AI Research
1h ago
MIT engineers create tool to simulate extreme weather events without past data
Aug 24, 2026
AI Summary
Engineers at MIT have developed a machine-learning algorithm that generates plausible extreme weather scenarios without relying on historical extreme event data. This tool aims to help planners and policymakers prepare for unprecedented events by estimating their characteristics, such as intensity and area of impact.
- The new method, called Extreme Event Aware or 'η-learning', generates worst-case scenarios for extreme weather events like storms, heat waves, and wildfires.
- It uses daily weather records and maps to learn statistical patterns, allowing it to create plausible future extreme events without needing past extreme data.
- The algorithm can model events that have never been recorded, helping planners understand potential risks and prepare infrastructure accordingly.
- The researchers demonstrated the method by generating maps of future extreme precipitation events across the continental United States.
- The approach can also be applied to other fields, such as financial markets and robotic navigation, to explore complex extreme events.
- The research highlights the importance of anticipating extreme events for national and economic resilience, especially as global systems become more optimized and less resilient to shocks.
algorithmextreme eventsinfrastructuresupply chainsanticipation