Page 232 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Use Case 12: AI for Weather Prediction for Africa (Project Umoya)

Organization: Ishango�ai

Country: United Kingdom

Contact Person(s):
Primary: Eunice Baguma Ball, eunice@ ishango .ai | +44 7857 449119

1 Use Case Summary Table

Item Details
Category Climate Change/Natural Disaster

Problem Addressed Lack of localized AI-based weather prediction (AIWP) models and
expertise for the African continent

Key Aspects of Solution - Community building among African AI experts
- Local dataset collection and curation
- Fine-tuning AIWP models like GraphCast for Africa
- Supporting local weather services and institutions

Technology Keywords AIWP, GraphCast, machine learning, climate adaptation, reanalysis
data, low-cost sensors, weather forecasting

Data Availability No public or private access yet (future datasets planned)
Metadata (Type of Data) Reanalyses, Automated Weather Station data, Trained model param-
eters
Model Training and Models such as GraphCast and Aurora will be trained or fine-tuned
Fine-Tuning using African weather datasets including ERA5 and station data
Testbeds or Pilot Deploy- Proof-of-concept in progress – community training, data sourcing,
ments and local forecasting capability building

2 Use Case Description

2�1 Description

Project Umoya (isiZulu for "wind") addresses a critical gap in African weather forecasting by
building local capacity in AI for Weather Prediction (AIWP). While models like DeepMind’s
GraphCast show excellent performance globally, Africa lacks locally tailored solutions due to
limited technical expertise, infrastructure, and data availability.

Objectives:
• Create Africa’s first open-source AIWP model, built by Africans for Africa

196