Data assimilation lies at the heart of numerical weather prediction, enabling the integration of observational data with dynamic atmospheric models to produce more accurate initial conditions and ...

Nvidia’s first model is Medium Range, which is aimed at weather predictions for up to 15 days in advance, across more than 70 weather variables, including temperature, pressure, wind, and humidity.

The intermittent nature of solar energy poses challenges to grid stability, making accurate ultra-short-term solar irradiance forecasting crucial for balancing supply and demand. However, traditional ...

AI weather models have correctly predicted a 20-day monsoon stall in India affecting 38 million farmers, demonstrating ...

The intermittent nature of solar energy poses challenges to grid stability, making accurate ultra-short-term solar irradiance forecasting crucial for balancing supply and demand. However, traditional ...

Researchers from the Hong Kong University of Science and Technology (HKUST) have developed a new AI model which they ...

The observational track of Typhoon "Danas" (solid line) along with forecasted paths (dashed lines) depicted on the FY-4B satellite visible light imagery at 08:00 BST on July 6, 2025. The dashed lines ...

A groundbreaking artificial intelligence model has achieved unprecedented accuracy in tropical cyclone intensity prediction, marking a significant ...

For weather forecasting, we traditionally use numerical weather prediction models that are based on physics principles and have seen large improvements over the last 20–30 years. We have been running ...

NVIDIA has open-sourced its Earth-2 AI weather models, claiming to outperform Google DeepMind's GenCast with 15-day forecasts running on standard GPUs.

After a merciful lack of hurricanes in the U.S. and an unusually quiet Gulf of Mexico, the 2025 Atlantic hurricane season wasn’t too memorable in Louisiana. But the season will be remembered for ...