Tag

Ai Weather Forecasting

All articles tagged with #ai weather forecasting

China bets on AI weather forecasts to sharpen typhoon warnings
technology1 month ago

China bets on AI weather forecasts to sharpen typhoon warnings

China is advancing AI-driven weather forecasting with models like Fengwu, Pangu, and Fuxi developed by Shanghai AI Laboratory, Huawei, and Fudan University, which can generate forecasts faster and sometimes more accurately than traditional systems and are being tested during Typhoon Dolphin’s approach; AI speeds up forecasts and reduces computing costs but lags in predicting storm intensity and long-range climate shifts, so experts expect continued use of both AI and conventional models for now.

science2 months ago

AI-Driven WeatherNext Cyclones Elevates Tropical Cyclone Forecasts via Massive Ensembles

Nature reports WeatherNext Cyclones (WN-C), an AI-powered operational weather model that generates state-of-the-art ensemble forecasts for the track, intensity, and size of tropical cyclones worldwide. Trained on global analysis data and historical cyclones, WN-C produces ensembles up to 1,000 members up to 15 days ahead, delivering about a day or more of lead time improvement over leading models. The system shows that high-resolution input is not strictly required for strong intensity forecasts, and that including WN-C in a weighted ensemble improves overall skill, marking a step-change toward more reliable and timely forecasts and warnings.

"Assessing the Safety and Visibility of the April 8th Total Solar Eclipse"
weather-forecasting2 years ago

"Assessing the Safety and Visibility of the April 8th Total Solar Eclipse"

The cloud cover forecast for the upcoming solar eclipse on April 8 is still uncertain, as predicting clouds is challenging due to small-scale atmospheric processes. Meteorologists are using ensemble models to analyze a range of possible weather outcomes and their probabilities. While deterministic forecasts show extensive cloud cover along the eclipse path, AI weather forecasting start-up Excarta claims its forecasts can be 20% more accurate than traditional models, with current predictions showing varying cloud cover percentages for cities in the eclipse path. The Washington Post will launch its own eclipse cloud forecast tracker, but forecast confidence may not be high until just a day or two before the event.