Page 226 - AI for Good Innovate for Impact
P. 226
AI for Good Innovate for Impact
Use Case 10: Quartz Solar AI Cloudcasting - predicting cloud
movements through satellite imagery
Organization: Open Climate Fix
Country: United Kingdom
Contact Person(s):
Primary Contact: Dan Travers – dan@ openclimatefix .org
Secondary Contact: Kasia Krasucka – kasia@ openclimatefix .org
Additional Contact: James Fulton, PhD – james@ openclimatefix .org
1 Use Case Summary Table
Item Details
Category Climate Change/Natural Disaster
Problem Addressed Inaccurate solar forecasts due to unpredictable cloud cover, creating
challenges in renewable energy integration and grid reliability.
Key Aspects of Solu- - AI model predicts future satellite imagery up to 3 hours ahead
tion - Secondary AI model forecasts solar PV generation based on cloud
predictions
Technology Keywords machine learning, neural network, computer vision
Data Availability EUMETSAT public satellite data:[4]
Metadata (Type of Visual, video, satellite
Data)
Model Training and - simVP architecture to forecast satellite images
Fine-Tuning - Two models: one for cloud prediction, one for PV power forecasting
Testbeds or Pilot Open Climate Fix. n.d. Cloudcasting App Repository.[4]
Deployments
Code Repositories Open Climate Fix. n.d. SAT Pred Repository. [6]
Alan Turing Institute. n.d. Cloudcasting Project Repository.[7]
190
