Abstract:
The 21st century is faced with various climate related problems, amongst them water stress brought about by increased frequency of extreme hydrological droughts. Freshwater resources are sensitive to variable climates and are bound to be strongly impacted by climate change. This in turn has negative impacts on both the quantity and quality of water resources required to meet human and environmental demands. Apart from climate driven factors, other forces contributing to the impairment of water resources availability include but are not limited to population growth, rapid urbanization, various land use activities and water pollution. Africa is considered one of the most vulnerable continents by virtue of an
inherently variable climate and because of low adaptive and coping capacities. The
consequences of high climatic variability and climate change are likely to be felt more
severely in arid and semi-arid regions such as Botswana, where water resources are
scarce and most livelihoods rely entirely on rain fed agriculture for sustenance. Equitable
and sustainable water allocation in such regions requires appropriate and timely water
management strategies that can adequately account for the available water resources.
Such strategies should be well informed by scientific findings as well as availability of
reliable hydrological records. However, most regions especially in southern Africa are
characterised by hydrological data paucity mainly because of inconsistent monitoring or
due to lack of resources to carry out the monitoring programme. Deficiencies in hydrological data can partly be addressed by calibrating a hydrological model which can then be used to simulate future hydrological conditions in a catchment. If the calibrated model responds well to the catchment specific characteristics and processes it can then be used as a baseline proxy for data upon which, certain hydrological variables can be predicted. Moreover, a well calibrated and validated model gives the assurance that the model can reliably represent the catchment hydrological characteristics at any given time. If prediction can be reliably carried out, data on some of the missing variables can be derived as this may be the case in data sparse regions. Hydrological modelling therefore becomes more important for solving problems related to climate variability and change on the one hand and addressing water and food security on the other hand. Hence, once calibrated model can be used for various water use scenarios and for predicting water availability in the future. The predictive model outputs can be used to inform decision making in water resources management particularly where the catchment is ungauged or lacks quality hydrological data. The main aim of this study was to calibrate a hydrological model for the sub basins of the Limpopo River located in Botswana. As an example of model application, the calibrated model was then used to predict streamflow in the future (2025 -2050) under the context of climate change. To further validate the calibrated model, two drought indices were compared consisting of the Standardized Streamflow Index (SSI,
generated from simulated streamflow) and the Standardised Precipitation
Evapotranspiration Index (SPEI).The study applied a three-fold process where, firstly the semi-distributed Pitman hydrological model was set up and calibrated for the sub-basins in the study area. Second, by coupling with downscaled and bias corrected Global Climate Model (GCM) datasets the calibrated model was then used to predict streamflow in the future under the context of climate change. The GCMs consisted of the Representative
Concentration Pathway 4.5 and were extracted from the Coordinated Regional Climate
Downscaling Experiment (CORDEX). Finally, a comparative analysis of drought was
performed using the SSI and SPEI. The Pitman model has been successfully calibrated for the Limpopo sub-basins in Botswana. This can be inferred from the statistical objective performance functions where both the Nash-Sutcliffe coefficient of efficiency (CE) and the coefficient of determination (R2) were > 0.5 while the percent bias was <20 % on average. Generally, streamflow is expected to decrease in the near future on average by ≥25%.
Despite slight differences in the frequency of occurrence of future droughts, this study revealed a close positive correlation between the streamflow derived drought index (SSI) and SPEI. There were no significance differences at the 95% confidence interval between SSI and SPEI, thus further reinforcing the capability of the calibrated model to simulate future hydrological conditions in the Limpopo sub-basins of Botswana. The study concludes that the calibrated model can adequately represent the hydrological response characteristics in the study area and can therefore be used to inform decision making in managing the basin’s water resources. The calibrated model can also be applied to predict different hydrological variables. Nonetheless, it is always necessary to acknowledge predictive uncertainty when simulating hydrological components under future conditions of climate change.