 2009; Chatterjeeet al., 2010; Choudhary et al, 2011; Ullah et al., 2013). The Dikgathong Dam lies downstream from expanding urban, industrial and mining areas, which poses a threats to the dam’s water quality (EHES, 2002). It is imperative for the Dikgathong Dam to continue to sustain its water quality in order to serve its purpose of supplying drinking water. Measures need to be put in place to monitor the water quality of this dam with deliberate interventions to ensure it does not continue to deteriorate (EHES, 2002). The use of traditional methods for monitoring water quality alone is not adequate (Ritchie et al., 2003). Integration of traditional methods and remote sensing have demonstrated the practicality of remote sensing in monitoring of water quality ( Dube et al., 2014; Dlamini et al., 2016). Previous studies on remote sensing applications on water quality have used empirical algorithms for predicting water quality parameters. The limitations of empirical methods of remote sensing applications in water quality are that the algorithms work best when applied to the site where sampling data was collected and the formulas derived and may not be applicable to other water bodies (Chang et al., 2014).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 3

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

1.3 Justification Botswana is a semi-arid country and its water resources are increasingly scarce. Future sites for dam construction are exhausted and use of shared water courses requires SADC protocol (DWA, 2013). The Dikgathong Dam is the largest domestic water supplier in the country. EHES, (2002) calls for measures to be put in place to monitor the water quality of the Dikgathong Dam to ensure its water quality does not deteriorate so that the dam continues to sustain its water quality in order to serve its purpose of supplying drinking water to people and livestock. Regular monitoring of the Dam will reduce negative ecological impacts and potential water borne disease hence improve water resources protection. Clean and safe water from the Dam will improve social well-being and livelihoods of the nation and who will work towards improving economic efficiency of the country.

The Departments of Water Affairs for Botswana and South Africa, supported by the United States Agency International Development (USAID) carried a joint monitoring survey in the LRB in 2012. Findings of the report indicate that the LRB catchment is polluted with concentrations of heavy metals, which fuels infestation of water hyacinth (DWA, 2012). These report recommended for a more detailed study to ascertain the actual sources of pollution in the LRB, which should include physical inspections of all potential point and non-point sources of pollution.

The Southern African Development Community (SADC) has a protocol on shared watercourse systems, which promotes the prevention of pollution on shared watercourses which may have detrimental effects on the ecosystem. Therefore there is need for trans-boundary pollution prevention before the water is released to Limpopo river basin (shared by Botswana, SA, Mozambique and Zimbabwe) to satisfy agreements for SADC Protocol on Shared Watercourses Management.

The African Group in Earth Observations (AfriGEOSS) earth greatly supports the use to environmental resources including water. The AfriGEOSS advocates for application of remote sensing in water quality monitoring in Africa. Several authors have also supported the importance of integration traditional methods of water quality monitoring with remote sensing techniques (Hellweger et al., 2004; He et al., 2008).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 4

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

1.4 Objectives

1.4.1 Main objective

The main objective of this study is to integrate the physiochemical assessment of the water quality with remote sensing techniques in the Dikgathong Dam in Botswana.

1.4.2 Specific objectives

i. To establish land use changes before construction (2010) and after construction (2015).

ii. To analyse selected water quality parameters in the Dikgathong Dam to ascertain its suitability for raw water supply for drinking purposes. iii. To perform a near real-time retrieval of Chlorophyll a and Total Suspended Matter in the dam using remote sensing techniques for improved water quality monitoring. iv. To test whether satellite retrieved water quality parameters significantly relate with ground measured parameters for future estimation of water quality parameters.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 5

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

CHAPTER 2

2.0 LITERATURE REVIEW

2.1 Water quality monitoring Integrated Water Resources Management (IWRM) calls for consideration of both quality and quantity for water usage (Masamba and Mazvimavi, 2006). Contaminated water is detrimental to health of living organisms and the environment (Massoud, 2012; Singh, 2014). Polluted water also increases treatment cost for drinking water (Matta, 2014). Water quality monitoring is vital for the protection of freshwater resources (Kannel et al., 2007). Protection of freshwater resources can be achieved through regular water quality monitoring (Elbag, 2006). Regular water quality assessment of water resources is important to determine suitability and safety for varying purposes (WMO, 2012). Conducting water quality assessment begins with evaluation of the suitability of water based on its intended use (WSR, 2003). Water quality monitoring can assist water authorities to identify and quantify sources and impacts of pollution as well as communicate to stakeholders about pollution status (Kadewa et al., 2005; Garizi et al., 2011). Water quality monitoring can further help review of policies and assist management to prioritize plans for effective management (Walakira and Okot-okumu, 2011).

Several methods have been used in the field of water quality monitoring around the world. Many researchers have used physico-chemical parameters for assessing the water quality status of freshwater systems (Adakole et al., 2003; Mustapha, 2008; Araoye, 2009; Choudhary et al., 2011; Patil et al., 2012; Al-Anzi, 2012). Other researchers employed the use of water quality indices as the indicators of water pollution by assessing spatial and temporal changes and classification of river water quality (Kannel et al., 2007; Gibrilla et al., 2011; Massoud, 2012). Recently most authors have added the used remote sensing techniques for assessment of water quality (Alparslan et al., 2007; Rosado-Torres, 2008; Larkin, 2014; Zheng et al., 2014; Hansen et al., 2015).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 6

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

2.2 Physico-chemical and biological characteristic of water

The selection of water quality assessment parameters depend on the needs and objectives of the assessment (Quevaulviller at el., 2006). Descriptions of key water parameters is given.

2.2.1 pH pH, is a measure of the concentration of hydrogen ions in the water or a measure of how acidic or basic it is on a scale of 0 to 14, with 7 being neutral (Talling, 2010). Naturally occurring fresh waters have a pH range between 6.5 and 8.5. The pH of the water is important because it affects the solubility and availability of nutrients, and how they can be utilized by aquatic organisms (Stone et al., 2013). Surface water pH can be relatively higher in low discharge waters, since water is rich in solutes characteristic of ground water. The pH value can sensitively indicate variations in water quality and is affected by dissolved substances (WHO, 2006).

2.2.2 Turbidity and Total Suspended Solids

Turbidity and transparency of water is determined by the concentration and nature of Total Suspended Solids (TSS). TSS contains soluble organic compounds as well as fine particles of organic and inorganic matter (Matta, 2014). TSS and turbidity differs with time based on biological activity in the water system and type of sediments carried by surface run-off. Turbidity highly is influenced by rainfall at a particular point. Turbidity can be related to TSS hence turbidity can be used as an indirect measurement for TSS (Chapman, 1996). Turbidity refers to the quantity of suspended material, which interferes with light penetration in the water column. Higher turbidity can cause temperature and DO stratification in water bodies (Tessema et al., 2014). High TSS levels in surface water absorb heat from sunlight, which increases water temperature and decreases levels of dissolved oxygen (Iqbal et al., 2010). This results with the water body losing its ability to support aquatic life (Jo-Anne et al., 2014).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 7

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

2.2.3 Electrical conductivity and Total Dissolved Solids

Electrical Conductivity (EC) can be used to estimate the total amount of Total Dissolved Solids (TDS) in water. EC is determined by the amount of dissolved solids in water. EC is related to salt content; i.e. the higher the salt content, the higher the EC will be (Anhwange et al., 2012). An increase in conductivity of the water indicates the addition of mineral salts to the water (Gupta et al., 2009). TDS value in mg/L is about half of the electrical conductivity (μS/cm) (Stone et al., 2013). TDS is formed due the ability of water to dissolve salts and minerals and these minerals produce un-wanted taste in water (Mohsin et al., 2013). The presence of high levels of TDS is not desirable as it causes scaling in water pipes, heaters, boilers and household appliances (Gupta et al., 2009).

2.2.4 Total Hardness, Calcium and Magnesium

Calcium (Ca) and Magnesium (Mg) salts are largely responsible for the Total Hardness (TH) of water (Aher and Deshpande, 2011). TH is the concentration of multivalent metallic cations in a solution. Bicarbonates and carbonates of Ca and Mg impart temporary hardness, while, sulphates, chlorides and other anions produce permanent hardness (Uchchariya and Saksena, 2012). The sources of Ca and Mg in natural water are various types of rocks, industrial waste and sewage.

(Gupta et al., 2009). Compounds of Ca become stable when CO2 is present in water, but Ca concentrations are lowered when CaCO3 precipitates due to rise in water temperature. Mg is formed by weathering of rocks having Mg minerals and from some CO3 rocks (Gupta et al., 2009).

2.2.5 Sodium and Potassium

Sodium (Na) is one of the most abundant elements on earth and is highly soluble in water. Increased of Na in surface waters may arise from sewage and industrial effluents. The Na ions are principally brought into water bodies from sodium salts percolated from rocks and occasionally as a result of industrial and domestic activities (Muhammad and Nadeem, 2015). WHO guidelines does not specify limits for Na in surface water. But for drinking water the average taste threshold for Na is about 200 mg/l (WHO, 2011). Potassium (K) levels in water bodies are generally very low as compared to Na since potassium salts are scarce in rocky deposits (Muhammad and Nadeem, 2015). High amounts of K in drinking water causes laxative effects. Potassium salts are

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 8

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

extensively used in industry and for agriculture. K deposits enter freshwaters through industrial discharges and run-off from cultivated fields (Mustapha and Usman, 2014).

2.2.6 Chloride Chlorides occur in most fresh waters, as the salts of sodium or calcium. High chloride content in water sample may be due to the pollution from chloride rich effluent of sewage and municipal waste. Chloride in excess imparts salty taste to water and beverages (WHO, 2011). The chloride concentration can be used as an important parameter for detection of contamination by sewage, prior to other test like BOD and COD (Verma et al.,2013). A maximum of 250 mg/l of Cl is allowable in surface waters as per EPA 2001.

2.2.7 Nitrates and Total Nitrogen

Total Nitrogen (TN) represents the summation of ammonia nitrogen, nitrite, nitrate nitrogen (NO3), and organic nitrogen. The transformation of nitrogen from one form to another does not change the TN concentration; but, loss of nitrogen may occur through sedimentation of particulate matter, uptake of inorganic nitrogen by algae, loss of un-ionized ammonia to the atmosphere, and the reduction of nitrite plus nitrate nitrogen to gaseous nitrogen (Lutz and Cummings, 2003). The main sources of NO3 in water are human and animal waste, use of fertilizers and industrial effluent.

Excess NO3 in the water is a source of fertilizer for aquatic plants and algae. The amount of NO3 in the water is what limits how much plants and algae can grow in most cases. If there is an excess level of NO3, plants and algae will grow excessively. Excess plants in a body of water can create many problems. An excess in the growth of plants and algae create an unstable amount of dissolved oxygen (WHO, 2006).

2.2.8 Total Phosphate and Orthophosphate Phosphorus usually occurs as phosphate, either organically bound as polyphosphates or as soluble orthophosphate, the latter being most readily available as a plant nutrient. Sources of phosphorus include nonpoint sources, such as overland run-off of agricultural fertilizers, which are often associated with sediment, as well as point sources such as sewage treatment plant effluent and various food processing plant discharges (Hoff, 2013). Because phosphorus is an essential nutrient for plant growth, aquatic plants may be stimulated to increase to nuisance levels when sufficient

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 9

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

phosphorus is present. Algal production is related to levels of phosphorus and nitrogen in water (Lutz and Cummings, 2003). 2.2.9 Sulphates 2- Sulphates are naturally present in surface waters as sulphate ions (SO4 ). Sulphates are formed from the leaching of sulphur compounds, sulphate or sulphide minerals such as gypsum and pyrite (Kipngetich et al., 2011). Sulphur is readily soluble in water in its stable and oxidised form. Significant amounts of sulphate are added to surface waters through industrial discharges and atmospheric precipitation (Georgieva et al, 2010). In natural waters, SO4 levels should range generally between 2 mg/l to 80 mg/l, though near industrial discharges they may exceed 1,000 mg/l (WHO, 2006).

2.2.10 Alkalinity

Alkalinity is the acid-neutralizing capacity of water and is usually expressed in mg/l CaCO3. For water with less buffering capacity total alkalinity and acidity are inter-related with pH. When natural waters contain weak acids, alkalinity is usually determined as well as pH in water quality assessment. Total Alkalinity (TA), the concentration of bases in water is composed mainly of bicarbonate, carbonate and hydroxyl ions (WHO, 2006). TA is affected by variations in flow regimes and its natural unevenness is linked to the presence or absence of carbonate rock (Dladla, 2009). The change in alkalinity depends on carbonates and bicarbonates, which in turn depend upon release of CO2. Change in carbonates and bicarbonates also depend upon release of CO2 through respiration of living organisms (Verma et al., 2013).

2.2.11 Chemical Oxygen Demand

The chemical oxygen demand (COD) is used to indirectly measure the amount of organic compounds in water. Most applications of COD determine the amount organic pollutants found in surface water, making COD a useful measure of water quality (Harrafi et al, 2012). It is expressed in mg/l, which indicates the mass of oxygen consumed per litre of solution. COD is the measurement of the amount of oxygen in water consumed for chemical oxidation of pollutants.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 10

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

COD determines the quantity of oxygen required to oxidize the organic matter in water samples, under specific conditions of oxidizing agent, temperature and time (Boyles, 1997).

2.2.12 Algae Algae and other microorganisms in the water greatly affect dissolved oxygen. Under algae bloom conditions, the algae have a negative effect on reservoir fisheries because of periodic oxygen depletion associated with algae respiration and decomposition (Shock et al, 2003). During the day algae picks up carbon dioxide and releases oxygen through photosynthesis so the dissolved oxygen in the water rises. At night their metabolism requires them to take up oxygen and release carbon dioxide. These fluctuations can be large. When an excess of algae grow and sink deeper into the water, their rate of photosynthesis can no longer be maintained, and they decompose (Shock et al., 2003).

2.3 Water quality challenges

Several studies have shown evidence that the effects of human activities like agricultural runoffs, sewage and industrial effluents contaminate freshwater resources (Dutta et al., 2005; Jaji et al., 2007; Dladla, 2009; Chatterjee et al., 2010; Ogleni and Topal, 2011; Al-Anzi, 2012). The effects of human activities on water quality differ in magnitude from one place to another. Changes in the physical, chemical, and biological characteristics of water negatively affect both human and ecosystem health (Wagner et al., 2013).

2.3.1 Agricultural use of fertilizer and runoff

Agricultural activities around the world contributes significantly to water-pollutant loads (Adamu et al., 2014). Agricultural activities are among the most frequently cited sources for degradation and pollution of fresh water systems (Mustapha et al., 2013). Hoff (2013) carried a study to establish the source and magnitude of pollution in Kranji Reservoir in Singapore. The report findings indicated that there were high levels for nutrients and bacterial concentration in the downstream from an intensive cropping vegetable production operation and it is a major contributor to non-point source of pollution in the reservoir. Nyakungu et al., (2013) investigated the impacts of human activities along Manyame River and its tributaries (Mukuvisi, Marimba,

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 11

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Ruwa and Nyatsime rivers) in Zimbabwe. The report established that the contamination is associated with agricultural activities among other sources.

2.3.2 Population growth, uncontrolled disposal of human and industrial wastes The world is getting urbanized at an alarming rate concurrently with the ever increasing population growth. The majority of the population is living in the urban areas, leading to more construction developments to meet the demands by population growth. High populations create more domestic and industrial waste which lead to contamination of freshwaters (Gomes and Ebrary, 2009). In , Damodhar and Reddy, (2013) investigated the impacts of pharmaceutical industry treated effluents on the water quality of river Uppanar and observed that the River Uppanar is highly polluted and unfit for domestic purpose. Findings by Emeka, (2015) indicated industrial effluent from Emene industrial area, in Nigeria polluted the downstream water resources. In Zimbabwe, Nhapi et al., (2007) measured nitrogen and phosphorus concentrations in Lake Chivero and results showed that sewage effluent is the major pollutant source in the Lake. Notwane River in Botswana is heavily polluted with sewage effluent from Glen Valley and Mochudi wastewater treatment plants rendering these rivers unfit for human use (Mladenov et al, 2005).

2.3.3 Mining activities Surface waters in the proximity to mining industries are at a great risk of contamination due to waste discharges from mining activities (UNEP, 2010). Mining industries require different amounts of water based on their operations. The end products of mines come with lot of wastewater that end up being discharged into open water courses (Johnston et al., 2008; Caruso et al., 2012) investigated the impacts of mining on water quality in the Caucasus Mountains in Georgia. Concentrations of manganese, iron and nickel were detected from public water supplies. High concentrations of Iron and manganese were detected on rivers downstream the mining industrial discharges. Nganje et al. (2010) studied the influence of mine drainage on water quality along river Nyaba in Nigeria. Results showed that the river water quality was polluted due to the presence of heavy metals such as manganese, nickel and chromium whose values were above WHO maximum permissible limits. In South Africa, Ochieng et al. (2010) found that the water quality of the Blesbokspruit, Klip and Wonderfontein is polluted due to acid generation from mining activities.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 12

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

The water qualities discharged by the mines into the rivers exceed the limits set by South African aquatic water quality standards of 1996.

