Latest updates · 21 October 2026
In the full issue →AI is sharpening weather and climate forecasts: AI models gave some of the most accurate track predictions for Hurricane Isaias, and New Zealand researchers produced high-resolution climate projections in six hours instead of years.
Research, industry & policy
- AI models are transforming hurricane forecasting, WBUR Here & Now (WUSF) · 20 Oct 2026: AI models provided some of the most accurate predictions of the track of Hurricane Isaias.
- Generative AI unlocks extreme-weather insights for New Zealand, Earth Sciences New Zealand · 20 Oct 2026: The AI produced more than 15,000 simulated years of daily rainfall and maximum temperature at 12 km resolution through to 2100, within six hours, a task that would have taken years conventionally.
Journals & preprints
- AirMPA: A Meteorology-to-Pollution Adapter for Global Air Quality Forecasting, arXiv 2610.25797 · 19 Oct 2026: Accurate real-time forecasting of atmospheric pollutants is essential for reducing exposure risks and improving public health.
- Zephyrus-R1: Agentic Reinforcement Learning for Weather Reporting, arXiv 2610.25625 · 19 Oct 2026: Reinforcement learning (RL) has emerged as a key method for training AI models in scientific applications, but RL remains difficult when outputs cannot easily be verified against a known ground-truth answer.
Why it matters
Faster, cheaper simulations let scientists study rare extremes such as 1-in-100-year rainfall that matter most for planning.
