An Approach for Including Uncertainty in Integrated Water Resources Assessments within Large River Basins of Southern Africa.

Abstract:

There are many large basins in southern Africa that are mostly ungauged but may have some streamflow observations either on the main river or on tributaries. Many of the streamflow records are, however, of poor quality or impacted by largely unquantified and non-stationary development impacts. All water resources assessments are therefore uncertain and model setups are difficult to validate in traditional ways. The paper presents a method for practical uncertainty assessment using a semi-distributed (sub-basin) model. The method uses a 2-stage approach where the first stage involves obtaining ‘behavioural’ parameter sets to represent the incremental natural streamflow for each sub-basin. The criteria for ‘behavioural’ are based on a series of constraints on model output that can be developed from the available gauged data or from regional assessments of natural hydrological functioning. The second stage simulates the whole basin based on sampling the ‘behavioural’ incremental flow parameter sets, as well as samples of additional individual parameter values representing downstream routing parts of the model and development impacts. One of the perceived advantage of the method is that all the ensembles at the total basin outlet are made up of behavioural inputs for all sub-basins. The method is also flexible in terms of the uncertainty range of the constraints, which might be expected to be narrow (low uncertainty) in well gauged sub-basins, or areas where our understanding of flow regime characteristics is good, but much wider (higher uncertainty) in other parts of the basin. The paper briefly explains the approach and discusses some of the issues associated with its application using examples from southern Africa.