Local Off-Grid Weather Forecasting
Multi-modal Earth observation data for weather prediction in remote areas
Local Off-Grid Weather Forecasting
This work addresses the challenge of weather forecasting in off-grid and remote locations where traditional meteorological infrastructure is limited or absent. By leveraging multi-modal Earth observation data, we enable accurate local weather predictions for underserved regions.
The Challenge
Billions of people worldwide live in areas with limited meteorological station coverage and no access to real-time weather data, yet face vulnerability to extreme weather events.
Our Approach
We develop a multi-modal Earth observation framework that combines:
- Satellite data from optical, thermal, and microwave sensors
- Deep learning architecture for data fusion
- Minimal computational requirements for edge deployment
- Local calibration and uncertainty quantification
Authors
Qidong Yang, Jonathan Giezendanner, Daniel Salles Civitarese, Johannes Jakubik, Eric Schmitt, Anirban Chandra, Jeremy Vila, Detlef Hohl, Chris Hill, Campbell Watson, Sherrie Wang