ASSESSING THE POTENTIAL IMPACT OF RAINFALL CHANGE ON SOUTHERN AFRICAN RIVER FLOWS RAKHEE LAKHRAJ-GOVENDER 11598 SUPERVISOR: PROF STEFAN GRAB A research report submitted to the Faculty of Science, University of Witwatersrand in partial fulfilment of the requirements for the degree of Master of Science. Johannesburg 2010
CONTENTS Content Declaration Acknowledgements Abstract List of Figures List of Tables List of Abbreviations List of Symbols ii iii iv v vii ix x x Chapter One: Introduction 1.1 Background to the study... 1 1.2 Aims... 2 1.3 Study area... 4 Chapter Two: Literature Review 2.1 Introduction... 11 2.2 Fluvial systems... 12 2.3 South African fluvial systems and global change... 13 2.4 Rainfall trends in South Africa... 17 2.5 Strategies to evaluate the impact of climate change on hydrology... 19 Chapter Three: Methodology 3.1 Data selection... 22 3.2 Analysis... 23 3.3 Methodological limitations... 26 Chapter Four: Results and Discussion 4.1 Introduction... 30 4.2 Trends in inter-annual rainfall... 30 4.3 Trends in seasonal rainfall and river flow... 50 4.4 Five year moving average for annual rainfall and river flow trends... 80 Chapter Five: Conclusion Conclusion...100 References...106 Appendices...117 ii
Declaration I hereby declare that this research report is my own, unaided work. I am submitting it in partial fulfilment of the requirements for a MSc degree in the University of the Witwatersrand, Johannesburg. It has not been submitted for examination in any other university before. Rakhee Lakhraj-Govender 12 May 2010 iii
Acknowledgements Thank you to the South African Weather Services for providing rainfall data and to Charlotte McBride for timeously emailing these data. Thank you to Department of Water Affairs and Forestry for the easily accessible river flow data. This study would not have been possible without the provision of such data. Thank you to the Almighty for carrying me through difficult times, for the courage and strength you have given me and for blessing me with so much. To my beautiful baby boy Tahir, thank you for bringing so much love, laughter and happiness into our lives. The journey we have taken together has been filled with so many challenges. Thank you to my dear husband Kasavan for your love, encouragement, support, time and patience throughout the duration of my degree. Thank you for taking care of Tahir when I had to write. Thank you to my parents for working so hard so we could get the best out of life. Thank you to my brother, Naren for always encouraging me to reach new heights. Thank you, Irene for baby sitting every time I needed to meet my supervisor. I would like to express my sincere gratitude to Prof Stefan Grab, my supervisor, for his encouragement, guidance, patience, enthusiasm and motivation. I sincerely appreciate it. Thank you. iv
Abstract The primary aims of this study are firstly to determine long-term rainfall and river flow trends for a variety of South African catchments over the period 1905 to 2008 and secondly to compare and correlate the rainfall and river flow data for these catchments. Historical rainfall trends were analysed using Mann Kendall statistic and Sen s method was used for the determination of the magnitude of the trend. The Kendall tau method was used to determine the correlation between rainfall and river flow. To this end, long-term annual, seasonal and 5-year moving average annual rainfall and river flow data for the Tugela, Mgeni, Orange, Vaal and Breede River catchments in southern Africa were analysed. The Western Cape region displayed an increase in annual rainfall over the period 1918 to 2008, whilst the eastern-central, western-central and eastern regions show decreasing annual rainfall trends over the period 1905 to 2008. Eighty percent of the rivers significantly declined in annual river flow over the period of record. The majority of the rainfall stations experienced insignificant changes for the seasonal analysis during early, mid and late rainfall seasons, with the early season defined as October and November, the mid season defined as December to February, and the late season defined as March and April for the summer rainfall regions. For the winter rainfall region the early rainfall season was defined as April and May, the mid rainfall season as June and July and the late rainfall season as August and September. Only the Villiers rainfall station exhibited significant decreasing trends and the rainfall at the Mistley station in the Mgeni River catchment displayed a significant