Google DeepMind 已开源其 WeatherNext AI 模型，该模型显著提升了气旋预报能力。WeatherNext 发表在《自然》杂志上，可提供平均多 24 小时的风暴路径和强度预报提前量，将三天预报的准确性与之前的两天预报相匹配。该模型在广泛的大气数据和历史气旋记录上进行了训练，能在不到一分钟的时间内生成 15 天的概率预报。此次发布包括代码和模型权重，可供学术和业务使用。 AI
影响 加速 AI 在天气预报和灾害准备中的应用。
排序理由 Google DeepMind 发布了带有系统卡和代码/权重的 AI 模型。
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
We’re open sourcing the code and model weights on @Github, making them freely available for anyone to build on. This could be for academic purposes, operational forecasting, or developing more specialized, localized models. Explore the research → https://t.co/hyllj4fVKW
During Hurricane Melissa, WeatherNext gave forecasters early predictions of its Category 5 landfall 5 days in advance with 80% confidence. This year, we’re providing 1,000 probabilistic predictions per storm to support forecasters, now accessible via WeatherLab → https://t.co/je…
The model learned from years of everyday global atmospheric data alongside a curated database of almost 5,000 historical cyclones. It generates each 15-day probabilistic forecast scenario in under a minute on a TPU. https://t.co/vptoun7H9G
WeatherNext delivers a decade's worth of forecasting progress in a single leap. 📈 On average, 3-day predictions now match the quality that prior models could only provide 2 days out. https://t.co/oHirwymtF7
Predicting cyclones accurately can help save lives - and every hour of lead time counts. Published in @Nature, our AI model WeatherNext achieves state-of-the-art accuracy in forecasting a storm’s track and intensity, giving us a critical extra 24 hours to prepare on average. 🧵 h…