2.4 Land use impacts on water quality Researches have been carried on the relationship between water quality and land use type and have established that there is a relationship between the two (Ding et al., 2015). It has been observed that land use greatly impacts the water quality of water systems and it is related to human activities. Conversion of land pattern from one use to another is regarded as the major factor in transforming surface runoff and water quality (Huang et al, 2013). Zamani et al. (2012) assessed land-use change and its impacts on surface water quality in the Ziarat Catchment in Iran. Results showed that about 980 ha of forests were converted to other classes of land use such as croplands, residential areas and roads. The results of this research suggest that land-use change is one of the key factors causing water quality changes in the study area. Gumindoga, (2010) investigated the impacts of land use changes on runoff generation in the Upper Gilgel Abay River Basin in Ethiopia. Results show that increases in agricultural land corresponded to increase in annual runoff volume. Kibena et al. (2014) revealed that the land use and the runoff changes in the same basin affect the water quality of lakes Chivero and Manyame and their tributaries. Runoff influences water quality by introduction of sediments and fertilizers into water bodies, leading to algal blooms and suspended solids (Chithra et al., 2015). Dube et al. (2014) studied land cover changes around Lake Mutirikwi in Zimbabwe from 1984 to 2011. Forest and shrubs were reduced from 310.8 km2 in 1984 to 77.3 km2 in 2011, cultivation increased by 51.44% between 1984 and 2011. The research attributed the Lake enrichment to runoff from surrounding farms.

2.5 Climate change

Climate change is a major threat to water and food security. Water is high susceptible to constantly changing climatic conditions (Misra, 2014). In the next hundred years climate change will affect rainfall patterns, river flows and sea levels and consequently agricultural production (IPCC, 2008). Effects of climate change results in hostile conditions for ecosystem health in water bodies which in turn affect human health (Yadav et al., 2013). Temperature affects surface water quality as it controls oxygen levels, which influences rate of chemical and biological reactions for aquatic

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 13

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

organisms (Yadav et al., 2013). Changes in temperature affect the variability of rainfall which in turn alters the salinity levels of surface water (Trenberth, 2005). Jun et al. (2010) assessed potential impacts of climate change on water quality and ecosystem in China. The study highlighted that water pollution problems caused by climate change are as a result of human activities and economic development. The report established that the magnitude of climate change impacts depends on the conditions of human activities. Urama and Ozor. (2010) studied climate change impacts on water resources in Africa. The report links increase in temperature to warming of lakes and rivers which eventually lead to salinity and nutrients enrichment of water systems. This transformation has an overwhelming impact on freshwater ecosystems, which result in alteration of the availability and distribution of fish populations.

2.6 Remote sensing of water quality

Several studies have highlighted the usefulness of remote sensing in water quality monitoring (He et al., 2008). Application of remote-sensing techniques to estimate water quality offers extended sampling coverage and cost savings. The traditional monitoring methods are not consistent and are time consuming (ZHOU et al., 2010). Remote sensing provide solution for future water resources planning and water quality management (Usali and Ismail, 2010). Predicting water quality parameters using remote sensing needs validation through ground truth (Blake et al., 2013). Therefore assessment and monitoring of water quality using the combination of remote sensing and in-situ measurements plays a significant role in providing reasonable and accurate optically constituents of water (Salama et al., 2009).

Remote sensing is based on the radiative transfer principle. Solar radiation on surface water is either absorbed or scattered back to the open space. The light scattered back comprises of absorption and scattering properties of all water constituents (Ritchie et al., 2003a). The sensor records signal from water that contains information on water and its constituents. Apparent Optical Properties (AOP) are measured from space using sensors. AOP are subsequently related to water quality variables through the Inherent Optical Properties (IOPs) of spectral absorption and scattering coefficients (Olet, 2010).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 14

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Several methods has been used to retrieve satellite data. Methods such as empirical, semi-empirical and analytical models have been applied for estimating and producing quantitative water quality maps using different sensors (He et al., 2008; Olet, 2010).

2.6.1 Availability of satellite data

Many suitable sensors are available for estimating water quality parameters (Hellweger et al., 2004). Sensors are compared and selected based on their spectral, spatial and temporal resolution. A high spectral resolution is desired to differentiate substances based on their reflectance spectrum. (Hellweger et al., 2004). Table 2.0 shows information of different selected bands.

Table 2.0 Satellite information of different selected sensors. Sensor Spatial resolution (m) Revisit time Spectral bands Landsat 30 16 days 8 and 11 MODIS 250 Twice daily 36 MERIS 240 35 days 15 SPOT 20 26 days 5

Hung and Tuyen, (2014) used Landsat multispectral images to estimate suspended sediment concentrations in Lake Trian in Vietnam. The results obtained can be used to create a map of suspended sediment distribution to evaluate water quality unlike traditional methods which can solve the problem on a small scale based on field surveys only. Tzortziou et al., (2007) used MODIS images to quantify lake chlorophyll-a concentrations. Dall’Olmo et al., (2005) assessed the potential of MODIS for estimating chlorophyll concentration in turbid productive waters using red and NIR bands. Findings suggest that, MODIS images could be used to quantitatively to monitor chlorophyll in turbid productive waters as long as atmospheric correction scheme specific for the red and NIR spectral region is available.

In this study, a moderate to high spectral resolution is desired because it has reported wide applications on characterizing water quality of water bodies (Cannizzaro and Carder, 2006; Moses, 2009 ; Thi et al., 2014) hence MODIS is the ideal sensor. According to Dlamini et al. (2016) MODIS is cheap and readily-available and has been intensively used to quantify water quality of reservoirs especially in data-scarce areas like Sub-Saharan Africa. MODIS provide a range of spatial resolutions of 250, 500, 1000 m. The MODIS sensor is on board the Terra and Aqua satellites (Hellweger et al., 2004).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 15

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

2. 6.2 Empirical Determination of water quality variables

A lot of empirical algorithms have been developed in the recent decades (IOCCG, 2006). The empirical algorithms are derived from the relationship of remote sensing reflectance and backscattering coefficients. Ritchie et al. (2003) developed an empirical approach for approximation of suspended sediments. Formation of the empirical algorithms is based on the following equation;

푌 = 퐴 + 퐵푋 ……………………………..1

Where Y is the remote sensing measurement and X is the water quality parameter of interest. A and B are empirically derived factors. Statistical relations are established between measured spectral properties and measured water quality parameters, which aid in the selection of best model in the empirical approach. The empirical characteristics of these relationships bounds their applications to the condition for which the data was collected. Empirical models are only used to estimate water quality parameters for water bodies with similar conditions.

2.6.3 Analytical models for water quality monitoring

A number of semi analytical models that link water leaving radiance to the IOPs of water have been developed. All of these models exploit hydro-optical properties to relate apparent optical properties to the inherent optical properties of water. These IOPs are related to the types and concentrations of dissolved and suspended matter constituents which are mainly; chlorophyll-a, detritus and dissolved organic matter and suspended particulate matter. The concentrations of the water quality parameters can then be derived from the IOPs through the specific inherent properties following the Lambert – Beer Law (Olet, 2010).

Quasi Analytical Algorithm

The Quasi-Analytical Algorithm (QAA), originally developed by Lee et al., (2002), is a semi- analytical model that derives IOPs, in particular the spectral absorption, a(λ), and backscattering, bb(λ), coefficients of water, from spectral remote-sensing reflectance Rrs (λ). QAA is based on the basic relationship of remote sensing reflectance with absorption and backscattering. QAA starts with the calculation of the total absorption coefficient (a) at a reference wavelength (λ0), and then

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 16

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

propagate the calculation to other wavelengths. Component absorption coefficients are further algebraically decomposed from the total absorption spectrum. Several studies have used QAA for analysis or retrieval of water quality in inland and oceanic waters. The model has been refined several times in the past by Lee et al in 2007, 2009, 2010, 2013 (Zheng et al., 2014). QAA is consist of the following elements:

1) The ratio of backscattering coefficient (b) to the sum of backscattering and absorption coefficients (b / (a + b)) at λ is calculated algebraically based on the models of (Lee et al., 2002)

bb(λ) (−0.0895 + √0.008+0.499 rrs(λ)) = … … … … . . (2) a(λ)+ bb(λ) 0.249

Here rrs (λ) is the nadir viewing spectral remote sensing reflectance just below the surface and is calculated from nadir – viewing Rrs (λ) through

(푅푟푠 (휆)) rrs(휆) = … … … … … … … … … … … … … . … (3) (0.52+1.7 푅푟푠 (휆)) 2) The spectral bb(λ) is modelled with the widely used expression ( Gordon and Morel, 1983; Smith and Baker, 1981),

λ0 η Bb (λ) = bbw (λ) + bbp (λ)[ ] … … … … … … … . . . (4) λ

Where bbw and bbp are the backscattering coefficients of pure seawater and suspended particles, respectively (Lee et al., 2002a). Step by step calculations in the QAA are shown on Table 2.1.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 17

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Table 2.1 Steps of the QAA to Derive Absorption and Backscattering Coefficients

Step Property Symbol Equation 1 Below surface Remote r 푅푟푠 rs = sensing reflectance 0.52 + 1.7푅푟푠 2 Ratio of backscattering µ (λ) (−푔0 + [(푔0)2 + 4푔1 푟푟푠]0.5 ) = coefficient to the sum of 2푔1 absorption and backscattering coefficients 3 a(555) = 0.0596 + 0.2[푎(440)푖 − 0.01] Absorption coefficient of the 푎(440)푖 = exp(−2.0 − 1.4푝 + 0.2푝2) total 푟푟푠(440) 푝 = 퐼푛[ ] 푟푟푠(550)

4 Backscattering coefficient of bbp(555) (푢(555)푎(555)) = − 푏푏푤 (555) suspended particles 1 − 푢(555) 5 Spectral power for particle Y 푟푟푠(440) = 22{1 − 1.2 exp [−0.9 ( ]} backscattering coefficient 푟푟푠(550) 푌 6 Backscattering coefficient of bbp(λ) 555 = 푏푏푝(555) ( ) suspended particles 휆 7 Absorption coefficient of the a(λ) [1 − 푢(휆)][푏푏푤(휆) + 푏푏푝(휆)] = total 푢(휆) NB: Table reproduced from Lee et al., (2002)

Rosado-Torres, (2008) evaluated and developed the QAA for chl_a retrieval for Mayagüez Bay data in western Puerto Rico. The QAA performed poorly when the derived IOPs were compared to measured data. The poor performance of the QAA was attributed to its inadequacy in the modeling of the spectral particle backscattering coefficients for Mayagüez Bay.

In conclusion it is shown by various researchers that remote sensing can be effectively used for water quality monitoring without necessarily eliminating in- situ and laboratory analysis entirely but can be used in combination (Hellweger et al., 2004; He et al., 2008). Future water quality monitoring programmes should incorporate both remote sensing techniques and field measurements. Zhu et al. (2014) evaluated the strengths and limits of several algorithms (empirical, semi-analytical, optimization, and matrix inversion methods). The algorithms were evaluated specifically for aquatic colored dissolved organic matter (CDOM) for varying water quality for Lake Huron, United States. Algorithms were evaluated through comparisons to in-situ

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 18

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

CDOM measurements. Results indicated that the majority of the algorithms underestimated high CDOM waters and overestimated low CDOM scenarios. Shareef et al. (2014) used the in situ measurements and IKONOS data assess water quality for Tigris River in France. Results showed that the use of both ground measured water quality parameters and satellite retrieved parameters were able to monitor and predict the distribution of water quality parameters in large freshwater systems.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 19

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

CHAPTER 3 3.0 STUDY AREA

3.1 Location

The Dikgathong Dam is located on the three kilometres below the confluence with the Tati River as shown in Figure 3.0. Shashe River flows through the village of Tonota village, which is located 27 kms south of the City of Francistown along the A1 road. Tati River in the northeast of Botswana flows through the city of Francistown and all the way to join with Shashe River a few kms upstream of Dikgathong Dam. The nearest settlement to the Dikgathong Dam is Robelela at 3km.

Figure 3.0: The Dikgathong Dam and its tributaries.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 20

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

3.1.1 Population The closest settlement to the Dikgathong Dam is Robelela village, with a population of 471 (CSO, 2011). Major settlements found within Dikgathong Catchment are Tonota and the city of Francistown, with population of 20,007 (growth rate of 2.51%) and 100,079 (growth rate of 1.89%) (CSO, 2011) respectively. The Dikgathong Dam supplies the greater Gaborone with drinking water. Gaborone had population of 231,626 in 2011 population census with a growth rate of 2.03 % (CSO, 2011).

3.1.2 Rainfall Rainfall season begins in October, but the rainfall is concentrated in the summer months from November to April. December, January and February are the months in which the heaviest rainfalls are likely to occur. The Dikgathong Dam itself is located where there is an average annual rainfall of 450 mm (Appendix 9). The rainfall is highly variable over time (EHES, 2002).

3.1.3 Soils, geology and vegetation

Predominant soils in the Dikgathong catchment are sandy loams to sandy clay loam. Dam basement rocks includes Karoo dolerite dykes. These rocks stretch east-west ranging in width from a maximum of about 100 m. The dominant plant assemblage in the dam area are the common mixed Mopane/Acacia trees, although considerable variations occur locally due to topography and geology (EHES, 2002).

3.1.4 Socio-economic issues The major economic activity of the people of Robelela, who are predominantly female, is the provision of labour in the Drought Relief Programme. (EHES, 2002). Residents are also engaged in livestock farming and fishing as a source of income and food. Fishing is practiced on a small scale using traps, baskets and hook-and-line (EHES, 2002).

3.2 Potential point and non-point sources of pollution in the Dikgathong Catchment Area

In the catchment water quality problems are mostly associated with farming activities, industrial and domestic effluent discharges (MRC, 2009), which have potential to contribute towards the deterioration of the water quality of the dam. Point and non-point sources of pollution in the area,

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 21

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

includes leachate from landfills, agricultural runoff, domestic and industrial discharges. The type of pollutants associated with non-point sources includes nutrients, sediments, toxic and chemical contaminants and pathogens. Factories and sewage treatment plants are two common examples of point sources. Factories, including textiles, beverages and chemicals, typically discharge pollutants sewage treatment facilities. The major industries in the city of Francistown are; Kgalagadi Breweries Limited, Botswana Meat Commission abattoir and Northern Textiles industry. All the wastewater water from these industries are discharged into sewer lines that convey them to Mambo Wastewater Treatment Plant, which discharges into Tati River. In Tonota village, there is B and M Garments, a large scale textile industry which discharges its wastewater to Tonota Stabilization Ponds, which discharges into Shashe River. Several mines in this area include Tati Nickel, Phoenix mining and Mupane Gold Mine which principally produce cobalt, copper and nickel. Most ranch farms are found along the Tati River while agricultural fields are mainly found along Shashe River.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 22

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

CHAPTER 4

4.0 MATERIALS AND METHODS

4.1 Land use and Land cover Classification for 2010 and 2015

4.1.1 Image acquisition and processing

Landsat Thematic Mapper (TM) satellite imagery with 30 m resolution was obtained from the United State Geological Survey (USGS) website (http://glovis.usgs.gov) for 04 May 2010 and 22 August 2015. Landsat images were used because of relatively high spatial resolution (30 m) suitable for land classification, and its temporal resolution of 15 days make it perfect for land change detection over time (Loveland and Dwyer, 2012). The 2010 and 2015 image were classified into agricultural fields, bare land, forest and shrub, grassland, irrigation, settlements and water. The image was classified through the maximum likelihood digital image classification using the supervised classification approach. This classification uses the training data by means of estimating means and variances of the classes, which are used to estimate probabilities and also consider the variability of brightness values in each class. This classifier is based on Bayesian probability theory. It is the most powerful classification methods when accurate training data is provided and one of the most widely used algorithm (Perumal and Bhaskaran, 2010).

4.1.2 Validation of the classification output

Ground control points were collected from different LULC classes using Garmin model Global Positioning System (GPS) and were used for validation of the 2015 classified map. ILWIS software was used to evaluate the accuracy of the image classification. The strong point of the confusion matrix is its ability to identify the nature of the classification errors, as well as their magnitudes. The confusion matrix is a suitable to produces adequate results for evaluation of accuracy (Tempfli et al., 2001; Erener and Düzgün, 2009).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 23

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

4.2 Physicochemical analysis of water quality parameters

4.2.1 Selection of study site

Botswana experiences generally low rainfall. The Mean annual rainfall varies from a maximum of over 650 mm in the north eastern area of Chobe District to a minimum of less than 250mm in the extreme south western part of Kgalagadi District. Botswana surface water resources are restricted to ephemeral and perennial rivers and water stored in reservoirs. The perennial rivers (Limpopo, Chobe, Zambezi and Okavango) are shared watercourses, and their management and use are subject to the SADC Protocol on Shared Water courses (DWA, 2012). Botswana’s surface water resources are limited and unevenly distributed over the country. Most sources are in northern Botswana, while most people live in south eastern Botswana in and around Gaborone. The average annual run off is 1.2 mm, ranging from zero in western and central Botswana to over 50 mm per annum in the north. The average annual run off implies a total annual run off of 696Mm3. Most of the run off cannot be captured due to the lack of suitable dam sites, high variability of run off in time as well as high evaporation (DWA, 2013).