decreasing trend during the late rainfall season. Of the catchments studied, the Tugela River indicated a 73% decline in flow during the early rainfall season, followed by the Vaal River (69 % decline during the early rainfall season), both of which were significant for all seasons. The Mgeni and Tugela Rivers displayed significant decreasing trends in seasonal flow at the 95% significance level, with the exception of the early rainfall season for the Mgeni at Table mountain gauge. The Orange River flow displayed an insignificant increasing trend during all 3 seasons. A study of the Orange River catchment rainfall data indicated that the record length has an important role in the outcome of the percentage change over time. The 5-year moving average examined the frequency, duration and amplitudes of each rainfall and river flow station in order to determine long-term trends. The frequency of wet and dry v
periods within the data sets was similar across the majority of stations. The Breede, Mgeni, Tugela and Vaal Rivers displayed significant decreasing trends for the 5-year moving average annual flow, whilst the Orange River 5-year moving average flow exhibited an increasing trend due to the duration of flow. The average rainfall over the Breede River catchment displayed a significant increase in rainfall with time, indicating wetter conditions in the western region of southern Africa. However, the Breede River flow significantly decreased over time. The average annual rainfall over the Vaal River catchment characterized by a long wet period followed by an extended dry period, as well as the insignificant increase in the amplitudes over time and the decrease in amplitudes of the wet periods, indicate drier conditions over the Vaal River The 5-year moving average rainfall over the Orange River catchment indicates an insignificant increasing trend, whilst the Orange River flow significantly increased over the period 1910 to 2008. The significant decline in rainfall over both the Mgeni and Tugela River catchments, together with significant declines in the Mgeni and Tugela River flows, indicate drier conditions with time over the eastern region of southern Africa. vi
List of Figures Figure 1: Map of South Africa showing rivers pertinent to this study. Figure 2: Google Earth image showing location of Tugela river flow and rainfall stations. Figure 3: Google Earth image showing location of Mgeni at table mountain river flow station and rainfall stations. Figure 4: Google Earth image showing the Western Cape river flow station and rainfall stations. Figure 5: Google Earth image showing location of Vaal river flow and rainfall stations. Figure 6: Google Earth image showing the Orange river station and rainfall stations. Figure 7: Time series showing the annual rainfall for the different stations in the Vaal River Figure 8: Time series showing the average rainfall in the Vaal River catchment and annual river flow at the Schoolplaats station. Figure 9: Time series showing the rainfall at different stations in the Orange River Figure 10: Time series showing the average rainfall in the Orange River catchment and annual river flow at the Orange at Aliwal North station. Figure 11: Time series showing the rainfall at the Lesotho stations in the Orange River Figure 12: Time series showing the rainfall at the New Hanover and Mistley stations and river flow in the Mgeni River Figure 13: Time series showing the rainfall at different stations in the Tugela River Figure 14: Time series showing the average rainfall and the Tugela River flow in the Tugela River Figure 15: Time series showing the annual rainfall at different stations and river flow in the Breede River Figure 16: Early, mid and late rainfall season for Villiers rainfall station in the Vaal River Figure 17: Early, mid and late rainfall season for Bloemhof rainfall station in the Vaal River Figure 18: Early, mid and late rainfall season for Klerksdorp rainfall station in the Vaal River Figure 19: The Vaal River flow at the Vaal at Schoolplaats gauge during the early, mid and late vii
