Rainfall uncertainty, Climate change and Water resources in the Limpopo basin, Botswana *Kebuang Piet Kenabatho, Neil McIntyre, Howard Wheater Department of Civil and Environmental Engineering, Environment and Water Resources Engineering, Imperial College London, SW7 2AZ, UK. 1.INTRODUCTION To keep up with the ever-escalating water demand in Botswana, there is currently significant investment towards developing more dams in the Limpopo basin. The major issues are that (1) the hydrology of this region is poorly understood, (2) there is limited information on historic rainfall observations yet with (3) extended periods of missing records. This could lead to (4) high uncertainties on water resources planning models. Furthermore, (5) knowledge on how uncertainty in future climate projections will affect water resources systems is completely inadequate. Focusing on rainfall and hydrological uncertainty, this project aims to apply rainfall and hydrological models to further understand the hydrology of this region under current and future climate states. Such work is necessary to ensure more robust water resources plans for the future. FIGURE1: The Limpopo basin A multi-site continuous time stochastic rainfall model, the generalised linear model (GLM) (Chandler and Wheater, 2002) was used to infill historic rainfall data, to generate multiple realisations of rainfall (with uncertainty) for the current rainfall series (Figure 2) in the study area (Figure 1) 3. HYDROLOGICAL MODEL A semi-distributed version of the IHACRES model (Ye et al., 1997) was used for the hydrology. Stochastic infilling of rainfall data allows calibration of a hydrological model under input uncertainty (Figure 3) FIGURE 5: Monthly temp (a) and rainfall (b) projections from 2 GCMs 10 CSMK3 HADCM3 5 Temp Changes(deg.Celc) 2. STOCHASTIC RAINFALL MODEL FIGURE 2: Rainfall results-GLMs 0 2 4 6 b a A2-2020s 8 10 12 8 10 12 10 CSMK3 HADCM3 5 0 2 4 A2-2050s 6 10 CSMK3 HADCM3 5 dam A2-2080s 0 2 4 6 8 10 12 [months] 4. CLIMATE CHANGE PROJECTIONS FIGURE 3: Calibration results for flow (left) and cumulated volume (right) The stochastic rainfall model was used to downscale outputs of global circulation models (GCMs) to generate rainfall at a basin scale under scenarios of climate change using multiple GCM experiments (Figure 5) 5. RESERVOIR PERFORMANCE The rainfall model, together with the uncertain hydrological model, is then used to generate multiple realisations of reservoir inflow over a 100-year period under the current and future rainfall scenarios. A proposed 382 106 m3 reservoir at the outlet of the catchment, which is part of Botswana’s national water resource strategy, is re-evaluated in light of the extended inflow data and the estimated uncertainty (Figure 4). FIGURE 4: Reservoir performance for stationary climate and A2 scenario 6. CONCLUSIONS Results show that the uncertainty has a considerable effect on the reliability of the reservoir; for example, the proportion of time for which demand for water was met ranged from [77 to 100% ]-stationary climate, [0-76%]- projected future changes, over the different flow realisations used. In view of these it is proposed that adaptation measures such as supply restrictions should be imposed when the reservoir level reaches a certain threshold to control shortfalls especially during dry periods ACKNOWLEDGEMENTS We thank the Department of Water Affairs and Department of Meteorological Services (Botswana) for providing data used in this study. We also thank the Commonwealth Scholarship Commission (UK) for sponsoring Piet’s research at Imperial College London. *kp.kenabatho06@imperial.ac.uk REFERENCES 1. Chandler, R. E. & Wheater, H. S. (2002) Analysis of rainfall variability using generalized linear models: a case study from the west of Ireland. Water Resources Research. 38, 1192, doi:10.1029/2001WR000906, 2002. 2. Ye, W., Bates, B. C., Viney, N. R., Sivapalan, M. & Jakeman, A. J. (1997) Performance of conceptual rainfall–runoff models in low-yielding catchments. Water Resour.ces Research. 33, 153–166.