The Dikgathong Dam was constructed in 2008 and started operation in 2012, With a full supply storage capacity of 400 Mm³, it is Botswana’s largest reservoir and is the main domestic water supplier in the country (Jeffares and Green, 2014). The Dikgathong Dam supports a variety of valuable common property resources such as water, fish and forests which are very important for the livelihoods of the population within the catchment and beyond. The completion and operation of the dam came at the right time when has reached its dead storage level. The Dikgathong Dam now supplies drinking water to the greater Gaborone and other settlements. Livelihood strategies in the catchment area include farming, fishing and fish trading, livestock rearing, hunting and self-employment. Therefore, evaluating the quality of water of this dam is important to ensure it does not deteriorate and to reduce impact on the health of both human, animals and the environment.

4.2.2 Selection of sampling sites

Ten sampling sites were selected for sampling. The precise sampling locations were selected such that they have the more representative sample, where the sampling site corresponded to a well-

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 24

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

mixed area as per the Botswana standard (BOS ISO 5667-6:2005). Four sites were selected to represent the upstream of the confluence after the tributaries Shashe and Tati Rivers (D4, D5, D6 and D7). Three sites were chosen in each of the rivers, D1, D2 and D3 to represent Shashe River, while D8, D9 and D10 represented Tati River. Figure 4.0 shows pictorial location of the sampling points within the Dikgathong Dam while Table 4.0 shows the coordinates location and description of each sampling point.

Figure 4.0. Map illustrating sampling points for the Dikgathong Dam.

Table 4.0 Coordinates of sampling points in the Dikgathong Dam Point Lat Lon Description D1 -212,197 2,754,014 Shashe River entrance into the dam D2 -2,149,183 2,755,743 2.5 km into the dam D3 -2,116,255 27,483,333 5.67 km before Tati and Shashe confluence D4 -2,140,775 2,768,539 2km south of confluence D5 -21,5 27,914 Confluence of Tati and Shashe rivers D6 -21,552 27,891 Dam intake point D7 -21,54 27,968 2.16 km before Tati and Shashe confluence D8 -21,574 27,945 5.32 km into the dam D9 -21,538 27,873 Tati River 3km into the dam D10 -21,498 27,887 Tati River entrance into the dam

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 25

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

4.2.3 Selected of parameters to be analysed

The water quality parameters were selected based on human activities that characterize the study area in relation with the possible parameters present in agricultural runoff, industrial and sewage effluents. Major water quality parameters have been selected for testing as shown in table 4.1. The methods were performed according to the Standard Methods for Examination of Water and Waste Water, 19th edition 2012 Laboratory Standard Operating Manual.

Table 4.1: Analyzed parameters and method used Parameter Method and equipment used Units Temperature YSI 650 water meter OC pH YSI 650 water meter pH Turbidity YSI 650 water meter NTU Electrical Conductivity YSI 650 water meter mg/l Total Dissolved Solids YSI 650 water meter, mg/l Total Suspended Solids Filtration mg/l Sodium Atomic Absorption Spectrometer- ISO 9964-3 mg/l Calcium Atomic Absorption Spectrometer- ISO 7980:1986 mg/l Magnesium Atomic Absorption Spectrometer- ISO 7980:1986 mg/l Chloride Ion chromatography- ISO 1034-2:1995 mg/l Potassium Atomic Absorption Spectrometer- ISO 9964-3 mg/l Chemical Oxygen Demand Titration mg/l

Total Hardness EDTA Titrimetric Method mg/l CaCO3 Total Alkalinity Titration mg/l Phosphates Ion chromatography- ISO 1034-2:1995 mg/l Nitrates Ion chromatography- ISO 1034-2:1995 mg/l Sulphates Ion chromatography- ISO 1034-2:1995 mg/l Total Nitrogen Colorimetric test mg/l Total Phosphorus Colorimetric test mg/l Algae Filter and Screen Counts/ml

4.2.4 Methods of sampling and frequency

Water samples were collected after every two weeks in the Dikgathong Dam starting from January 15th to April 06th 2016. Subsequently, a total of seven (7) sampling campaigns were conducted

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 26

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

during that period. Composite sampling method was used to collect all samples. Composite samples provide a more representative sampling of heterogeneous matrices in which the concentration of the analytes of interest may vary over short periods of time and/or space. Composite samples were obtained by combining portions of multiple grab samples using specially designed extendable hand sampler. Depth integrated samples made of two equal parts were collected at predetermined depth intervals between surface and bottom of the dam.

Preservation of samples followed the Botswana Standards, BOS ISO 5667-3:2003. Samples were put in a cooler box (Figure 4.1) for preservation during transportation to the laboratory. The methods used for testing and analysis are illustrated in Table 4.1. The values of pH, turbidity, EC, DO and TDS were measured directly on site using YSI 650 (SONDE) multi-parameter water meter (Figure 4.2). Samples were collected in a set of one (1) litter plastic sampling bottles containing 2ml of nitric acid for sample preservation. The acidified sample was used for determination of cations (Na+, K+, Ca2+, and Ma2+). The other set of bottles with no acid were used to collect the - 2- - - water sample to determine anions (Cl , SO4 , NO3 and PO4 ).

Figure 4.1: Some of the equipment used include cooler box and polypropylene bottles.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 27

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Figure 4.2: Water field meter YSI 650 for measurement of physical water quality parameters.

4.2.5 Water quality testing

Water samples were submitted to the Department of Water Affairs, Water Utilities Corporation and Betach laboratories for testing and analysis. All these laboratories have implemented their laboratory quality management system in accordance with BOS ISO 17025:2005, which outlines the general requirements for testing and calibrating laboratories. The Laboratories also participates in a proficiency testing scheme with Botswana Bureau of Standards for all the tested parameters. Botswana does not have standard specifically for surface water. Therefore, the results for this study were compared to the international Environmental Protection Agency (EPA) surface water standard of 2001 and the Swaziland Water Quality Objectives (SWQO) of 1999 for surface water.

4.3 Near real-time retrieval of Chl_a and TSM using Quasi Analytical Algorithms (QAA) Near real-time is whereby satellite data is used to retrieve water quality parameters on the same date that the in-situ sampling was conducted ref. It can be the same day or a day before or after the in-situ sampling was conducted. In this study MODIS data is used because it has a revisit time of twice a day. This study is focused mainly on the long term monitoring of Chl_a and total suspended matter (TSM). Chl_a is a biological parameter and the necessary pigment used by most photosynthetic organisms for the release of chemical energy. It exhibits two main absorption maxima positioned at 433nm (blue) and 686 nm (red) of the spectrum (Hunter et al. 2008). The concentration of Chl_a is used as an indicator for the description of bio production and is linearly related to the biomass, the age of algae communities (Theologou et al., 2015).

4.3.1 Downloading MODIS images

Cloud free MODIS images were downloaded from http://modis.gsfc.nasa.gov. Cloud cover provides an enormous challenge in the application of remote sensing methods as it is impossible for visible and near infrared radiation to penetrate thus hindering the registration of water leaving reflectance at the sensors (Olet, 2010). MODIS images were converted to tiff format using the SEBS manual. The HDF viewer was used extract the relevant reflectance scales and offsets.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 28

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

4.3.2 Image processing and extraction of reflectance values at specific points

The ILWIS 3.7.2 software was used for calibrating images under SEBS tools. Imported images were calibrated using the reflectance coefficients. Azimuth and zenith angles were corrected using a scale factor of 0.01. Each zenith and azimuth angles were multiplied by 0.01. Atmospheric correction is one of the most important steps in water quality remote sensing. SMAC tool in ILWIS was used for atmospheric correction of the visible and near visible bands. After the radiometric calibration and atmospheric correction, pre-processed bands with reflectance were imported in ILWIS to convert them to raster maps. The image raster maps were then geometrically corrected using Nearest Neighbours resampling technique (Madamombe and Rwasoka, 2005).

4.3.3 Input of MODIS reflectance into the QAA Remote sensing reflectance obtained from atmospherically corrected MODIS images were input into the QAA programmed excel file for retrieving the absorption and backscattering values for all sampling points in the dam. QAA uses 555 nm as the reference wavelength for all calculations. Wavelengths used in QAA and the corresponding MODIS bands are shown in Table 4.2.

Table 4.2 QAA wavelengths corresponding to MODIS band

QAA_v5 wavelength MODIS band

410 8 440 9 490 10 530 11 550 4 670 1

Backscattering and absorption values calculated from QAA were used to derive water quality parameters (Chl_a and TSS). The Water color simulator model (WASI_v4.1) was used to derive backscattering and absorption coefficients in order to calculate Chl_a and TSM from QAA and MODIS data. The WASI model is described in detail.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 29

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

4.3.4. The Water Colour Simulator (WASI)

WASI is a windows-based software program which was developed by Gege, (2004) for modeling and evaluating optical in situ measurements in freshwater systems. It supports numerous forms of bands such as remote sensing reflectance above and below the surface, irradiance reflectance, specular reflectance at the water surface, absorption, attenuation, and bottom reflectance (Gege, 2004). Equation 5 and 6 are used to calculate Chl_a and TSM respectively. Coefficients of absorption by phytoplankton (a*i (λ)) and bb, Mie* are found in the WASI model.

Absorption

Absorption of a mixture of water constituents is the sum of the components' absorption coefficients:

awc =∑ ci.a*i (λ) + Y.a*y (λ) + D.a*(λ)………………5

λ denotes wavelength. Three groups of absorbing water constituents are considered: phytoplankton

(a*i (λ)), Gelbstoff (Y.a*y (λ)), and detritus (D.a*(λ)).

Phytoplankton. The high number of species that occur in natural waters causes some variability in phytoplankton absorption properties are accounted for by specific absorption spectra ai*(λ). Ci indicates pigment concentration for chlorophyll. Gelbstoff absorption is calculated in WASI by default. Detritus absorption spectrum is provided with WASI and calculated as exp [-SD (λ-λ0)] using λ0 = 440 nm and SD = 0.008 nm-1 (Gege, 2015).

Backscattering

Backscattering bb of a water body is the sum of backscattering by pure water and suspended matter. For the latter, a mixture of two spectrally different types is implemented (Gege, 2015).

bb (λ) = bb, W (λ) + X · bb, X* · bX (λ) + CMie · bb, Mie* · (λ/λS) n ……….6

The first type is defined by a scattering coefficient with arbitrary wavelength dependency, bX (λ), the second type by a scattering coefficient following the Angstrom law (λ/λS) n.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 30

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

For pure water, bbw (λ) = b1 · (λ/λ1) −4.32. The specific backscattering coefficient, b1, depends on salinity. It is b1 = 0.00111 m–1 for fresh water and b1 = 0.00144 m–1 for oceanic water with a salinity of 35–38 ‰, when λ1 = 500 nm is chosen as reference wavelength.

Suspended particles of Type II are defined by the normalized scattering coefficient (λ/λS) n, where the Angstrom exponent n is related to the particle size distribution. CMie is the concentration and bbMie* the specific backscattering coefficient. The parameters are set by default to bbMie* = 0.0042 m2/g, λS = 500 nm, n = –1.

The retrieved chl_a and TSM were then interpolated on QGIS using grid interpolation and the spatial maps produced from ArcMap.

A number of model parameters can change regionally or seasonally, in particular the IOPs of water constituents and the AOPs of the bottom and the atmosphere (Gege, 2014). The WASI model has been designed using data derived from in-situ measurements from lakes in Southern Germany for shallow lakes. If no site-specific information is available, it can be used as a first approximation for other ecosystems as well (Gege, 2004). However, region or season specific information should be used whenever available. Preferably, the optical properties should be measured at the test site close to the airplane or satellite overpass. However this is not always possible (Gege, 2014). In this study the default settings of the WASI model were not changed as there is no regional information for Dikgathong Dam. Other researchers such as (Albert, 2004) have used the WASI model before using the default parameters and were satisfied with the results. Madamombe and Rwasoka, (2005) have also used the default settings for the WASI model for estimating Chl_a and TSM for Lake Chivero in Zimbabwe.

4.4 Relationship between satellite and ground measured water quality parameters

A statistical analysis was performed to establish the relationship between water quality parameters obtained through satellite and those measured on the ground. Spearman’s correlation test was performed to define the rate of association between two variables, to find out whether it is a positive or negative linear relationship (Kibena et al., 2014). The SPSS software 23.0 was used to

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 31

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

calculate the spearman’s correlation rho (r). A critical value was determined for a 95 % confidence interval from the sample size, n, and the r (Faul et al., 2009).

4.5 Methods of data analysis and interpretation

4.5.1 Analysis of the water quality parameters

Water quality results were analysed using a one-way Analysis of Variance (ANOVA), Cluster Analysis (CA) and Principal Component Analysis (PCA). Statistical Package for Social Scientists (SPSS) 23.0 and XLSTAT 2016 tools were used for data analysis and interpretation of results.

A one-way Analysis of Variance (ANOVA)

ANOVA was applied to compare means and test for significance of each water quality parameter. The method enables the difference between two or more sample means to be analysed. The purpose is to test for significant differences between class means, and this is done by analysing the variances. The basis of ANOVA is the partitioning of sums of squares between-class and within- class (Park, 2005).

Cluster Analysis

Cluster Analysis (CA) was used to find natural groupings of samples such that samples within a group are more similar to each other (Akbulut et al., 2010). The resulting clusters of objects should then exhibit high internal homogeneity and high external heterogeneity (Duan et al., 2016). Each cluster thus describes, in terms of the data collected, the class to which its members belong (Salah et al., 2012). According to Key (2009), CA allows for the strategic planning of a future spatial sampling approach in an optimal manner, at reduced cost, without losing any significance of the outcome. K-means was used to assign number of clusters in order to come with the best cluster of sampling points (Salem and Ramadan, 2009). With the k-means you have to know in advance the number of clusters you want. The user input cluster numbers during the analysis. In the k-means, k is the number of clusters you want (Leskovec and Rajaraman, 2014).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 32

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Principal Component Analysis

Principal Component Analysis (PCA) was performed using XLSTAT 2016 excel tool. The PCA tries to convert a large set of inter-correlated indicators into a smaller set of composite indicators, uncorrelated variables called principal components (Fataei, 2011). Principal Components (PCs) provides information on the most meaningful parameters, which describe the whole data set affording data reduction with minimum loss of original information. Eigenvalues of the PCs measures associated variances and the sum of the eigenvalues coincides with the total number of variables. Correlation of PCs and original variables is given by loadings (Kebede and Kebedee, 2010). Findings of a research by Fataei, (2011) on the assessment of water quality of Gharasou river, recommends that the PCA method can be used with high confidence.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 33

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

CHAPTER 5

5.0 RESULTS AND DISCUSSION 5.1.0 Landuse classification Land use patterns surrounding the Dikgathong Catchment Area were assessed before construction (2010) and after construction (2015). Results showed that forest and shrubs were the dominant landuse covering 73.7 % of the total area, followed by settlements (21.1 %), agricultural fields (2.76 %), water (1.88 %), grassland (0.45 %), bareland (0.16 %) and irrigation (0.008 %). These results showed an overall increase of 7 % in settlements from 14.1 % to 21.1 % from 2010 to 2015 respectively. This trend corresponds with the decrease in agricultural fields, forests and bare land. Forest and Shrubs were the dominant LULC type covering 74.5 % in 2010 and 73.7 % in 2015. A slight reduction of 0.8 % in forest and shrubs during that period can be attributed to increase in settlements and water bodies. Agricultural fields has undergone the highest reduction of 8.5 % from 11.2 % in 2010 to 2.76 % in 2015. The changes on agricultural fields can be attributed largely to increase in settlements. Figure 5.0 shows classified thematic map for 2010 and 2015. Agricultural fields, bare land, forest and shrubs, grassland, irrigation, settlements and water body are the major LULC classes which were easily identified and defined on Landsat 5TM imagery. Table 5.0 and Figure 5.1 shows percentage and total area for each landuse for 2010 and 2015 respectively.

