Assessment of Uncertainties Associated with the use of Alternative Satellite-Based Rainfall Estimates in Pitman Modelling
Date
2024-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch University
Abstract
South Africa is a semi-arid water-scarce country experiencing increased climatic variability and hence the management of its water resources is important to the livelihood of its population. For effective management, there is a need to have a robust system that allows for continuous monitoring and simulation of the available water resources. However, the number of suitable observation stations in South Africa has been observed to be on a declining trend over the past few decades. This impedes the efficient management of water resources. South Africa makes use of the WR2012 database as the reference in the management of its water resources. The database contains at present monthly flows for the period 1921-2009. This database has been outdated for more than a decade and continues to age. The WR2012 report recommends a continuous update of this database. However, the cost and sustainability of a continuous update, given the declining trend in available observed rainfall data, is challenging. This research seeks to find a solution to the identified ever-increasing challenge of declining rainfall datasets. This research identified the availability of easily accessible satellite-based rainfall data products as an alternative to the dwindling observed rainfall data, for the simulation of hydrological flows in the Pitman model. It then developed a CHIRPS-based Pitman model framework, which builds on the existing experience and knowledge of the Pitman model and CHIRPS (Climate Hazards Group Infrared Precipitation with Station data). The developed framework aimed at providing an alternative dynamic, quantifiable and up-to-date rainfall database while simplifying the process of simulating flows by removing the bottlenecks associated with data accessibility. The outcome of this research also focused on improving credibility, hence confidence in the use of CHIRPS-based simulated flows, for efficient water resources management, by assessment of associated uncertainties. The successful complementary use of CHIRPS rainfall estimates for simulating flows in this research thus provides a positive advancement towards embracing modern technology in the monitoring, simulation and management of water resources for South Africa. In performing this research, study drainage regions (G, B, V & L) were selected based on the existing primary drainage regions and rainfall zones of South Africa. Drainage region G belongs to the winter rainfall zone, drainage region L ‘All year’ rainfall zone, while drainage regions B and V belong to the summer rainfall zone. These rainfall zones have different rainfall mechanisms influencing how CHIRPS estimates correspond to observed rainfall. However, CHIRPS estimates illustrated a strong correlation (R2 > 0.7) in all the study drainage regions apart from drainage region L, experiencing an ‘All-year’ rainfall. The relationships between CHIRPS and observed catchment rainfall were used to adjust CHIRPS data. Calibrated CHIRPS data was then used in setting up and running the Pitman model. Calibration (1981-2009) and validation (2010-2019) were done on the Pitman model and results indicated good similarity between observed flows and simulated CHIRPS-based flows for the period of analysis. Eight hydrological indices were identified and used in assessing the ability of the simulated CHIRPS-based flows to reproduce different aspects of observed catchment hydrological responses. Using a ±15% uncertainty window, hydrological indices were evaluated and CHIRPS-based flows illustrated 78%, 73% and 80% suitability for drainage regions B, V and G respectively. Drainage region L had suspect results owing to the poor quality of data from the available gauge stations. Further, climate change simulations were done in representative quaternary catchments in study drainage regions B, V & G. The Coupled Model Intercomparison Project-Phase 5 (CMIP5) climate change data was downloaded and using a baseline scenario of 1981-2019, climate change signals for the near (2021-2060) and far (2061-2099) future scenarios were computed. The climate change signals were then used to adjust input climate data for the Pitman model and simulate climate change-induced flows. The results of climate change analysis illustrated a general decrease in simulated flows ranging from -0.45% (RCP 4.5, 2061-2099) in drainage region B to -40.31% (RCP 8.5, 2061-2099) in drainage region V. These results are in concurrence with previous research findings towards a consensus of a drier climate change future for South Africa. This research finds CHIRPS-based rainfall estimates to be a suitable alternative to the decreasing observed rainfall data and recommends it for use in the Pitman model simulation of observed and climate change flows for South Africa for the effective management of the available water resources. A novel contribution to the water resources of South Africa is made by this research in providing a framework that allows for the collaboration of the already published WR2012 data with CHIRPS estimates in generating updated flows and climate change-induced flows in a simplified process.