(AGENPARL) – Thu 22 May 2025 Media Release from Wits University
Immediate release
22 May 2025
Wits Climate Risk Modelling project receives Bezos Earth Fund grant to research AI-aided weather forecasting
The Wits Climate Risk Modelling project has been named one of the inaugural recipients of the prestigious Bezos Earth Fund’s AI Grand Challenge for Climate and Nature, securing a US$50,000 (approximately R925,000) Phase I grant to lead a ground-breaking project aimed at transforming weather forecasting across the African continent.
The project, led by Professor Rendani Mbuvha from the Wits School of Statistics and Actuarial Science will make use of advanced artificial intelligence – including Graph Neural Networks and diffusion models – to deliver accurate, high-speed weather and climate predictions in regions where current systems are insufficient.
“We are deeply grateful to the Bezos Earth Fund for selecting us as a grantee in this highly competitive AI Grand Challenge for Climate and Nature,” says Mbuvha. “Together with our partners, we’re thrilled to democratise skilful AI-driven weather forecasts that will generate multisectoral impact across the Global South. This grant award is a vote of confidence in our work and inspires us to accelerate inclusive, community-informed solutions for climate resilience.”
The Bezos Earth Fund launched the AI Grand Challenge with a total commitment of US$100 million (around R1.85 billion) to support bold, AI-powered ideas aimed at addressing climate change and biodiversity loss. The Climate Risk Modelling project was selected for its potential to address a major climate and development challenge in Africa – the lack of reliable weather data and forecasting infrastructure, particularly in rural and underserved areas dependent on rain-fed agriculture.
At present, Africa has just one-eighth of the weather station density cover recommended by the World Meteorological Organisation – and that number is decreasing. This scarcity limits access to accurate early-warning systems that are critical for climate adaptation.
To help close this gap, the Wits-led initiative will expand observation networks by deploying automatic weather stations in under-served areas, working closely with grassroots research communities and national weather services. It will also combine ground-based, satellite, and historical climate data to build robust, open-access forecasting models that can be used across the continent and beyond.
“I am delighted that the Bezos Earth Fund is supporting the climate risk modelling work of Professor Mbuvha,” says Professor Stephen Jurisich, Head of the Wits School of Statistics & Actuarial Science. “It really strengthens the research of the School in this important area. Many congratulations to Rendani and his collaborators.”
The project is also notable for its focus on ethical and inclusive AI development, ensuring that data privacy, environmental sustainability, and equitable access are prioritised. All outputs – including data, models, and forecasts – will be open-sourced to maximise their value to global and local efforts to improve climate resilience.
By securing this Phase I grant, the Wits team is now eligible to further compete in an innovation sprint for a Phase II implementation grant of up to US$2 million (approximately R37 million) to further develop and scale the project over the next two years.
This recognition from the Bezos Earth Fund affirms Wits University’s leadership in applying frontier technologies to real-world challenges, and marks a significant step forward in building climate resilience for vulnerable communities across Africa.
For more information or media interviews contact:
Rendani Mbuvha
Links
Bezos Fund Official announcement
https://www.bezosearthfund.org/news-and-insights/phase-i-grants-ai-grand-challenge-climate-nature
Article on Wits website
https://www.wits.ac.za/news/latest-news/research-news/2025/2025-05/wits-named-recipient-of-prestigious-grant-for-ai-climate-research.html