Figure 5.0 Classified thematic map for 2010 and 2015

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 34

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Table 5.0 Land use change in area from 2010 to 2015 2010 2015 Landuse Area in km2 % Area in km2 % % change Agricultural Fields 323 11.2 80 2.76 -8.5 Bareland 5.7 0.20 4.7 0.16 -0.04 Forest & shrub 2237 74.5 2124 73.7 -0.8 Grassland 12.7 0.44 13 0.45 0.007 Irrigation 2 0.08 0.23 0.008 -0.07 Settlements 407 14.1 608 21.1 7.0 Water 15 0.53 54 1.88 1.35 Total area 2883 100 2883 100

2500 2237 2124 2000

1500

1000

Area km2 in 608 407 500 323 80 5.7 4.7 12.7 13 2 0.2 15 54 0 Agricultural Bareland Forest & Grassland Irrigation Settlements Water fields shrubs Landuse

2010 2015

Figure 5.1: 2010 and 2015 land use classification

Several studies have been carried on the relationship between water quality and land use type and have established that there is a relationship between the two. Ding et al. (2015) assessed the impacts of landuse on surface water quality in Dongjiang River Basin, in China. Urban settlements were positively correlated to temperature (r = 0.59), EC (r = 0.66), TN (r = 0.40), TP (r = 0.27), chl_a (r = 0.46) and negatively correlated to NO3 (r = - 0.17). Agricultural land was positively correlated to temperature (r = 0.25), EC (r = 0.28), TN (r = 0.12), TP (r = 0.00), chl_a (r = 0.37) and negatively correlated to NO3 (r = - 0.10). Forest land was negatively correlated temperature (r= - 0.54), EC (r= - 0.65), TN (r = - 0.35), TP (r = - 0.20), chl_a (r = - 0.50) and positively

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 35

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

correlated to NO3 (r = 0.23). Haidary et al. (2013) assessed the impacts of land use types on the water quality of wetlands in Japan. The findings indicated that there was a significant positive relationships between urban areas and EC (r =0.67), TDS (r =0.69), TN (r =0.92), NO2 (r =0.52) while negative relationships were observed between forest areas and EC (r = - 0.67), TDS (r = - 0.68) and TN (r = - 0.68).

Kibena et al. (2014) assessed the relationship between water quality parameters and changes in landuse patterns in the upper Manyame River in Zimbabwe. Results indicated that settlements had a strong positive correlation with COD (r =0.97), TSS (r =0.87), TP (r =0.90) and TN (r =0.76). The relationship was attributed to both point and non-point sources of pollution in the catchment from sewage effluent from wastewater treatment plants. Forested had a significant negative correlation with COD (r = - 0.44) and TSS (r = - 0.95). Agricultural activities had strong positive relationship with TSS (r =0.78), TP (r =0.96) and TN (r =0.97). The relationship was attributed to extensive use of fertilizers from upstream urban agricultural activities.

Xiao, et al. (2016) analyzed the relationship between landscape pattern and water quality in Taihu Lake in China. Results indicated positive correlation between settlements and COD, BOD and

NH3. The results showed that the increase in built-up land increases the concentration of COD and BOD. The forest had negative correlation with COD and BOD, indicating that the increase of forest will decrease the concentration of these water quality parameters. Huang et al. (2013) evaluated the impacts of landuse on water quality in Chaohu Lake Basin. Results showed that the built-up area was positively related to TP, TN, and COD, indicating that the increase of the built up area tends to degrade the water quality. Results also showed that forest land and grassland were negatively related to TP, TN, and COD. The significant negative relationship between the forest land and grassland area and TP, TN and COD indicates that the forest land and grassland played a key role in reducing the nitrogen pollutants and phosphorus. Forest land and grassland can effectively reduce the nutrient salts brought into the river by the surface runoff since they play an important role in reducing the surface runoff, conserving the water and soil, and absorbing the pollutants. Therefore, the increase of the forest land and grassland area will reduce the concentration of both TP and TN and consequently improve the water quality.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 36

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Based on previous studies results on relationship between landuse and water quality and the current conditions of landuse in the Dikgathong Catchment Area; chl_a, COD, EC, TP, TN, TSS, NO3 and

NO2 were selected for testing and analysis for this study. These parameters were analyzed together with other major water quality parameters such as Algae, Ca, Cl, Mg, Na, K, SO4, PO4, turbidity, hardness and alkalinity.

5.1.1 Validation of the classification output

The verification of the accuracy of the derived land use maps was performed for 2015 using the confusion matrix. The overall classification accuracy for all classes is 64.9 % as presented in Table 5.1. The accuracy of 65 % is very close to new introduced weighted accuracy of 66.9 %. The International Geosphere–Biosphere Program (IGBP) introduced an area weighted accuracy of 66.9%. Accuracy for LULC studies is considered fair for values above 50 % and good for values above 70% (van Vliet et al., 2011). Agriculture has accuracy of 64.6 %, bare land 50 %, Forest has 69%, grassland has 57.9%, Irrigation has 56 %, and settlements at 72% and water has 100%. Thomlinson. (1999) set a target of an overall accuracy of 85% with no class less than 70% accurate. The IGBP set the accuracy for the individual classes from 40% to 100%. Many other studies discuss classifications with overall accuracies below the general target of 85% and have a large range in the accuracy with which the individual classes have been classified (Foody, 2002).

Table 5.1 Results for LULC accuracy assessment

Classified Total % Accuracy Agricultural Fields 31 48 64.6 Bareland 19 38 50 Forest & Shrub 38 55 69.1 Grassland 11 19 57.9 Irrigation 14 25 56 Settlements 36 50 72 Water 10 10 100 Total 159 245 64.9

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 37

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

5.2 Water quality of the Dikgathong Dam A total of twenty water quality parameters temperature, pH, EC, turbidity, TDS, TSS, Ca, Mg, Na,

K, Cl, NO3, PO4, SO4, COD, TA, TH, TP, TN and algae were measured. ANOVA was applied with Post Hoc Multiple Comparisons using Student-Newman-Keuls (S-N-K) tests to find out if there were any significant variations of the parameters among the ten sampling points. Descriptive statistics for average water quality parameters from ten sampling sites are summarised in Table 5.2. The Principal Component Analysis (PCA) was performed to find out from the twenty selected parameters which ones mostly influenced the variability of the water quality of the dam. The PCA grouped the parameters based on their influence to similar pollution characteristics. From the twenty parameters PCA identified ten parameters (TA, Mg, Ca, TH, SO4, NO3, TSS, EC, COD and turbidity) which are discussed in this section. Table 5.2: Descriptive statistics for average water quality parameters from ten sampling points Units Range Minimum Maximum Mean Std. Deviation O Temp C 3 25 28 26.53 1.039 pH pH 1 7 8 7.32 0.19 EC µS/cm 43 175 218 202.18 12.342 Turbidity NTU 185 2 187 51.35 67.821 TDS mg/l 48 107 155 131.2 13.665 TSS mg/l 214 13 227 72.24 74.555 Ca mg/l 6 19 25 22.2 1.956 Mg mg/l 1 5 7 5.82 0.447 Na mg/l 6 7 13 9.62 1.769 K mg/l 3 5 9 7.07 1.109 Cl mg/l 5 5 10 6.13 1.519 mg/l NO3 2 0 2 0.52 0.522 mg/l SO4 1 1 2 1.45 0.504 TA mg/l CaCO3 19 82 101 91.34 6.186 TH mg/l CaCO3 24 90 114 103.39 8.997 COD mg/l 80 2 82 21.63 22.782 Algae Counts/ml 130 21 151 79.92 38.714 TP mg/l 0 0 0 0.06 0.022 TN mg/l 1 0 1 0.74 0.205

5.1.1 Chemical Oxygen Demand Measured values ranged from 2.00 mg/l to 82.0 mg/l with an average of 21.6 mg/l. One way ANOVA showed no significant variation (p=0.411) among the COD values. Figure 5.2 shows

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 38

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

spatial variation of COD in the dam with point D9 having highest concentration. High levels of COD were due to the sewage effluent discharge from City of Francistown Wastewater Treatment Plant and Tonota Village Ponds. Similar trends to the current study were observed by (Anhwange et al., 2012) who studied impacts of human activities on Benue water quality in Nigeria. The authors recorded COD values of between 91.60 mg/l to 128.83 mg/l. The authors cited inflow of livestock waste, domestic and industrial waste as the sources of COD concentrations. COD is an indicator of organic pollution, which is caused by the inflow of domestic, livestock and industrial waste that contains elevated levels of organic pollutants Garg et al. (2010). The high concentration levels in the Dikgathong Dam makes the dam polluted.

Figure 5.2: Spatial variation of COD in the Dikgathong Dam

5.1.2 Electrical Conductivity EC values range from 175 µS/cm to 218 µS/cm with an average of 202 µS/cm and within a limit of 1500µS/cm set by EPA, 2001 as illustrated in Figure 5.3. One way ANOVA showed no significant variation (p=0.284) among the values of EC. Variations show a slight decrease in measured EC values from January to April. Similar observations to the current study were made by Klake et al. (2015), who studied the seasonal variation in water quality of the Weija Dam in

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 39

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Ghana. The authors observed slightly high conductivity values during the dry season and low values in the wet season. High EC values during dry season are due to evaporation resulting in high concentration of ions (Verma et al, 2013). The lower conductivity in the wet season might be due to high rainfall which reduces the level of dissolved solids by dilution of water in the dam through runoff which increases the volume of water (Anhwange et al., 2012). The concentration of EC in the Dikgathong Dam were within allowable limits.

250 200 150 100

EC EC (mg/l) 50 0

Sampling dates

Figure 5.3 Temporal variation of average values of EC in the Dikgathong Dam 5.1.3 Turbidity

Turbidity values range from 2.0 NTU to 187 NTU with an average of 51.4 NTU and way above 5.0 NTU by Swaziland Water Quality Objectives of 1999 for surface waters. One way ANOVA showed no significant difference (p=0.063) among the turbidity values. The highest value for turbidity was measured on the 10th of March when the highest rainfall of 22 mm was recorded (Appendix 10). Figure 5.4 shows spatial variation of turbidity concentration levels in the dam. Water quality assessment of the Owena Dam in Nigeria by Irenosen et al. (2012), also reported similar observations of high turbidity (5.60 NTU to 8.80 NTU)during rainy season which were above the stipulated 5.0 NTU. During rainy season silt, clay and other suspended particles contribute towards high turbidity values, while during dry seasons settlement of silt, clay results low turbidity (Thirupathaiah et al., 2012). Turbidity levels were beyond set limits of 5.0 NTU and therefore the dam is contaminated due to high turbidity levels.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 40

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Figure 5.4: Spatial distribution of Turbidity in the Dam.

5.1.4 Total Suspended Solids TSS values ranges from 13.0 mg/l to 227 mg/l with an average of 72.2 mg/l and above 50 mg/l set by EPA 2001. ANOVA showed no significant variation (p=0.103) between the TSS values. Spatial variation of TSS in the dam is shown in figure 5.5. The highest value was recorded on March 10th when peak rainfall of 22 mm was recorded (Appendix 10) and corresponds with that of high turbidity. Similar findings were made by Anhwange et al., (2012) who studied impacts of human activities on Benue water quality in Nigeria and observed that TSS values for wet season were high than that of dry season. TSS can often be related to turbidity as was proved by Hui et al., (2011) who established a positive relationship between TSS concentration and turbidity. The author suggested that the measurement of turbidity is possibly the most economic option for estimating total suspended solids concentration in a water body. The dam is polluted due to TSS levels that were above the EPA limit.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 41

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Figure 5.5: Spatial variation of average values of TSS in the Dikgathong Dam

5.1.5 Total Hardness Total Hardness (TH) is due to the presence of predominantly calcium and magnesium. TH values ranged from 90 mg/l to 114 mg/l with an average of 103.4 mg/l and below 350 mg/l CaCO3 by EPA standard. The Dikgathong Dam water is classified as moderately soft hardness. The water in the dam can be classified as moderately soft (50 – 110 mg/l) to moderately hard (151 – 250 mg/l) according to the EPA standard. ANOVA showed no significant variation (p=0.436) between the TH values. High values were observed in February and declined towards April (Figure 5.6), depicting high values in dry season and low values in wet season. Similar results were obtained by Khound et al., (2012) who assessed physico chemical parameters for Jia-Bharali River Basin, in India. The reporter observed low TH values during wet season and high values during dry season and linked the pattern to high dilution during wet season. Total hardness is due to the presence of bicarbonate, sulphate, chlorides and nitrates of calcium and magnesium (Olawale,2016). TH values were within allowable limits in the Dikgathong Dam.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 42

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

140 120 100 80 60 40 20

0 TH (mg/l TH (mg/l CaCO3)

Sampling dates

Figure 5.6: Temporal variation of average values of TH in the Dikgathong Dam

5.1.6 Calcium

Measured Ca values ranges from 19.0 mg/l to 25.0 mg/l with an average of 22.2 mg/l. ANOVA showed no significant variation (p=0.486) between the Ca values obtained in all sampling points (Figure 5.7). Low Ca values was recorded on 10th March when peak rainfall was recorded. Khound et al. (2012) analysed physico-chemical parameters of Jia-Bharali River Basin in India. The researcher who reported high Mg values in dry season and low values in wet season.

30 25 20 15 10 Ca (mg/l) 5 0

Sampling dates

Figure 5.7: Temporal variation of average values of Ca in the Dikgathong Dam 5.1.7 Magnesium Mg values ranges between 5.0 mg/l to 7.0 mg/l, with an average of 5.82 mg/l. ANOVA showed no significant variation (p=0.168) between the Mg values as shown in figure 5.8. Low Mg values was recorded on 10th March when peak rainfall was recorded. To support this findings, Olawale,

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 43

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

(2016) analyzed physico chemical parameters of Asa River water quality in Nigeria and also recorded high Mg values in dry season and low values in wet season.

8 7 6 5 4 3

Mg(mg/l) 2 1 0

sampling dates

Figure 5.8: Temporal variation of average values of Mg in the Dikgathong Dam

5.1.8 Total Alkalinity

Alkalinity levels varied from 82.0 mg/l to 101 mg/l, with an average of 91.3 mg/l. There is no limit specified for alkalinity. ANOVA showed no significant variation (p=0.069) between the TA. TA values shows a slight decrease from the beginning of the sampling campaign (Figure 5.9), therefore the alkalinity is high in the dry season and lower in the rainy season, when the dam water levels increases. These results agree with findings by Sawant and Chavan, (2013), who attributed low values in the rainy season to dilution and high values to the increase of carbonates and bicarbonates in the water in the dry season. Alkalinity increases as the amount of dissolved carbonates and bicarbonates increase (Pulugandi, 2014).

120 100 80 60 40

TA TA (mg/l) 20 0

Sampling dates

Figure 5.9: Temporal variation of average values of TA in the Dikgathong Dam

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 44

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

5.1.9 Nitrates

NO3 concentrations ranged from 0.00 mg/l to 2.00 mg/l, with an average of 0.52 mg/l. ANOVA showed a significant variation (p=0.001) between the NO3 values as shown in Figure 5.10. Similar trends were observed by (Mustapha, 2008) who assessed water quality using physico chemical parameters for Oyun Reservoir in Nigeria. The researcher also recorded high nitrates during rainy season. NO3 is found in little amounts in natural waters and mostly it is of mineral origin, while most coming from organic and inorganic sources, such as effluent discharges and fertilizer runoff (Irenosen, 2012).

1.4 1.2 1 0.8 0.6 0.4

NO3(mg/l) 0.2 0

Sampling dates

Figure 5.10: Temporal variation of averages values of NO3 in the Dikgathong Dam

5.1.10 Sulphates

Measured SO4 values range from 1.00 mg/l to 2.00 mg/l with an average of 1.45 mg/l and very low compared to the EPA standard of 400 mg/l. ANOVA showed a significant variation (p=0.039) between the SO4 values as shown in Figure 5.11. Patil et al. (2013) assessed water quality parameters in Kolhapur, India and found similar variations (0.6 mg/l to 8 mg/l) with high recordings during rainy season while lowest dry summer season.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 45

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

2.5 2 1.5 1

SO4 (mg/l) 0.5 0

Sampling dates

Figure 5.11: Temporal variation of average values of SO4 in the Dikgathong Dam

Cluster analysis

The aim of the cluster analysis was to find natural groupings of samples such that samples within a group are more similar to each other. The resulting clusters of objects should then exhibit high internal homogeneity and high external heterogeneity (Akbulut et al., 2010). The resultant of CA is a cluster membership (Table 5.3, grouping sampling points with similar concentration levels together into clusters. As a result, five groups of sites were formed from ten sites. Table 5.4 shows final cluster centres portraying concentration levels for each parameter in the group. Cluster 1 is made of point D1 only, which is the entrance of the Shashe River tributary into the Dikgathong Dam. Cluster 2 is made of point D9 only. Point D9 has the highest levels of concentrations compared to other points. Cluster 3 is made up of points D3, D6 and D8. The points D3 and D8 are points before the confluence of the tributaries while point D6 is the intake point (D6). These are the points with fairly low concentration levels per each parameter. Cluster 4 is consist of the points D2, D4, D5 and D7. Points D4, D5 and D7are located close to the centre of the dam. Cluster 5 is made of only point D10, which is the entrance of Tati River into the dam.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 46

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Table 5.3: Cluster membership Case Number Points Cluster Distance 1 1 1 0.000 2 2 4 34.993 3 3 3 40.674 4 4 4 24.511 5 5 4 18.768 6 6 3 34.795 7 7 4 37.853 8 8 3 26.171 9 9 2 0.000 10 10 5 0.000

Table 5.4: Final cluster centers Cluster

1 2 3 4 5 Temp 27.4 27.1 26.23 26.14 27.6 pH 7.06 7.31 7.35 7.32 7.43 EC 175.3 205 205.3 205.6 203 Turbidity 186.8 126.7 7.76 13.9 121.1 TDS 107.3 124 139.3 129.4 145 TSS 171 227 37.8 24.1 114.6 Ca 20.5 21 21.7 23.6 20.8 Mg 5.17 5.74 5.74 6.18 5.41 Na 7.47 11.48 9.44 8.93 13.2 K 6.44 8.2 6.94 7.03 7.06 Cl 6.06 7.38 5.79 5.22 9.63 NO3 1.77 0.87 0.321 0.173 0.89 SO4 2.16 1.75 1.34 1.06 2.33 TA 82.01 89 92.63 95.03 84.4 TH 92.1 110.8 100 109.2 94.4 COD 20 81.66 16.17 11.56 19.9 Algae 52.8 42.5 124.2 77.6 21 TP 0.077 0.029 0.057 0.061 0.085 TN 0.952 0.411 0.659 0.761 0.998

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 47

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Salah et al., (2012) performed CA on ten sampling sites in the Euphrates River. The ten sampling sites were group into two statistically significant sites. Cluster I included sampling site (S7). Cluster II comprised the sampling sites S1 - S6, and S8 - S11. Among the sampling sites, site 7 (S7) had the lowest pollution while the other sites (S1 - S6 and S8 - S11) had the highest pollution levels. The result were in agreement with the variation in water quality parameters measured in the sampling sites. Principal Component Analysis (PCA)

Table 5.5 summarizes the PCA results that includes eigenvalues, the amount of variance explained by each F and the cumulative variance. The results are complemented by the examination of the loadings of the three retained components (Table 5.6). The PCA grouped the parameters in each cluster and formed new variables using the correlation matrix based on their influence to similar pollution characteristics. The PCA indicates that the first three principal components together account for 80.8 % of the total variance in the dataset.