rainfall seasons. Figure 20: Rainfall trends for the early, mid and late rainfall seasons at Lille rainfall station in the Orange River Figure 21: Rainfall trends for the early, mid and late rainfall seasons at Middelplaats rainfall station in the Orange River Figure 22: Early, mid and late rainfall season for Zastron rainfall station in the Orange River. Figure 23: The Orange River flow trends during the early, mid and late rainfall seasons. Figure 24: Rainfall trends for the early, mid and late rainfall seasons at the Thaba Tseka rainfall station in Lesotho. Figure 25: Rainfall trends for the early, mid and late rainfall season at the Semonkong rainfall station in Lesotho. Figure 26: Rainfall trends for the early, mid and late rainfall seasons at Moorside rainfall station in the Tugela River Figure 27: Rainfall trends for the early, mid and late rainfall seasons at Swartwater rainfall station in the Tugela River Figure 28: Rainfall trends for the early, mid and late rainfall seasons at Tugela Ferry rainfall station in the Tugela River Figure 29: The Tugela River flow trends during the early, mid and late rainfall seasons. Figure 30: Rainfall trends for the early, mid and late rainfall seasons at New Hanover rainfall station in the Mgeni River Figure 31: Rainfall trends for the early, mid and late rainfall seasons at Mistley rainfall station in the Mgeni River Figure 32: The Mgeni River flow trends during the early, mid and late rainfall seasons. Figure 33: Rainfall trends for the early, mid and late rainfall seasons at Malabar rainfall station in the Western Cape. Figure 34: Rainfall trends for the early, mid and late rainfall seasons at Touwsrivier rainfall station in the Western Cape. Figure 35: Breede River flow trends during the early, mid and late rainfall seasons in the Western Cape. Figure 36: The 5-year moving average of total annual rainfall for the different stations in the Vaal River viii
Figure 37: The 5-year moving average of total annual average rainfall for the Vaal River Figure 38: The 5-year moving average annual flow at the Vaal River at the Schoolplaats gauge. Figure 39: The 5-year moving average annual total for the different rainfall stations in the Orange River Figure 40: The 5-year moving average annual total for the average rainfall over the Orange River Figure 41: The moving average rainfall for the for the different rainfall stations in Lesotho. Figure 42: The 5-year moving average of Orange River flow at the Aliwal North station. Figure 43: The 5-year moving average annual rainfall for the different rainfall stations in the Mgeni River Figure 44: The 5-year moving average of total annual average rainfall for the Mgeni River Figure 45: The 5-year moving average of Mgeni River flow at the Table Mountain gauge. Figure 46: The rainfall for the different rainfall stations in the Tugela River Figure 47: The 5-year moving average of total annual average rainfall for the Tugela River Figure 48: The 5-year moving average for the Tugela River flow at the Tugela Ferry station. Figure 49: The rainfall for the different rainfall stations and the river flow in the Breede River Figure 50: The 5-year moving average of total annual average rainfall for the Breede River Figure 51: The 5-year moving average annual flow at the Bree at Ceres river flow gauge. List of Tables Table 1: The spatial distribution of rainfall and river flow stations in southern Africa. Table 2: Rainfall and river flow data sets showing data gaps. Table 3: Mann Kendall trend statistic results for the annual rainfall across southern Africa. Table 4: Mann Kendall trend statistic results for the annual river flow across South Africa. Table 5: The Kendall tau correlation coefficients for annual rainfall at the various rainfall stations and river flow stations in southern Africa. ix
Table 6: Rainfall and river flow percentage changes observed over selected time periods for the early, mid and late rainfall seasons at southern African stations. Table 7: Mann Kendall trend statistic results for the seasonal rainfall across South Africa and Lesotho. Table 8: Mann Kendall trend statistic results for the seasonal river flow across South Africa. Table 9: Kendall tau correlation coefficients for river flow and rainfall at stations during different rainfall seasons. List of Abbreviations CSM: Climate Systems Model DWAF: Department of Water Affairs and Forestry DEAT: Department of Environmental Affairs and Tourism ENSO: El Niño and Southern Oscillation RHP: Rivers Health Programme UK: United Kingdom USA: Unites States of America List of Symbols Sgn: Sign Z: Mann Kendall normalized test statistic S: sum of the difference between data points n: the number of values in the data set p-value: probability value Sens method Q: slope B: constant indicating the vertical axe crossing of the slope line. x
Kendall tau τ: Kendall tau S: Kendall score P :the number of times y increases as x increases M : the number of times y decreases as x increases xi