Vibeleaderboard
← All Intel
Intel / post

WeatherNext 3 moves to hourly forecasts at 5km resolution

Source
x.com
Date
GoogleDeepMind@GoogleDeepMind
Thread · 5 parts

WeatherNext 3 is a major breakthrough in how we forecast global weather. ⛅ Developed with @GoogleResearch, the model learns directly from real-world, real-time observations to give more localized highly accurate predictions faster. 🧵

While traditional compute constraints limit standard weather updates to six-hour intervals, WeatherNext 3 ingests real-time satellite data directly to launch a brand-new forecast every single hour.

By training directly on raw weather station observations, the system captures localized microclimates across typically underserved areas. It also delivers a 5 times resolution boost in temperature forecasts - from 25km down to 5km - in a single pass. 🏔️

Predicting rain accurately is notoriously difficult for global weather models, with previous methods producing blurry estimates or missing severe storm boundaries. WeatherNext 3 achieves a major leap in global precipitation forecasting, delivering up to a 50% reduction in error - with the greatest improvements in regions where forecasts have historically been less reliable.

Read the full thread on X

Context

Google DeepMind’s WeatherNext 3 uses live satellite observations to initialize a new global forecast every hour. Its release article explains that weather-station observations also help it learn local conditions, including differences associated with terrain.

The added detail varies by quantity. Temperature and moisture forecasts reach a 5 km grid, while other surface variables use 10 km and atmospheric variables use 25 km. Compared with WeatherNext 2’s 25 km grid and six-hour increments, this gives selected forecasts both finer local detail and more frequent updates. The 5 km figure describes spatial resolution, not a uniform accuracy improvement for every weather variable.

More from GoogleDeepMind
Recommended reads
Comments

Checking sign-in…

Loading comments…