Hydrological Dynamics Assessment of Basin Upstream–Downstream Linkages Under Seasonal Climate Variability

Abstract The impacts of climate change are already being felt, not only in terms of increase in temperature but also in respect of inadequate water availability. The Mkomazi River Basins (MRB) of the KwaZulu-Natal region, South Africa serves as major source of water and thus a mainstay of livelihood for millions of people living downstream. It is in this context that the study investigates water flows abstraction from headwaters to floodplains and how the water resources are been impacted by seasonal climate variability. Artificial Neural Network (ANN) pattern classifier was utilized for the seasonal classification and subsequence hydrological flow regime prediction between the upstream–downstream anomalies. The ANN input hydroclimatic data analysis results covering the period 2008–2015 provides a likelihood forecast of high, near-median, or low streamflow. The results show that monthly mean water yield range is 28.6–36.0 m3/s over the Basin with a coefficient of correlation (CC) values of 0.75 at the validation stage. The yearly flow regime exhibits considerable changes with different magnitudes and patterns of increase and decrease in the climatic variables. No doubt, added activities and processes such as land-use change and managerial policies in upstream areas affect the spatial and temporal distribution of available water resources to downstream regions. The study has evolved an artificial neuron system thinking from conjunctive streamflow prediction toward sustainable water allocation planning for medium- and long-term purposes.