Modeling our future: Advancing climate research with AI
summary
Neural nets predict maps’ year from climate-model snapshots and Explainable AI backpropagates to show which regions drive that signal (Southern Ocean, North Atlantic, SE Asia), proving the models learn real, transferable fingerprints. The same approach detects stratospheric-aerosol impacts within ~10 years for temperature and ~20 years for precipitation. A hybrid framework called AiBEDO pairs a spherical U‑Net + LSTM + physics constraints trained on CMIP6 (terabytes on AWS) to search inverse cloud‑brightening strategies to avoid tipping points like West Antarctic ice sheet collapse and delivers one‑second inference.
AI-written summary · about
chapters (6)
- 0:00Introduction to climate risk
- 6:33Advanced technology applications
- 9:48Machine learning in climate
- 12:59Explainable AI and detection
- 24:54AIBedo and hybrid modeling
- 38:15Panel discussion on cloud role
related talks
09mGen AI Workflows for Climate Risk Modeling in Food Systems
22mLet's Terminate Climate Change with AI
20mFoundations for AI+Science
28mAI vs Climate Change: Insights from the Cutting Edge
17mAI Weather Forecasting
14mAI for Science
22mKEYNOTE: How to Leverage Data to Combat Climate Change 🌳
45mGreen AI: Making Machine Learning Environmentally Sustainable