The first principal component (F1) explained 41.9 % of total variance, with strong positive loading on TA, Mg, Ca, TH, SO4, NO3, TSS, EC, and turbidity. This factor group is highly and positively contributed by the variables related to natural factors. TA increases as the amount of dissolved carbonates and bicarbonates increase (Pulugandi, 2014). An increase in EC reflects the pollution load as well as tropic levels of aquatic body (Anhwange et al., 2012). TH is due to the presence of bicarbonate, sulphate, chlorides and nitrates of calcium and magnesium. F1 pollution can be associated with quantities of sewage effluents and detergents in the reservoir from settlements through surface runoff water.

The second principal component (F2) explained 25.2 % of total variance. F2 had strong positive loading of COD, Na and K and moderate loading of pH, turbidity and TSS. TSS concentrations will increase turbidity level. The association of these variables may have occurred as a result of run-off around the dam, which may increase the levels of SS and TS. The source of these sediments includes natural and human activities in the watershed, such as natural or excessive soil erosion from urban runoff, industrial effluents, or excess phytoplankton growth (UNEP/GEMS, 2006).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 48

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Pollution for F2 can related to human activities in the watershed, such as erosion from urban runoff, sewage and industrial effluents.

The third principal component (F3) had strong loading on TDS, Na and Cl and explained 13.8% of the total variance. High Cl were attributed to rainfall runoff. Increased concentration of Cl is always regarded as an indicator for pollution due to sewage discharge, irrigation, and solid waste (Uchchariya, 2012). High values may be owing to loss of water due to heat and concentration of salts present in water, while low vales may be due to dilution of water during rainy season (Sawant and Chavan, 2013).

Table 5.5: Eigenvalues for principal components

F1 F2 F3 F4 F5 F6 F7 F8 F9 Eigenvalue 7.953 4.780 2.618 1.568 1.216 0.457 0.170 0.131 0.108 Variability (%) 41.857 25.156 13.779 8.254 6.400 2.407 0.894 0.687 0.567

Cumulative % 41.857 67.013 80.792 89.046 95.445 97.852 98.746 99.433 100.000

Table 5.6: Contribution of the variables (%)

F1 F2 F3 pH 4.762 6.074 3.045 EC 8.109 3.366 6.084 Turbidity 7.981 5.422 1.352 TDS 0.241 0.921 27.551 TSS 6.800 5.901 3.776 Ca 8.201 0.717 0.105 Mg 10.231 0.638 0.140 Na 0.206 8.313 18.950 K 1.854 10.574 1.172 Cl 3.312 6.975 14.601 NO3 9.434 1.942 1.876 SO4 9.618 2.415 3.055 TA 11.598 0.387 0.003 TH 7.924 4.321 0.872 COD 0.155 13.622 3.810 Algae 2.846 5.735 0.558 TP 4.225 5.962 7.685 TN 2.405 2.156 5.338

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 49

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

For this study, the first principal component (F1) will be used determine the monitoring program for the Dikgathong Dam. Therefore the nine parameters (TA, Mg, Ca, TH, SO4, NO3, TSS, EC, and turbidity) with strong positive loading are recommended for regular monitoring by Water Utilities Corporation (WUC). F1 pollution can be associated with quantities of sewage effluents and detergents in the dam from settlements through surface runoff water. Several researchers used PCA to assess water quality parameters. Mustapha and Abdu. (2012) applied PCA on the surface water quality data to identify pollution sources and their contribution toward water quality variation. Fifteen physico-chemical water quality parameters were selected for analysis. PCA was applied to examine the source of each water quality parameters and generated three components factors with 80% total variance. PCA linked pollution to erosion, domestic, dilution effect and agricultural run-off. The application of PCA has allowed an analysis of the possible correlations between the parameters measured. CA helped to realize the relationship between sampling sites and sampling time as far as the homogeneity of water composition is concerned.

5.2 Water quality parameters retrieved from MODIS images

Water leaving reflectance, were extracted from atmospherically corrected MODIS images and then input into the QAA excel sheet to determine the absorption and backscattering values. Chl_a and Total Suspended Solids (TSS) were then calculated using equation 5 and 6, using coefficients obtained from the lookout table from WASI model.

Results for March 10th 2016 data are shown on Table 5.7, while the rest of the results are shown in Appendix 11 to Appendix 14. The results for 10th March have high reflectances compared to other sampling dates. Reflectances varied from 0.0586 to 0.3175 for aph. Chl_a ranged between 1.74 mg/m3 and 9.48 mg.m3. Backscattering reflectance varied from 0.0327 to 0.2898 while TSM ranged between 5.68 mg/l and 53.1 mg/l as shown in Table 5.7. High TSM values can be attributed to the amount of rainfall measured during that week; 3.05 mm, 12.7 mm, 14.0 mm and 22.4 mm recorded on the 7th, 8th, 9th and 10th of March respectively. Low chl_a values can be linked to dilution due to rainfall.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 50

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Table 5.7: Derived IOPs and Chl_a for 10th March 2016

Derived absorption Chl_a Derived bbp backscattering TSM aph coefficient (mg/m3) QAA coefficient (mg/l) QAA

D1 0.3175 0.0335 9.48 0.2898 0.004773 53.1 D2 0.2597 0.0335 7.75 0.1702 0.004773 31.1 D3 0.1397 0.0335 4.17 0.0742 0.004773 13.3 D4 0.0766 0.0335 2.27 0.0689 0.004773 12.3 D5 0.0723 0.0335 2.16 0.0535 0.004773 9.51 D6 0.0586 0.0335 1.75 0.0327 0.004773 5.68 D7 0.1528 0.0335 4.56 0.0816 0.004773 14.7 D8 0.1316 0.0335 3.93 0.0423 0.004773 7.45 D9 0.1933 0.0335 5.77 0.1449 0.004773 26.4 D10 0.2910 0.0335 8.69 0.0992 0.004773 17.9

Overall chl_a concentrations varied from 1.74 to 24.4 mg/m3 within dam. Figure 5.12 shows spatial variation for chl_a in the dam for the entire sampling period. The lowest chl_a of 1.74 mg/m3 is found at point D6, which is the intake point of the dam. The highest reading of 24.4 mg/m3 is found at point D9, few kilometers of the mouth of the dam. There is little variation in chl_a values for different sampling points and dates, but generally lowest values are recorded at the center of the dam after the confluence of Tati and Shashe Rivers. Highest values are found at the mouth of the dam where Tati and Shashe Rivers enter the dam. The EPA classifies a lake as oligotrophic when chl_a is less than 8 mg/m3 and mesotrophic in the range of 8 – 25 mg/m3. For this study there are no in situ or laboratory analysis measurements for Chl_a to compare with remote sensing retrieved data. Dikgathong Dam can be classified as both oligotrophic and mesotrophic, with low to moderate algal growth, while the level of pollution is classified as very low to low (EPA, 2001). Chl_a is a key indicator of phytoplankton productivity and is the major light-absorbing pigment in green plants useful for photosynthesis (Boyer et al., 2009).

Overall TSM ranged from 2.34 mg/l to 59.2 mg/l in the dam. Figure 5.13 shows spatial variation for TSS in the dam for the entire sampling period. Highest readings are found on the edges of the dam, while lowest values are found at the center of the after the confluence of tributaries. Most of the high values were measured on the 10th March when the peak rainfall of 22mm was recorded.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 51

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

The results for TSM retrieved from MODIS are low compared to the results from laboratory analysis which ranged from 13 to 227 mg/l.

Figure 5.12: Spatial variation of Chl_a in the Dikgathong Dam

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 52

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Figure 5.13: Spatial variation of TSS in the Dikgathong Dam

5.3 Relationship between satellite and ground measured water quality parameters

Spearman’s correlation was used to establish if there exists a relationship between remote sensing retrieved data and ground measured water quality parameters. Results are shown in table 5.8 and 5.9. Strong positive significant correlation was observed between chl_a and turbidity (r=0.794 and

0.830), TSS (r = 0.819 and 0.770), SO4 COD (r=0.781 and 0.769), SO4 (r= 0.851 and 0.646) and alkalinity (r= 0.847). Moderate positive and non-significant relationship is observed for temp (r= 0.055), pH (r= 0.587), EC (r= 0.409), TDS (r=0.348), Na (r= 0.406) and Cl (r= 0.394).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 53

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Table 5.8. Spearman’s correlation between chl_a and ground measured water quality

15-Jan 26-Jan 11-Feb 10-Mar 6-Apr r Sig. r Sig. r Sig. r Sig. r Sig. N Chl_a 1 . 1 . 1 . 1 . 1 . 10 TSM 0.394 0.26 0.406 0.244 0.285 0.425 0.927** 0 0.673* 0.033 10 Temp -0.559 0.093 -0.267 0.456 0.597 0.068 0.556 0.095 0.055 0.881 10 pH -0.413 0.236 -0.222 0.538 0.032 0.929 -0.683* 0.03 0.587 0.074 10 EC -0.281 0.431 -0.285 0.425 0.238 0.509 -0.375 0.285 0.409 0.241 10 Turbidity 0.794** 0.006 -0.19 0.599 0.479 0.162 0.830** 0.003 0.127 0.726 10 TDS 0.125 0.73 -0.628 0.052 0.15 0.679 -0.256 0.475 0.348 0.325 10 TSS 0.819** 0.004 -0.078 0.83 0.528 0.117 0.770** 0.009 -0.225 0.532 10 Ca -0.113 0.757 -0.367 0.297 0.4 0.252 -0.681* 0.03 -0.042 0.907 10 Mg -0.35 0.321 -0.354 0.315 0.138 0.705 -0.5 0.141 -0.055 0.881 10 Na -0.388 0.268 -0.541 0.107 0.063 0.864 -0.037 0.92 0.406 0.244 10 K -0.850** 0.002 0.217 0.546 0.049 0.894 -0.573 0.083 -0.527 0.117 10 Cl 0.138 0.705 -0.354 0.315 0.12 0.74 0.012 0.973 0.394 0.26 10 TA -0.125 0.731 0.847** 0.002 0.205 0.57 -0.463 0.177 -0.394 0.26 10

NO3 0.564 0.089 -0.683* 0.029 0.078 0.831 0.549 0.1 -0.024 0.947 10 NO2 0.528 0.117 -0.701* 0.024 0.406 0.244 0.097 0.79 . . 10

SO4 0.851** 0.002 -0.155 0.668 0.114 0.754 0.646* 0.043 -0.079 0.829 10

PO4 . . -0.522 0.122 0.174 0.631 . . 10 COD 0.781** 0.008 0.103 0.777 0.51 0.132 0.769** 0.009 -0.061 0.866 10 TH -0.25 0.486 0.563 0.09 0.302 0.397 -0.596 0.069 -0.115 0.751 10 **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

Similar results to this study were observed by Faragallah et al., (2009), who tested correlation between physicochemical parameters and chl_a in the Damietta in Egypt. Relatively high levels of chl_a concentrations was recorded in the surface layer during the period of study and negative correlation was found between chl_a and both NO3, PO4 and SO4 (r= -0.58, -0.38 and -0.58, respectively). The study observed moderate positive significant correlation between chl_a and COD (r = 0.559).

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 54

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Table 5.9. Spearman’s correlation between TSM and ground measured turbidity and TSS

15-Jan 26-Jan 11-Feb 10-Mar 6-Apr Turbidity TSS Turbidity TSS Turbidity TSS Turbidity TSS Turbidity TSS TSM r -0.006 0.106 -0.37 -0.234 0.564 -0.112 0.867** 0.733* 0.394 -0.097 Sig. 0.987 0.77 0.293 0.515 0.09 0.757 0.001 0.016 0.26 0.789 Turbidity r 1 0.613 1 0.508 1 0.467 1 0.927** 1 -0.401 Sig. . 0.06 . 0.134 . 0.173 . 0 . 0.25 N 10 10 10 10 10 10 10 10 10 10

Relationship was also established between satellite measured TSM and laboratory analysed TSS and turbidity. Strong positive and significant correlation was observed for 10th March data, with turbidity (r= 0.867) and TSS (r= 0.733). Moderate positive correlation for TSS (r= 0.106) for 15th January, turbidity (r= 0.564) and (r= 0.394) for 10th March and 6th April respectively. There was a strong positive significant correlation between TSS and turbidity with (r=0927) on 10th March. Strong positive significant correlation between satellite measures chl_a and TSM (r= 0.927) was also observed on the 10th March 2016. On the 10th March a peak rainfall of 22 mm was recorded. During rainy season silt, clay and other suspended particles contribute towards high turbidity values, while during winter and summer seasons settlement of silt, clay results low turbidity (Thirupathaiah et al., 2012). In summary there has been positive correlation between chl_a, turbidity, TSM and TSS and mostly observed during inflow into the dam after experiencing rainfall. Wickramaarachchi et al. (2013) carried a study to define the relation between turbidity and total suspended solid (TSS) concentration in Gin River at Baddegama. Results showed strong positive correlation (R2 = 0.98) between turbidity and TSS concentration. According to the authors the results strongly suggested turbidity is a suitable monitoring parameter for TSS. In conclusion, remote sensing retrieved data showed positive relationship with lot of in-situ measurements, which means that satellite data can be used to predict water quality of the dam in future.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 55

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

CHAPTER 6

6.0 CONCLUSIONS AND RECOMMENDATIONS

6.1 CONCLUSIONS From this study the following can be concluded

1. Current landuse patterns for the Dikgathong Catchment are dominated by forest and shrub (73.7 %), Settlements (21.1 %) and agricultural fields (2.67 %). Results showed increase in settlements in the catchment due to urbanization and population growth. 2. The water quality of the Dikgathong Dam is polluted due to high levels of COD, TSS and turbidity that exceeds set standards. Five different groups of sampling sites were formed from ten sites using cluster analysis. The Principle Component Analysis identified ten parameters out of twenty. 3. Remote sensing techniques through the use of Quasi Analytical Algorithm and MODIS helped to classify the water quality status of the dam. 4. There exists strong positive correlation between parameters retrieved through remote sensing and in-situ measurements. Strong positive significant correlation was observed

between chl_a and turbidity (r=0.794 and 0.830), TSS (r = 0.819 and 0.770), SO4 COD

(r=0.781 and 0.769), SO4 (r= 0.851 and 0.646) and alkalinity (r= 0.847).

