| dc.description.abstract | This thesis developed and tested a reproducible, basin-to-municipality workflow for mapping
compound water-security risk in South Africa, a data-sparse country spanning 13 climate
zones. Two complementary hydrological models were first benchmarked: the locally
calibrated ACRU and the global PCR-GLOBWB 2. After climate-weighted gauge quality-
control and raster–vector harmonisation, six CMIP5 (RCP 8.5) and five CMIP6 (SSP 5-8.5)
forcings were propagated through the models to produce a 63-member runoff ensemble for
2026-2045, 2054-2073 and 2081-2099. Runoff, groundwater recharge, desalination
withdrawals and discharge from PCR-GLOBWB were combined into a gridded water-
availability layer, while a Water-Security Index (WSI) -capturing accessibility, affordability
and acceptability of supply- was compiled from national household surveys and regulatory
audits. Physical and social layers were overlaid using Local Moran’s I to detect statistically
significant high-risk (HH) and low-risk (LL) clusters.
The results confirmed a robust south-west drying / north-east wetting dipole across models,
yet showed compound risk peaking in the Limpopo–Gauteng mining belt, where low
adaptive capacity offsets projected runoff gains, whereas many Western-Cape municipalities
avoid hotspot status thanks to diversified supply portfolios. Hotspots expand between 26 °S
and 30 °S under all scenarios, while LL refugia contract to a narrow Atlantic fringe by 2099.
Limitations include equal weighting of WSI components and static socio-economic baselines;
future work should deploy explainable machine-learning classifiers and downscale SSP
demand trajectories to create fully dynamic risk surfaces. Even so, the study delivers the first
nationally consistent maps of future compound water risk for South Africa, demonstrates how
social context reshapes hydrological hazard rankings, and offers an open, updateable
framework for equity-centred water-adaptation planning. | |
