Modelling uncertainty and reliability for water resource assessment in South Africa (MURRA)

J. Glenday, J.L. Tanner, D. Gwapedza

Sponsor: Water Research Commission
Collaborators: A. Rebelo (ARC, South Africa), P. Holden (UCT ACDI, South Africa), P. Metho (Zutari, South Africa), S. Gokool (UKZN CWRR, South Africa), F. Jumbi (SAEON, South Africa)

April 2022–March 2024

This project comprised several engagement activities to develop a better shared understanding of uncertainty in hydrological modelling and ways to address it. Hydrological modelling has become central to water resources management and catchment management in South Africa. With ever-growing pressure on the nation’s water supply systems, basing decisions on reliable estimates of surface and groundwater resource availability is critical. In recent decades, South Africa’s meteorological and hydrological monitoring infrastructure has declined severely. The sector relies on models to fill in gaps in monitoring and to forecast the future, predicting streamflow and groundwater under future climates, land uses, and management practices. As reliance on modelling grows, so too should attention to model uncertainty: quantifying it, finding practical ways to reduce it, and accounting for it in decision making.

Despite a growing body of global and local research on uncertainties in hydrological modelling, it remains common practice in applied and operational contexts for model outputs to be presented as single values or timeseries, without indicating the uncertainty inherent in the prediction. Without any quantitative indication of uncertainty around a predicted value, decision makers are ill-equipped to consider risks when using these values to make their decisions. This can lead to inappropriate management with negative consequences.

To address issues around insufficient consideration of hydrological modelling uncertainty, this project had the following aims and objectives:

Aims: 

• To improve shared understanding of model structural uncertainty, and the role of modeller decision making, and the potential scale of impact across the hydrological modelling community;
• To empower the community to identify practical strategies for assessing, communicating, and ideally reducing modelling uncertainty.

Objectives:

• Design and host an open ‘model-a-thon’ activity in which participants from across the water sector model the same catchment area, given the same input and calibration data, and apply the model to the same alternative scenario.
• Assess and discuss the diversity of modelling approaches, structures, and output predictions in the ‘model-a-thon’ activity as a means of collective scoping of the issue of structural uncertainty.
• Host synthesis engagements to discuss the implications of the findings around modelling uncertainty for future water resources studies and initiate visioning of how these studies can be done in a way that accounts for and, where feasible, aims to reduce uncertainty.
• Synthesise the engagement outcomes into a policy brief.
• Promote online resources that can build capacity across the South African hydrological modelling community, specifically a ‘wiki’ website about modelling tool options and capabilities (https://hydromodel-sa-wiki.saeon.ac.za/) and the online question-and-answer platform (Stack Exchange) to facilitate information sharing.

Last Modified: Tue, 06 Aug 2024 14:57:17 SAST