6.2 RECOMMENDATIONS 1. Landuse planning should form the basis of management plan for dam water quality control and protection since the land cover in the catchment area is characterized by increase in settlements, which include urban developments, sewage and industrial effluents which are threats to the dam water quality. The growth rate of urbanization should be slowed down within the catchment and be extended to areas outside the catchment. 2. Water quality monitoring program is recommended for Water Utilities Corporation mainly based on the statistical analysis results. Instead of the initial ten sites sampled in this study, five points are recommended sampling focusing mainly on ten parameters instead of the initial twenty as specified by the PCA.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 56

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

3. Remote sensing techniques through the Quasi Analytical Algorithm helped to classify the water quality status of the dam and as such the method can be deployed as a mechanism for near real time monitoring of water quality in Botswana reservoirs. 4. Based on the significant correlation between the remote sensing retrieved water quality parameters and in-situ measurements, it could be concluded that MODIS images can be used in monitoring and assessment of the water quality in the lake at a point in time

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 57

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

REFERENCES Adakole, J. A., Mbah, C. E.,and Dalla, M. A. (2003). Physicochemical limnology of Lake Kubanni, Zaria-Nigeria. Presented at the 29th WEDC International Conference, Abuja Nigeria. http://wedc.lboro.ac.uk/resources/conference/29/Adakole.pdf.04/06/2016 Aher, K. R and Deshpande, S. M. (2011). Assessment of Water Quality of the Maniyad Reservoir of Parala Village, District Aurangabad: Suitability for Multipurpose Usage. International Journal of Recent Trends in Science and Technology, Volume 1, Issue 3, pp. 91-95. Akbulut, M., Kaya, H., Çelýk, E. Þ., Odabaþi, D. A., Sagir-Odabasi, S. and Selvi, K. (2010). Assessment of Surface Water Quality in the Atikhisar. Ekoloji, 19, 139–149. Akintola, F. O and Gbadegesin, A. (1997). Land-use changes and water quality in impounded water-supply dams in southwest Nigeria. Presented at the Proceedings of Rabat Symposium S4, April-May 1997. IAHS Publ. no, 243, 313–319. Al-Anzi, B. (2012). Assessment of Wastewater Reuse in Kuwait and Its Impact on Amounts of Pollutants Discharged into the Sea. Journal of Environmental Protection, (3) 935-939. Alparslan, E., Aydöner, C., Tufekci, V., and Tüfekci, H. (2007). Water quality assessment of Omerli Dam using remote sensing techniques. Environmental monitoring and assessment, 135(1-3), 391–8. Anhwange, B. A., Agbaji, E. B, and Gimba, E. C. (2012). Impact Assessment of Human Activities and Seasonal Variation on River Benue, within Makurdi Metropolis. International Journal of Science and Technology, 2(5), 248–254. Araoye, P. (2009). The seasonal variation of pH and dissolved oxygen concentration in Asa Lake Ilorin, Nigeria. International Journal of Physical Sciences, 4(5), 271–274. Asif, S., Abed, S., and Pardhan, V. (2012). Ground Water Quality Assessment of Babalgaon , District Latur, Journal of Chemical , Biological and Physical Sciences, 2(1), 501–504. Bernard, A. (2010). Assessing the performance of Dompoase wastewater treatment plant and its effect on water quality of the Oda River in Kumasi. Unpublished MSc Thesis, Kwame Nkrumah University of Science and Technology, Ghana http://wwwdatad.aau.org/handle/123456789/8636. 06/12/2015. Botswana, and South Africa. (2012). Joint Water Quality Baseline Report for Limpopo Basin between Botswana and South Africa. http://www.water.gov.bw/images/Reports/Botswana-

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 58

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

SA_Joint_Water_Quality_Baseline_Report_for_Limpopo_Transboundary_River_2013_(Fi nal).pdf. 29/06/2015. Boyer, J. N., Kelble, C. R., Ortner, P. B., and Rudnick, D. T. (2009). Phytoplankton bloom status: Chlorophyll a biomass as an indicator of water quality condition in the southern estuaries of Florida, USA. Ecological Indicators, 9(6 suppl.), 56–67. Boyles, W. (1997). The science of chemical oxygen demand. Technical Information Series, Booklet, (9), 24. http://www.hach.com/asset-get.download.jsa?id=7639984471. 01/06.2016 Caruso, B. S., Mirtskhulava, M., Wireman, M., Schroeder, W., Kornilovich, B., and Griffin, S. (2012). Effects of Manganese Mining on Water Quality in the Caucasus Mountains, Republic of Georgia. Mine Water and the Environment, 31(1), 16–28. Central Statistics Organisation. (2011). Population and Housing Census, 2011. Statistics Botswana. http://www.cso.gov.bw/images/seminar_report.pdf. 05/06/2016 Chapman. (1996). Water Quality Assessment. Guide to use of biota, sediments and water in enviromental monitoring. http://doi.org/10.4324/9780203476710 02/05/2016. Chatterjee, S. K., Bhattacharjee, I and Chandra, G. (2010). Water quality assessment near an industrial site of Damodar River, India. Environmental Monitoring and Assessment, February 2010, Volume 161, Issue 1, pp 177-189 Chithra, S. V, Nair, M. V. H., Amarnath, A., and Anjana, N. S. (2015). Impacts of Impervious Surfaces on the Environment. International Journal of Engineering Science Invention, 4(5), 2319–6726.

Choudhary, R., Rawtani, P., and Vishwakarma, M. (2011). Comparative study of Drinking Water Quality Parameters of three Manmade Reservoirs i.e. Kolar , Kaliasote and Kerwa Dam, Current World Environment, an International Research Journal of Environmental Science, 6(1), 145–149. Chu, H., Liu, C., and Wang, C. (2013). Identifying the Relationships between Water Quality and Land Cover Changes in the Tseng-Wen Reservoir Watershed of Taiwan, (3), 478–48. Corcoran, E., C. Nellemann, E. Baker, R. Bos, D. Osborn, H. Savelli (eds). 2010. Sick Water? The central role of wastewater management in sustainable development. A Rapid Response Assessment. United Nations Environment Programme, UN-HABITAT, GRID-Arendal. www.grida.no.ISBN: 978-82-7701-075-5. 21/05/2016.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 59

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Dall’Olmo, G., Gitelson, A. A., Rundquist, D. C., Leavitt, B., Barrow, T., and Holz, J. C. (2005). Assessing the potential of SeaWiFS and MODIS for estimating chlorophyll concentration in turbid productive waters using red and near-infrared bands. Remote Sensing of Environment, 96 (2005) 176 – 187 Damodhar, U. and Reddy, M. V. (2013). Impact of pharmaceutical industry treated effluents on the water quality of river Uppanar , South east coast of India. Applied Water Science, June Volume 3, Issue 2, pp 501–514. Ding, J., Jiang, Y., Fu, L., Liu, Q., Peng, Q., and Kang, M. (2015). Impacts of Land Use on Surface Water Quality in a Subtropical River Basin: A Case Study of the Dongjiang River Basin, Southeastern China. Water, 7, 4427-4445. Dladla, B. (2009). The impact of sub-catchment activity on river water quality: A case study of Mbabane River in the Ezulwini Valley. Unpublished MSc Thesis. University of Zimbabwe. Harare. http://www.ir.uz.ac.zw/xmlui/bitstream/thesisdocument.pdf. 15/10/2015 Dlamini,S. Nhapi, I. Gumindoga, W. Nhiwatiwa, T and Dube, T. (2016). Assessing the feasibility of integrating remote sensing and in-situ measurements in monitoring water quality status of Lake Chivero, Zimbabwe. Physics and Chemistry of the Earth, 93 (2016) 2 -11 Duan, W., He, B., Nover, D., Yang, G., Chen, W., Meng, H., Liu, C. (2016). Water Quality Assessment and Pollution Source Identification of the Eastern Poyang Lake Basin Using Multivariate Statistical Methods. Sustainability, 8(2), 133. Dube, T., Gumindoga, W., and Chawira, M. (2014). Detection of land cover changes around Lake Mutirikwi, Zimbabwebased on traditional remote sensing image classification techniques, African Journal of Aquatic Science, volume 39, Issue 1 (2014) pp 89 – 95.

Dube. T. (2009). An Assessment of the Effect of Industrial and Sewage Effluent on Aquatic Invertebrates : A Case Study of A Southern Urban Stream, Journal of Sustainable Development (April), 3(2) 210–214. Dutta, S., Kole, R. K., Ghosh, S., Nath, D., and Vass, K. K. (2005). Impact assessment of lead on water quality of river Ganga in West Bengal, India. Bulletin of Environmental Contamination and Toxicology. November 2005, Volume 75, Issue 5, pp 1012–1019 DWA. (2013). Republic of Botswana Botswana Integrated Water Resources Management & Water Efficiency Plan (Vol. 1). http://www.gwp.org/Global/IWRMWEPlan.pdf. 29/07/2015.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 60

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

EHES (Pty) LTD. Consulting Engineers, Hydrologists and Environmentalists in association with Mantswe Natural Resources (2002). Republic of Botswana Department of Water Affairs EIA study in respect of detailed design of lower Shashe Dam final EIA report. Gaborone, Botswana. EHES (PTY) LTD. Consulting Engineers, Hydrologists and Environmentalists. Elbag, M. (2006). Impact of Surrounding Land Uses on Surface Water Quality. Unpublished MSc Thesis, Worcester Polytechnic Institute.https://www.wpi.edu/Pubs/ETD/Available/etd- 050306-155834/unrestricted/Thesis_Final_draft2.pdf. 23/04/2015 Emeka, E. E. (2015). Comparative Analysis of Effluent Discharge from Emene Industrial Area of Enugu , Nigeria , With National and International Standards, Civil and Environmental Research, 7(9), 84–93. EPA. (2001). Parameters of water quality. Environmental Protection, 133. http://doi.org/10.1017/CBO9781107415324.004. 09/06/2016. Fataei, E. (2011). Assessment of Surface Water Quality Using Principle Component Analysis and Factor Analysis, 3(2), 159–166. Faul, F., Erdfelder, E., Buchner, A., and Lang, A.-G. (2009). Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149–60. Foody, G. M. (2002). Status of land cover classification accuracy assessment. Remote Sensing of Environment, 80 (2002) 185 – 201. Garg, R. K., Rao, R. J., Uchchariya, D., Shukla, G., and Saksena, D. N. (2010). Seasonal variations in water quality and major threats to Ramsagar reservoir , India. African Journal of Environmental Science and Technology, 4(2), 61–76. Garizi, A. Z., Sheikh, V and Sadoddin, A. (2011). Assessment of seasonal variations of chemical characteristics in surface water using multivariate statistical methods. International Journal of Environmental Science and Technology, Volume 8, Issue 3, pp 581-592. Gege, P. (2004). The water color simulator; an integrating software tool for analysis and simulation of optical in situ spectra. Computers and Geosciences, 30 (2004) 523–532. Georgieva, N., Yaneva, Z., and Dospatliev, L. (2010). Analyses of natural water quality in Stara Zagora region according to the parameters sulfates and chlorides, Trakia Journal of Sciences, Vol. 8, Suppl. 2, pp 517-524.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 61

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Gibrilla, A., Bam, E. K. P., Adomako, D., Ganyaglo, S., Osae, S., Akiti, T. T, Agyeman, E. K. (2011). Application of Water Quality Index and Multivariate Analysis for Groundwater Quality Assessment of the Birimian and Cape Coast Granitoid Complex: Densu River Basin of Ghana. Water Quality, Exposure and Health, (2011) 3: 63. Gomes, K., and Ebrary, I. (2009). Wastewater management: A UN-Water Analytical Brief. UN Water, 55. Retrieved from. http://ubc.summon.serialssolutions.com/link/0. 01/06/2016. Gumindoga, W. (2010). Hydrologic impacts of Landuse change in the Upper Gilgel Abay River Basin , Ethiopia.Unpublished MSc Thesis. International Institute for Geoinformation Science and Earth Observation, Netherlands..https://www.itc.nl/library/msc/wrem/gumindoga.pdf. 07/06/2016 Hansen, C. H., Williams, G. P., Adjei, Z., Barlow, A., Nelson, J and Miller, A. W. (2015). Lake and Reservoir Management Reservoir water quality monitoring using remote sensing with seasonal models : case study of five central-Utah reservoirs Reservoir water quality monitoring using remote sensing with seasonal models, 2381(November). http://doi.org/10.1080/10402381.2015.1065937. 7/12/2015. Harrafi, H., Khedri, M., and Karaminejad, K. (2012). Determination of Chemical Oxygen Demand in Spent Caustic by Potentiometric Determination. World Academy of Science, Engineering and Technology, 6(7), 190–192. He, W., Chen, S., Liu, X., and Chen, J. (2008). Water quality monitoring in slightly-polluted inland water body through remote sensing . Guanting Reservoir , Beijing , China. Environmental Science and Engineering (2008)1. http://citeseerx.ist.psu.edu/viewdoi=10.1.1.469.91.pdf. 26/10/2015. Hellweger, F. L., Schlosser, P., Lall, U., & Weissel, J. K. (2004). Use of satellite imagery for water quality studies in New York Harbor. Estuarine, Coastal and Shelf Science Volume 61, Issue 3, November 2004, Pages 437–448. Hoff, M. A. (2013). Control of Agricultural Nonpoint Source Pollution. Unpublished MSc Thesis. Massachusetts Institute of Technology http://www.dspace.mit.edu/bitstream/handleMIT.pdf. 01/06/2016. Huang, J., Zhan, J., Yan, H., Wu, F., and Deng, X. (2013). Evaluation of the Impacts of Land Use on Water Quality: A Case Study in The Chaohu Lake Basin. Scientific World Journal, 2013.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 62

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

http://doi.org/10.1155/2013/329187. 27/05/2016. Hui, L., Daphne, X., Utomo, H. D., Zhi, L., and Kenneth, H. (2011). Correlation between Turbidity and Total Suspended Solids in. Journal of Water Sustainability, 1(3), 313–322. http://doi.org/10.11912/jws.1.3.313-322. 09/06/2016. Hung, T. Le, and Tuyen, V. D. (2014). Estimating suspended sediment concentrations in surface water of trian Lake (Vietnam) using Landsat multispectral images, Vestnik OrelGAU, 3(48), June 2014 57 3(June), 57–64. IOCCG. (2006). IOCCG Report Number 05: Reports of the International Ocean-Colour Coordinating Group Remote Sensing of Inherent Optical Properties : Fundamentals, Tests of Algorithms, and Applications. IOCCG Report 5 (Vol. 5). http://www.ioccg.org/report5.pdf. 30/05/2015. IPCC. (2008). Climate Change and Water. Climate change and water. http://doi.org/10.1016/.2010.08.039. 21/05/2016. Iqbal, J., Mumtaz, M. ., Mukhtar, H., Iqbal, T., Mahmood, S., and Razaq, A. (2010). Particle Size Distribution Analysis and Physico- Chemical Characterization of Chenab River Water At Marala Headworks. Pakistan Journal of Botany, 42(2), 1153–1161. Irenosen, O. G. (2012). Water Quality Assessment of the Owena Multi-Purpose Dam, Ondo State, Southwestern Nigeria. Journal of Environmental Protection, 03(01), 14–25. J. Leskovec, A. Rajaraman, J. U. (2014). Clustering. Mining of Massive Datasets. http://doi.org/10.1007/SpringerReference_34708. 04/06/2016. Jaji, M. O., Bamgbose, O., Odukoya, O. O., and Arowolo, T. A. (2007). Water quality assessment of Ogun river, South West Nigeria. Environmental Monitoring and Assessment. http://doi.org/10.1007/s10661-006-9602-1. 06/12/2015. Jeffares and Green, B. A., and Venture, and G. J. (2014). dikgatlhong_water. Retrieved from http://www.civildesigner.com/press/dikgatlhong_water.pdf. 06/12/2016. Jo-Anne, E. C. K. L. H. J. S. H. (2014). Effects of Suspended Sediment on Freshwater Fish and Fish Habitat. New Zealand Ltd and West Coast Regiona, Landcare Research. http://www.envirolink.govt.nz/. 14/05/2016. Johnston, D., Potter, H., Jones, C., Rolley, S., Watson, I., Pritchard, J., and Agency, T. E. (2008). Abandoned mines and the water environment. Science Report. http://doi.org/978-1-84432-

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 63

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

894-9. 08/06/2016. Jun, X., Shubo, C., Xiuping, H., Rui, X., and Xiaojie, L. (2010). Potential Impacts and Challenges of Climate Change on Water Quality and Ecosystem: Case Studies in Represantative Rivers in China. Journal of Resources and Ecology, 1(200801001), 31–35. Kadewa, W., Moyo, H. ., Phiri, O., & Mumba, P. (2005). Assessment of the Impact of Industrial Effluents on Water Quality of Receiving Rivers in Urban Areas of Malawi, (September). http://doi.org/10.1007/BF03325882. 02/09/2015. Kannel, P. R., Lee, S., Lee, Y., Kanel, S., and Khan, P. (2007). Application of water quality indices and dissolved oxygen as indicators for river water classification and urban impact assessment. Environmental Monitoring and Assessment. http://doi.org/10.1007/s10661-006-9505-1. 06/12/2015. Kebede, Y.K and Tesfu Kebedee, T. (2012). Application of Principal Component Analysis in Surface Water Quality Monitoring, Principal Component Analysis - Engineering Applications, Dr. Parinya Sanguansat (Ed.), ISBN: 978-953-51-0182-6, InTech, Available from: http://www.intechopen.com/books/principalcomponent-analysis-engineering- applications/application-of-principal-component-analysis-in-surface-waterquality- monitoring. 29/04/2016. Khound, N. J., Phukon, P., and Bhattacharyya, K. G. (2012). Physico-chemical studies on surface water quality in the Jia-Bharali River Basin , North Brahmaputra Plain , India. Scholars Research Library, 4(2), 1169–1174. Kibena, J., Nhapi, I., and Gumindoga, W. (2014). Assessing the relationship between water quality parameters and changes in landuse patterns in the Upper Manyame River, Zimbabwe. Physics and Chemistry of the Earth, 67-69, 153–163. Klake, R. K., Nartey, V. K., Doamekpor, L. K., and Dinku, D. E. (2015). Seasonal Variation in Water Quality of the Weija Dam , Ghana, 5(20), 82–88. Larkin, J. (2014). Detecting Long Term Trends in Water Quality Parameters Using Remote Sensing Techniques. Unpublished MSc Thesis. University of Illinois at Urbana-Champaign. Lee, Z., Carder, K. L and Arnone, R. A. (2002b). Deriving inherent optical properties from water color : a multiband quasi-analytical algorithm for optically deep waters, 41(27), 23–30. Ljee, T. (2011). Revue scientifique et Technique. Décembre 2011. Revue Scientifique Et

