Mohomi et al. (2026) Projections of extreme rainfall in South Africa using CMIP6 ISIMIP models

Identification

• Journal: Climate Dynamics
• Year: 2026
• Date: 2026-03-01
• Authors: Tumelo Mohomi, V M Stepanenko, А. I. Medvedev, Inos Dhau, Hector Chikoore, Mary‐Jane M. Bopape
• DOI: 10.1007/s00382-026-08079-2

Research Groups

• Department of Geography and Environmental Studies, University of Limpopo, Sovenga, South Africa
• Department of Geography and Environmental Science, University of Venda, Thohoyandou, South Africa
• Research Computing Center, Lomonosov Moscow State University, Moscow, Russia
• Faculty of Geography, Lomonosov Moscow State University, Moscow, Russia
• Moscow Center of Fundamental and Applied Mathematics, Moscow, Russia
• Department of Geography, University of South Africa, Florida, South Africa
• South African Environmental Observation Network, National Research Foundation, Pretoria, South Africa
• College of Science, Engineering & Technology, University of South Africa, Florida, South Africa

Short Summary

This study projects extreme rainfall in South Africa using CMIP6 ISIMIP global climate models under SSP1-2.6 and SSP5-8.5 scenarios, finding an overall increase in extreme rainfall events, with a trend towards a drier west and a wetter east, posing significant risks to water resources and infrastructure.

Objective

• To project future rainfall and extreme events in South Africa by evaluating the performance of CMIP6 ISIMIP models and their multi-model ensemble against observational data, and then studying projected future rainfall patterns under low- (SSP1-2.6) and high-mitigation (SSP5-8.5) scenarios, focusing on changes in extreme rainfall events in the eastern interior and complex topography.

Study Configuration

• Spatial Scale: South Africa, with a primary focus on the eastern plateau (24–30.4°E and 23.5–30°S). Model data was at 0.5° × 0.5° horizontal resolution, and reanalysis data at 0.1° × 0.1° horizontal resolution.
• Temporal Scale:
  • Historical/Evaluation Period: 1979–2014 (36 years).
  • Future Projection Periods: 2020–2100, divided into near-future (2020–2055) and far-future (2065–2100).

Methodology and Data

• Models used:
  • Phase 6 of the Coupled Model Intercomparison Project (CMIP6) Inter-Sectoral Impact Model Intercomparison Project (ISIMIP3b) Global Climate Models (GCMs).
  • Specific models: GFDL-ESM4, MPI-ESM-1-2-HR, MRI-ESM2-0, IPSL-CM6A-LR, UKESM1-0-LL.
  • Multi-model ensemble mean (MME) was used for projections.
  • Emission scenarios: Shared Socioeconomic Pathway 1-2.6 (SSP1-2.6, low forcing) and SSP5-8.5 (high forcing).
  • Future rainfall change quantified using 5th, 50th, and 95th percentiles (corresponding to drought, normal, and flood conditions).
  • Statistical validation: Taylor diagrams, Root Mean Square Error (RMSE), Coefficient of Determination (R²), Percent of Bias (PBIAS), Nash-Sutcliffe Efficiency (NSE), and bootstrapping significance tests.
• Data sources:
  • South African Weather Service (SAWS) daily rainfall observations from seven stations on the eastern plateau (1979–2014).
  • ERA5-Land reanalysis dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) (0.1° × 0.1° resolution, 1979–2014).

Main Results

• Model Performance: The multi-model ensemble (MME) consistently outperformed individual models and showed strong agreement with observations (correlation coefficients between 0.55 and 0.7 for stations, and 0.8 for reanalysis). Individual models exhibited a similar spatial bias, underestimating rainfall over the escarpment (exceeding -30% per year) and overestimating it in lowlands and coastal areas (exceeding 20% per year). Models struggled to accurately reproduce rainfall variability (maximum and minimum extremes) and fine-scale details in complex terrain.
• Annual Rainfall Projections: Generally drier conditions are projected (5th and 50th percentiles), particularly along the coast and northeast in the far-future. The eastern half of South Africa is likely to experience increased extreme wet conditions (95th percentile), with annual rainfall projected to rise from about 100 to 220 mm per year in the far-future under SSP5-8.5. The southwestern Cape is vulnerable to dryness, with decreases up to 200 mm per year under SSP5-8.5 in the far-future.
• Seasonal Rainfall Projections (DJF - Austral Summer): Pronounced wet and dry extremes are projected over eastern South Africa. Extreme wet conditions (95th percentile) are projected to increase by 40 to 80 mm per month in the eastern interior. Conversely, a severe decrease in rainfall (up to -100 mm per month) is projected in northeastern South Africa (5th percentile).
• Seasonal Rainfall Projections (JJA - Austral Winter): Persistent and intensifying dry and wet extremes are projected along the coast. The southwestern Cape is particularly vulnerable to severe dry conditions, with decreases up to -16 mm per month under SSP5-8.5. Coastal areas show a complex shift, with the 5th percentile projecting increased dryness and the 95th percentile projecting increased wetness.
• Monthly Rainfall Projections: January projects rainfall increases in the eastern interior, with extreme wet conditions (95th percentile) increasing by 35–45 mm across scenarios. October may experience increased extreme dry conditions (5th percentile), with more than 48 mm decrease in the far-future under SSP5-8.5.
• Heavy Rainfall Days (> 20 mm/day): Projected to increase significantly in eastern South Africa, exceeding 10 days in the eastern interior, and 20 to 50 days in the eastern region (east of 30°E) under SSP5-8.5. A general decrease is projected for much of drier western South Africa. Gauteng, Limpopo, Mpumalanga, and KwaZulu-Natal provinces are identified as hotspots for increased heavy rainfall days.

Contributions

• This study is the first to apply CMIP6 ISIMIP models for extreme rainfall projections in South Africa, providing robust projections by leveraging their enhanced resolution and multi-model ensemble approach.
• It delivers actionable insights for policymakers to develop targeted adaptation and mitigation strategies against climate change impacts on water resources, agriculture, and infrastructure.
• The research highlights the persistent challenges of GCMs in resolving fine-scale atmospheric dynamics over complex terrain and emphasizes the need for improved model parameterization schemes.
• It underscores the significant need for enhancing model parameterization schemes and reducing biases, especially for complicated landscape aspects and mesoscale elements in South Africa for future hydrological studies.

Funding

• South Africa’s National Research Foundation (NRF)

Citation

@article{Mohomi2026Projections,
  author = {Mohomi, Tumelo and Stepanenko, V M and Medvedev, А. I. and Dhau, Inos and Chikoore, Hector and Bopape, Mary‐Jane M.},
  title = {Projections of extreme rainfall in South Africa using CMIP6 ISIMIP models},
  journal = {Climate Dynamics},
  year = {2026},
  doi = {10.1007/s00382-026-08079-2},
  url = {https://doi.org/10.1007/s00382-026-08079-2}
}