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 64

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Technique, 100–115. Loveland, T. R., and Dwyer, J. L. (2012). Landsat: Building a strong future. Remote Sensing of Environment, 122(October 2000), 22–29. Lutz, D. S., and Cummings, M. R. (2003). Annual Report. Water Quality Studies – Red Rock and Saylorville Reservoirs Des Moines River , Iowa Water Quality Studies. http://home.engineering.iastate.edu/~dslutz/dmrwqn/Annual%20Report%20CP44/cp44%20 annual%20report.pdf. 27/05/2016 MacKenzie, R. (2010). Clearing the waters over hygiene debate. Nursing times (Vol. 96). Madamombe, C. E., and Rwasoka, D. T. (2005). a case study of lake chivero, paper presented at the 13th WaterNet-WARFSA-GWP-SA Symposium on the 31.10.2012 - 02.11.2012, Johannesburg, South Africa. Category: pp.1–17. Martinez, F. B., Mijares, M. B. B., and Galera, I. C. (2011). Assessment of the Water Quality of Mamba River of Mts . Palaypalay / Mataasna Gulod , Southern Luzon , Philippines, 2011 presented at the International Conference on Chemistry and Chemical Process IPCBEE vol.10 (2011) IACSIT Press, Singapore. Masamba, W. R. L., and Mazvimavi, D. (2006). Water Quality Thamalakane River.An outlet of the Okavango delta. http://www.ubrisa.ub.bw/bitstream/handle/10311/740/. 06/12/2015. Masere, T. P., Munodawafa, A., & Chitata, T. (2012). Assessment of human impact on water quality along Manyame River. International Journal of Development and Sustainability, 1(3), 754–765. Massoud, M. A. (2012). Assessment of water quality along a recreational section of the Damour River in Lebanon using the water quality index. Environmental Monitoring and Assessment. 2012 Jul; 184 (7):4151-60. Matta, G. (2014). A study on physico-chemical Characteristics to assess the pollution status of river Ganga in Uttarakhand, 7(3), 210–217. Mero, R. (2011). Assessment of water quality and spatial distribution of the major pollutants in Ngerengere River Catchment, Tanzania. MSc Thesis, University of Zimbabwe, Harare. www.ir.uz.ac.zw/jspui/bitstream/10646/789/1/Rose%20thesis-27.7.11.pdf. 06/12/2015. Misra, A. K. (2014). Gulf Organisation for Research and Development. International Journal of Sustainable Built Environment, 3(1), 153–165.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 65

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Mladenov, N., Strzepek, K., and Serumola, O. M. (2005). Water quality assessment and modeling of an effluent-dominated stream, the Notwane River, Botswana. Environmental Monitoring and Assessment. http://doi.org/10.1007/s10661-005-5842-8. 06/12/2015. Mohamed, M. F. (2015). Satellite data and real time stations to improve water quality of Lake Manzalah. Water Science, 29(1). http://doi.org/10.1016/j.wsj.2015.03.002. 23/11/2016. Mohammad, M. J., Krishna, P. V, Lamma, O. A., and Khan, S. (2015). Analysis of Water Quality using Limnological Studies of Wyra Reservoir, Khammam District , , India, Original Research Article, 4(2), 880–895. Mohsin, M., Safdar, S., Asghar, F., and Jamal, F. (2013). Assessment of drinking water quality and its impact on residents health in Bahawalpur City. International Journal of Humanities and Social Science, 3(15), 114–128. MRC. (2009). The Limpopo Challenge Program . Baseline Study of the water chemistry of the Limpopo Basin : Country studies from Botswana , Mozambique , South Africa and Zimbabwe . A report prepared by the Mineral Resources Centre , University of Zimbabwe . http://www.waternetonline.ihe.nl/challengeprogram/IR39/Water/quality/baseline.pdf. 16/11/2015. Mustapha. (2008). Assessment of the Water Quality of Oyun Reservoir , Offa , Nigeria , Using Selected Physico-Chemical Parameters, Turkish Journal of Fisheries and Aquatic Sciences 8: 309-319 (2008), 319, 309–319. Mustapha, A., and Abdu, A. (2012). Application of Principal Component Analysis and Multiple Regression Models in Surface Water Quality Assessment. Journal of Environment and Earth Science, 2(2), 16–24. Mustapha, A., and Usman, B. (2014). Sources and Pathway of Environmental Pollutants into Surface Water Resources : A Review, 1(2), 54–59. United Nations. (2015). The United Nations World Water Development Report 2015, Water for a sustainable world, http://unesdoc.unesco.org/images/0023/002318/231823E.pdf. 24/10/2015 Nganje, T., Adamu, C., and Ntekim, E. (2010). Influence of mine drainage on water quality along River Nyaba in Enugu South-Eastern Nigeria. African Journal of Environmental Science and Technology Vol. 4(3), pp. 132-144. Nhapi, I., Siebel, M. A., and Gijzen, H. J. (2007). the Impact of Urbanisation on the Water Quality

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 66

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

of Lake Chivero, Zimbabwe. Water and Environment Journal, 18(1), 44–49. Njoku, J., Ebe, T., and Pat-Mbano, E. (2010). Detection and Mapping of Land Use and Land Cover Classes of a Developing City in Southeastern Region of Nigeria, using Multi-band Digital Remotely-sensed Data. African Research Review, 4(3), 315–330. Nyakungu, T., Wuta, M., and Mudyazhezha, S. (2013). Assessment of the Impact of Waste Disposal on the Concentration of Anions in Ruwa River , Zimbabwe, 4(2), 11–16. Ochieng, G., Seanego, E., and Nkwonta, O. (2010). Impacts of mining on water resources in South Africa: A review. Scientific Research and Essays, 5(22), 3351–3357. Retrieved from http://www.academicjournals.org/sre/pdf/pdf2010/18Nov/Ochieng et al.pdf. 08/06/2016. Ogleni, N., and Topal, B. (2011). Water Quality Assessment of the Mudurnu River, Turkey, Using Biotic Indices. Water Resources Management. http://doi.org/10.1007/s11269-0. 06/12/2015. Olawale, S. A. (2016). Physico Chemical Analysis of Water from Asa River, Ilorin, Nigeria, Imperial Journal of Interdisciplinary Research (IJIR) Vol-2, Issue-3, 2016 (3), 122–129. Olet, E. (2010). Water Quality Monitoring of Roxo reservoir using Landsat Images and In-situ Measurements; MSc Thesis, International Institute of Geoinformation and Science http://cdn.intechopen.com/pdfs/45146/I.pdf..07/12/2015. Owa, F.D,. (2013). Water Pollution : Sources , Effects , Control and Management, Mediterranean Journal of Social Sciences, 4(8), 65–68. Ozbay, Nurgul and Yerel, Suheyla and Ankara, Huseyin (2009) Investigation of Cluster Analysis in Surface Water in Yesilirmak River. Presented at 1st International Symposium on Sustainable Development, June 9-10, 2009, Sarajevo, Bosnia and Herzegovina. http://www.eprints.ibu.edu.ba/id/rreprint/468/contents. 12/05/2016.

Park, H. M. (2005). Comparing Group Means : The T-test and One-way. The Trustees of Indiana University,1–57.http://stat.smmu.edu.cn/DOWNLOAD/ebook/statistics_course.pdf. 04/06/2016. Patil, P., Sawant, D., and Rn, D. (2012). Physico-Chemical Parameters for Testing of Water–a Review. International Journal of Environmental Sciences, 3(3), 1194–1207. Patil, S., Patil, S. S., and Sathe, T. V. (2013). Limnological status of Khanapur freshwater reservoir from Ajara tahsil, (MS), India. International Journal of Science, Environment, 2(6), 1163–1174.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 67

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Perumal, K., and Bhaskaran, R. (2010). Supervised classification performance of multispectral images. Journal of Computing, 2(2), 126–129. Pulugandi, C. (2014). Analysis of water quality Parameters in Vembakottai water reservoir , Virudhunagar district, Research Journal of Recent Sciences, (ISC-2013), 242-247 (2014) Quevaulviller, O., Thomas, O., and Beken, van der. (2006). Wastewater Quality Monitoring and Treatment. John Wiley and Sons, Ltd.Brussels, Belgium. Richardson, L.L and LeDrew E.F. (eds.) (2006), Remote Sensing of Aquatic Coastal Ecosystem Processes: Science and Management Applications, 81-109. 2006 Springer. The Netherlands. Ritchie, J. C., Zimba, P. V, and Everitt, J. H. (2003). Remote Sensing Techniques to Assess Water Quality, Photogrammetric engineering and remote sensing 69(6), 695–704. Rosado-Torres, M. A. (2008). Evaluation and development of bio-optical algorithms for chlorophyll retrieval in western Puerto Rico. Unpublished PhD Thesis, University of Puerto Rico. Mayagüez Campus. http://www.uprm.edu/bio-optics/docs/thesis2marcos.pdf. 31/05/2016. Salah, E. A. M., Turki, A. M., and Al-othman, E. M. (2012). Assessment of Water Quality of Euphrates River Using Cluster Analysis, 2012(December), 1629–1633. Salem, F. A. M., and Ramadan, A. M. T. F. A. (2009). An Efficient K-Means with Good Initial Starting Points, Georgian Electronic Scientific Journal: Computer Science and Telecommunications 2009, No.2 (19) 2(2), 47–57. Sawant, R., and Chavan, N. (2013). Water Quality Status Of Mahagaon Reservoir From Gadhinglaj Tahsil From . International Journal of Science, Environment, 2(6), 1196–1204. Schaeffer, B. A., Schaeffer, K. G., Keith, D., Lunetta, R. S., Conmy, R. and Gould, R. W. (2013). Barriers to adopting satellite remote sensing for water quality management) Barriers to adopting satellite remote sensing for water quality management Barriers to adopting satellite remote sensing for water quality management. International Journal of Remote Sensing, 3421(21), 7534–7544. Shock, C. C., Pratt, K., and Station, M. E. (2003). Phosphorus effects on surface water quality and phosphorus tmdl development, Western Nutrient Management Conference. 2003. Vol. 5. Salt Lake City, UT.5, 211–220.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 68

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Singh, P. (2014). Studies on seasonal variations in physico-chemical parameters of the river Gomti (U.P.), International Journal of Advanced Research (2014), Volume 2, Issue 2, 82-862(2). Stone, N., Shelton, J. L., Haggard, B. E., and Thomforde, H. K. (2013). Interpretation of Water Analysis Reports for Fish Culture. SRAC Publication, 4606(4606), 1–12. Suthar, S., Sharma, J., Chabukdhara, M., and Nema, A. K. (2010). Water quality assessment of river Hindon at Ghaziabad, India: Impact of industrial and urban wastewater. Environmental Monitoring and Assessment. http://doi.org/10.1007/s10661-009-0930-9. 06/12/2015. Tabriz, S., Mojid, M., and Wyseure, G. (2012). Irrigation suitability of North Bengal Sugar Mill’s effluent and its impact on soil properties. Journal of the Bangladesh Agricultural University. 9(2): 283–290. Tali, J. A., Divya, S., and Murthy, K. (2013). Influence of Urbanization on the Land Use Change : A Case Study of Srinagar City, American Journal of Research Communication, 1(7), 271– 283. Talling, J. F. (2010). pH, the CO 2 system and freshwater science. Freshwater Reviews, 3, 133– 146. Tessema , Mohammed, B. T. (2014). Assessment of Physico-chemical Water Quality of Bira Dam, Bati Wereda. J Aquac Res Development, 5(6), 1–4. Thirupathaiah, M., Samatha, C., and Sammaiah, C. (2012). Analysis of water quality using physico-chemical parameters in lower manair reservoir of , . International Jornal of Environmental Sciences, 3(1), 172–180. Thomlinson, J. R., Bolstad, P. V., and Cohen, W. B. (1999). Coordinating methodologies for scaling landcover classifications from site-specific to global: Steps toward validating global map products. Remote Sensing of Environment, 70(1), 16–28. Trenberth, K. E. (2005). The Impact of Climate Change and Variability on Heavy Precipitation, Floods, and Droughts. Encyclopedia of Hydrological Sciences, 1–11. http://www.cgd.ucar.edu. 01/06/2016. Tzortziou, M., Subramaniam, A., Herman, J. R., Gallegos, C. L., Neale, P. J., and Harding, L. W. (2007). Remote sensing reflectance and inherent optical properties in the mid Chesapeake Bay, (72), 16–32. Uchchariya, D. K., and Saksena, D. N. (2012). Study of Nutrients and Trophic Status of Tighra

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 69

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Reservoir , Gwalior ( Madhya Pradesh ), India, 2(8), 97–111. Ullah, Z., Khan, H., Waseem, A., Mahmood, Q., and Farooq, U. (2013). Water Quality Assessment of the River Kabul at Peshawar, Pakistan: Industrial and Urban Wastewater Impacts. Journal of Water Chemistry and Technology, Vol. 35, No. 4, pp. 170–176. UNEP. (2010). Clearing the waters. A focus of water quality solutions. Nursing times (Vol. 96). http://www.unep.org/PDF/Clearing_the_Waters.pdf. 04/06/2016. UNEP/GEMS. (2006). United Nations Environment Programme G E M S. http://www.gemswater.org/. 29/04/2016. Urama, K. C., and Ozor, N. (2010). Impacts of climate change on water resources in Africa : The Role of Adaptation, (December), 1–29. Usali, N and Hasmadi Ismail, M. (2010). Use of Remote Sensing and GIS in Monitoring Water Quality. Journal of Sustainable Development, 3(3). pp. 228 – 238. Verma, P. U., Chandawat, D. K and Solanki, H. A. (2013). Pollution Status of Nikol Lake Located in Eastern Ahmedabad , 2(8), 3603–3609. Wagner, P. D., Kumar, S and Schneider, K. (2013). An assessment of land use change impacts on the water resources of the Mula and Mutha Rivers catchment upstream of , India. Hydrology and Earth System Sciences, 17(1), 2233–2246. Walakira, P and Okot-okumu, J. (2011). Impact of Industrial Effluents on Water Quality of Streams in Nakawa-Ntinda, Uganda. Applied Science Environmental Management, 15 (2) 289 – 296 Water Safety Report. (2003). International Life Science Institute. Retrieved from http://www.ilsi.org/Documents/ILSIWaterSafety_FactSheet_January14_V2.pdf. 25/10/2015. WHO. (2006). Guidelines for drinking-water quality, 1. Retrieved from http://www.who.int/water_sanitation_health/dwq/gdwqvol32ed.pdf. 14/05/2016. WHO. (2011). Guidelines for Drinking-water Quality. www.who.int/water_sanitation_health/publications/2011/dwq_chapters/en/. 27/05/2016. WMO. (2012). Technical material for water resources assessment. Technical Report Series. WMO 1095. Retrieved from http://www.wmo.int/pages/prog/hwrp/publicatios. 25/10/2015. Xia, L. L., Liu, R. Z and Zao, Y. W. (2012). Correlation Analysis of Landscape Pattern and Water

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 70

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Quality in Baiyangdian Watershed. Procedia Environmental Sciences, 13(2011), 2188–2196. Yaakub, A., Norulaini, N and Ab, N. (2012). Water Quality Status of Kinta River Tributaries Based on Land Use Activities, 33, 178–182. Yadav, R. K., Kishor, J and Yadava, R. L. (2013). Effects of temperature variations on microstrip antenna. International Journal of Networks and Communications, 3(1), 21–24. Zamani, M., Sadoddin, A and Garizi, A. Z. (2012). Assessing land-cover / land-use change and its impacts on surface water quality in the Ziarat catchment, IRAN. 2012 International Congress on Environmental Modelling and Software Managing Resources of a Limited Planet, Sixth Biennial Meeting, Leipzig, Germany R. Seppelt, A.A. Voinov, S. Lange, D. Bankamp (Eds.) http://www.iemss.org/society/index.php/iemss-2012-proceedings. 27/05/2016. Zheng, G., Stramski, D and Reynolds, R. A. (2014). Remote Sensing of Environment Evaluation of the Quasi-Analytical Algorithm for estimating the inherent optical properties of seawater from ocean color : Comparison of Arctic and lower-latitude waters. Remote Sensing of Environment, 155, 194–209. http://doi.org/10.1016/j.rse.2014.08.020. 28/05/2016. ZHOU, Z., LIU, L., and ZHAO, Y. (2010). Design of the water quality monitoring system for inland lakes based on remote sensing data. 3rd International Conference on cartography and GIS 15-20 June, 2010, Nessebar, Bulgaria. http://cartography- gis.com/pdf/53_Zhou_China_paper.pdf. 26/10/2015.

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 71

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

LIST OF APPENDICES

Appendix 1: Results of physical, chemical and microbiological sampled on the 07th April 2016

Parameter Units D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 Temp C 22.6 22.1 23.1 23.1 22 22.8 23 23.6 21.8 24.2 pH pH 6.73 7.07 6.97 6.97 6.94 7.04 6.82 6.82 7.2 7.2 EC uS/cm 167 178 179 185 189 189 181 181 209 211 Turbidity NTU 12.47 28.06 2.96 1.97 2.85 2.8 1.84 1.72 15.17 15.06 TDS mg/l 114 124 110 122 130 122 104 98 130 136 TSS mg/l 18 0 20 16 26 32 50 20 36 14 Ca mg/l 21.73 23.31 21.76 22.62 23.15 22.9 21.62 21.94 16.7 16.69 Mg mg/l 5.12 5.68 5.48 5.71 5.78 5.98 5.49 5.64 4.04 4.07 Na mg/l 7.44 8.04 7.21 8.02 8.32 8.57 8.31 7.98 21.56 22.21 K mg/l 8.86 7.19 6.52 6.42 7.27 6.04 6.82 6.38 5.6 5.42 Cl mg/l 3.64 4.28 4.05 4.61 5.02 5.46 4.74 4.69 16.57 16.51 TA mg/l 122 87.33 82.41 85.61 86.43 85.77 81.92 85.36 70.6 71.01 NO3 mg/l 0.18 0.15 0.25 0.2 0.21 0.18 0.29 0.26 0.24 0.23 NO2 as N mg/l 0 0 0 0 0 0 0 0 0 0 SO4 mg/l 1.22 1.07 1.11 1.1 1.13 1.31 1.19 1.18 4.45 4.46 PO4 mg/l 0 0 0 0 0 0 0 0 0 0 COD mg/l 7.6 7.6 7.6 15.2 30.4 38 7.6 22.8 7.6 TH mg/l CaCO3 96.34 104.9 99.39 103.42 105.32 106.34 98.22 101.2 74.91 75.13 Algae counts/ml 66.3 134.4 192.3 56.9 48.5 52.3 35.5 189.5 70 28 TP mg/l 0.077 0.043 0.056 0.027 0.029 0.085 TN mg/l 0.952 0.561 0.986 0.483 0.411 0.998

Appendix 2: Results of physical, chemical and microbiological sampled on the 23rd March 2016

Parameter Units D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 Temp C 27 27.3 26.7 27.2 26.8 26.8 27 26.9 26.9 26.8 pH pH 7.41 7.21 7.43 7.48 7.29 7.19 7.23 7.57 6.97 7.05 EC uS/cm 190 200 198 213 210 212 216 210 160 120 Turbidity NTU 23.42 3.9 11.94 2.03 3.01 3.72 2.18 8.11 108 293.3 TDS mg/l 150 106 124 156 134 146 134 150 121 93 Ca mg/l 21.77 21.93 20.99 23.1 33.38 22.79 23.46 23.17 17.42 15.59 Mg mg/l 5.68 7.11 5.85 6.24 8.93 6.43 6.32 5.87 5.78 4.36 Na mg/l 7.42 8.67 7.85 8.73 8.62 8.99 8.98 9.73 8.59 4.73 K mg/l 4.73 5.12 4.68 4.51 4.96 5.14 5.18 5.28 4.3 2.51

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 72

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Cl mg/l 4.12 4.75 4.58 5.25 5.18 5.44 5.35 5.61 4.66 6.44 TA mg/l 88.4 98.48 90.04 97.5 94.79 96.02 99.96 97.01 71.75 50.43 NO3 mg/l 0.15 0.14 0.07 0.1 0.16 0.002 0.14 0.1 0.73 1.46 NO2 as N mg/l 0 0 0 0 0 0 0 0 0 0 SO4 mg/l 0.96 0 0.96 1 0.99 0.97 0.97 1.12 1.33 1.64 PO4 mg/l 0 0 0 0 0 0 0 0 0 0 COD mg/l 41.8 57 11.4 11.4 26.6 11.4 57 19 26.6 19 mg/l TH CaCO3 101.6 133.22 100.51 105.98 131.49 109.78 110.54 106.11 91.03 58.39 Algae mg/l 39.2 50.4 110.13 120.4 100.8 173.6 73.7 27.1 14.9 14

Appendix 3: Results of physical, chemical and microbiological sampled on the 10th March 2016 Parameter Units D1 D2 D3 D5 D6 D8 D9 D10 Temp C 27 27 26.7 26.8 26.8 26.9 26.9 26.8 pH pH 6.81 7.52 7.46 7.7 7.67 7.2 7.04 7.17 EC uS/cm 99 226 196 220 220 231 143 204 Turbidity NTU 513 86 27.12 517.2 215.9 TDS mg/l 65 175 180 178 140 216 100 190 TSS mg/l 620 20 100 12 2 44 808 308 Ca mg/l 10.4 22.24 13.88 21.31 24.24 20.43 10.73 12.7 Mg mg/l 3.39 5.67 4.44 5.38 5.1 4.58 4.01 4.33 Na mg/l 3.48 9.46 8.56 9.64 10.19 14.89 10.5 15.43 K mg/l 3.86 5.85 10.64 9.6 9.82 5.5 6.73 9.02 Cl mg/l 3.94 7.1 5.39 5.64 6.12 11.07 5.8 11.04 TA mg/l 41.33 107.83 88.07 100.94 105.86 92.58 58.71 79.7 NO3 mg/l 3.33 0.26 0.5 0.19 0.73 1.01 2.16 2.07 NO2 as N mg/l 0 0 0.11 0 0 0.47 0 0.14 SO4 mg/l 1.97 1.61 1.6 1.15 1.22 2.99 2.51 3.2 PO4 mg/l 0 0 0 0 0 0 0.05 0 COD mg/l 38.71 30.97 23.23 15.48 30.97 46.45 30.97 TH mg/l CaCO3 53.84 102.15 71.17 97.44 100.54 88.66 59.77 67.32

Appendix 4: Results of physical, chemical and microbiological sampled on the 24th February 2016 Parameter Units D1 D2 D5 D6 D9 D10 Temp C 29.1 29.2 29.2 29.4 29.2 28.9 pH pH 6.59 7.59 7.64 7.62 7.19 7.29 EC uS/cm 136 226 226 224 242 225 Turbidity NTU 367 3.11 38.71 TDS mg/l 66 122 102 114 110 140 TSS mg/l 272 6 8 6 8 62 Ca mg/l 17.33 26.42 26.84 25.23 27.38 32.91

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 73

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Mg mg/l 4.24 6.14 6.36 5.98 6.24 7.14 Na mg/l 3.94 9.81 9.32 8.3 9.65 10.15 K mg/l 6.01 10.99 11.84 7.11 12.78 8.34 Cl mg/l 2.41 5.66 5.64 5.62 6.24 6.75 TA mg/l 55.1 103.48 103.48 100.45 107.17 112.18 NO3 mg/l 3.94 0.13 0.14 0.09 1.1 0.1 NO2 as N mg/l 0.42 0 0 0 0 0 SO4 mg/l 1.78 0.95 0.97 0.95 0.99 1.03 PO4 mg/l 0 0 0 0 0 0 COD mg/l 24 16 8 8 8 TH mg/l CaCO3 72.55 116.44 119.3 112.15 119.66 140.87

Appendix 5: Results of physical, chemical and microbiological sampled on the 11th February 2016 Parameter Units D1 D2 D5 D6 D9 D10 Temp C 30.2 29.8 29.9 30.1 29.6 30.6 pH pH 7.57 7.72 7.73 7.67 7.63 8.04 EC uS/cm 229 219 220 218 230 224 Turbidity NTU 18.1 3.15 56 TDS mg/l 134 120 116 104 126 106 TSS mg/l 6 0 0 0 0 93 Ca mg/l 25.79 25.21 23.94 25.41 26.37 26.53 Mg mg/l 6.51 6.22 6.03 6.63 6.6 6.55 Na mg/l 9.99 9.88 10.14 9.16 9.96 11.03 K mg/l 8.4 7.74 8.08 7.38 7.83 8.44 Cl mg/l 5.69 5.91 5.56 5.56 6 6.74 TA mg/l 107.01 103.4 97.42 105.86 106.44 98.65 NO3 mg/l 0.06 0.13 NO2 as N mg/l 0 0 0 0 0 0.23 SO4 mg/l 1.02 1.2 1.05 1.02 0.97 1.06 PO4 mg/l 0 0 0 0 0 0 COD mg/l 0 0 0 17 4 TH mg/l CaCO3 117.92 114.08 109.35 117.96 120.11 120.09

Appendix 6: Results of physical, chemical and microbiological sampled on the 27th January 2016 Parameter Units D1 D2 D5 D6 D9 D10 Temp C 30.5 30 30.2 30.3 30.1 30.8 pH pH 7.61 7.7 7.68 7.61 7.71 8.14 EC uS/cm 266 212 216 216 223 234 TDS mg/l 146 130 136 124 142 162 TSS mg/l 40 12 24 56 56 56 Ca mg/l 28.23 24.73 23.97 24.65 25.58 29.36

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 74

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Mg mg/l 7.03 6.63 6.49 6.56 6.7 6.98 Na mg/l 15.98 10.06 9.63 8.9 10.28 13.68 K mg/l 7.21 11.61 7.15 6.84 12.93 6.81 Cl mg/l 20.23 5.61 5.48 5.58 6.06 9.76 TA mg/l 93.73 100.45 98.56 100.45 101.93 99.55 NO3 mg/l 0.96 0.17 0.17 0.17 0.2 NO2 as N mg/l 0.1 0 0 0 0 0.13 SO4 mg/l 6.6 1.02 0.93 0.98 1.07 2.74 PO4 mg/l 0 0 0 0 0 0.001 COD mg/l 0 0 71 150.5 467 39.6 TH mg/l CaCO3 122.67 116.26 113.21 115.48 188.96 128.28

Appendix 7: Results of physical, chemical and microbiological sampled on the15th January 2016 Parameter Units D1 D2 D5 D6 D9 D10 Temp C 25.5 25.3 25.5 25.7 25.3 25.3 pH pH 6.67 7.72 7.75 7.76 7.46 7.15 EC uS/cm 140 217 215 215 233 200 TDS mg/l 76 124 120 120 142 185 TSS mg/l 72 4 54 Ca mg/l 18.01 24.57 24.17 24.41 25.91 12.3 Mg mg/l 4.24 6.41 6.73 6.42 6.83 4.42 Na mg/l 4.03 9.47 9.87 12.92 9.84 15.4 K mg/l 6.02 8.51 10.73 12.48 7.21 8.91 Cl mg/l 2.41 5.32 5.18 5.2 6.34 9.8 TA mg/l 66.5 103.4 97.33 101.02 106.44 79 NO3 mg/l 3.75 0.13 0.1 0.14 0.14 2.1 NO2 as N mg/l 0.42 0 0 0 0 0.14 SO4 mg/l 1.56 0.9 0.86 0.88 0.96 2.2 PO4 mg/l 0 0 0 0 0 0 COD mg/l 28 30.1 TH mg/l CaCO3 80 114.05 115.68 113.73 120.85 70

Appendix 8: Standards and Guidelines for surface waters

Parameter Units SWQO 1999 EPA 2001

pH pH 6.5 – 8.5 5.0 – 9.0 Turbidity NTU 5 EC uS/cm 1800 1000 TDS mg/l 500

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 75

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

TSS mg/l 25 – 50 Sodium mg/l 200 Calcium mg/l Magnesium mg/l Chloride mg/l 250 Potassium mg/l COD mg/l 10 40 1000 Total Hardness mg/l CaCO3 50 – 350

Total Alkalinity mg/l CaCO3 Phosphates mg/l Nitrates mg/l 10 50 Sulphates mg/l 200 – 250 Total Nitrogen mg/l Total Phosphorus mg/l Algae Counts/ml

Appendix 9: Botswana rainfall map

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 76

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Appendix 10: Rainfall data for Francistown airport from January to April 2016

January February March April Days (mm) (mm) (mm) (mm) 1 0 0 0 0 2 0 0 0 0 3 0.254 0 0 0 4 0 0 0 1.52 5 0 1.016 0 1.02 6 0 4.32 0 0 7 0 10.7 3.05 0.51 8 0 1.016 12.7 11 9 1.016 0 13.97 0 10 1.27 0 22.4 0 11 0 4.32 3.3 0 12 3.302 0 3.81 0 13 2.794 0.254 0 0 14 1.27 0 1.016 0

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 77

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

15 11.7 0 0 0 16 8.6 0 0.762 0 17 1.27 0 11.4 0 18 0 0 9.4 0 19 0 0 18.8 0 20 0 0 7.4 0 21 0 0 0 0 22 0 4.1 0 0 23 0 5.08 0.51 0 24 7.62 5.8 0 0 25 15.5 4.06 0 0 26 5.59 14.5 0 0 27 0 3.3 0 0 28 0 2.54 0 0 29 3.81 0.51 0 0 30 3.81 0 0 31 0 0 0 http://www.accuweather.com/en/bw/francistown/33329/weather-forecast/33329

Appendix 11: derived iops at 440nm and chl_a for 13th january 2016

Derived absorption Chl_a Derived backscattering TSM aph QAA coefficient (mg/m3) bbp QAA coefficient (mg/l)

D1 0.59 0.0335 17.8 0.1320 0.004773 23.9 D2 0.62 0.0335 18.5 0.1286 0.004773 23.4 D3 0.39 0.0335 11.9 0.1342 0.004773 24.4 D4 0.37 0.0335 11.0 0.1228 0.004773 22.3 D5 0.24 0.0335 7.30 0.0237 0.004773 4.01 D6 0.28 0.0335 8.40 0.1269 0.004773 23.1 D7 0.32 0.0335 9.48 0.3229 0.004773 59.2 D8 0.26 0.0335 7.66 0.1051 0.004773 19.0 D9 0.40 0.0335 12.0 0.0248 0.004773 4.21 D10 0.53 0.0335 15.7 0.1263 0.004773 22.9

Appendix 12: derived iops at 440nm and chl_a for 26th january 2016

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 78

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Derived Derived absorption Chl_a Backscattering TSM aph QAA coefficient (mg/m3) bbp QAA coefficient (mg/l) D1 0.1377 0.0335 4.11 0.0147 0.004773 2.34 D2 0.2361 0.0335 7.05 0.0872 0.004773 15.7 D3 0.2640 0.0335 7.88 0.0662 0.004773 11.9 D4 0.2254 0.0335 6.73 0.0524 0.004773 9.30 D5 0.2118 0.0335 6.32 0.1414 0.004773 25.7 D6 0.3716 0.0335 11.1 0.1369 0.004773 25.0 D7 0.7743 0.0335 23.1 0.1374 0.004773 24.9 D8 0.5263 0.0335 15.7 0.1359 0.004773 24.7 D9 0.8183 0.0335 24.4 0.1276 0.004773 23.2 D10 0.1259 0.0335 3.76 0.1263 0.004773 22.9

Appendix 13: derived iops 440 nm and chl_a for 11th february 2016

Derived Derived bbp absorption Chl_a backscattering TSM aph QAA coefficient (mg/m3) QAA coefficient (mg/l) D1 0.3062 0.0335 9.14 0.0274 0.004773 4.70 D2 0.2253 0.0335 6.72 0.1709 0.004773 31.2 D3 0.2065 0.0335 6.16 0.0662 0.004773 11.9 D4 0.1432 0.0335 4.27 0.0621 0.004773 11.1 D5 0.1148 0.0335 3.43 0.0545 0.004773 9.70 D6 0.2009 0.0335 5.99 0.0532 0.004773 9.46 D7 0.3340 0.0335 9.97 0.1137 0.004773 20.6 D8 0.2022 0.0335 6.03 0.0663 0.004773 11.9 D9 0.1933 0.0335 5.77 0.1276 0.004773 23.2 D10 0.3259 0.0335 9.73 0.1173 0.004773 21.3

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 79

Assessment of the Water Quality of the Dikgathong Dam in Botswana using Remote Sensing Techniques

Appenidx 14: derived iops and chl_a for 05th april 2016

Derived Derived bbp absorption Chl_a Backscattering TSM aph QAA coefficient (mg/m3) QAA coefficients (mg/l) D1 0.0689 0.0335 2.06 0.0068 0.004773 0.9 D2 0.5922 0.0335 17.7 0.1099 0.004773 19.9 D3 0.2125 0.0335 6.34 0.0286 0.004773 4.92 D4 0.8038 0.0335 23.9 0.0572 0.004773 10.2 D5 0.3181 0.0335 9.49 0.0677 0.004773 12.1 D6 0.3370 0.0335 10.1 0.0711 0.004773 12.8 D7 0.4003 0.0335 11.9 0.0362 0.004773 6.32 D8 0.5605 0.0335 16.7 0.1016 0.004773 18.4 D9 0.7768 0.0335 23.2 0.1845 0.004773 33.7 D10 0.5771 0.0335 17.2 0.1844 0.004773 33.6

MSC IWRM 2015/16, KAGISO MOSIMANEGAPE 80